El manto de nieve en la península ibérica: climatología y sensibilidad a la variabilidad climática

PROGRAMA DE DOCTORADO EN CONSERVACIÓN Y USO SOTENIBLE DE SISTEMAS FORESTALES TESIS DOCTORAL: The snowpack over the Iberian Peninsula: climatology and sensitivity to climate variability El manto de nieve en la península ibérica: climatología y sensibilidad a la variabilidad climática Presentada por Esteban Alonso González para optar al grado de Doctor/a por la Universidad de Valladolid Dirigida por: Dr. Juan Ignacio López Moreno Dr. Antonio Ceballos Barbancho Índice Agradecimientos:..................................................................................................................................1 Resumen...............................................................................................................................................1 Abstract.................................................................................................................................................5 Capítulo 1: Introducción.....................................................................................................................11 1.1 La importancia del manto de nieve..........................................................................................11 1.2 Información disponible del manto de nieve.............................................................................13 1.2 Alternativas a las observaciones..............................................................................................17 1.2.1 Uso de información por satélite en el estudio del manto de nieve...................................17 1.2.2 Uso de simulaciones numéricas del manto de nieve para generar información del espesor de nieve y del SWE...................................................................................................................20 1.3 El manto de nieve en la península ibérica................................................................................23 1.4 Impactos de la variabilidad climática en el manto de nieve de la península ibérica...............27 1.5 Objetivos..................................................................................................................................28 Referencias:...................................................................................................................................30 Capítulo 2: Rejillas de datos diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014....................................................................................................43 Daily gridded datasets of snow depth and snow water equivalent for the Iberian Peninsula from 1980 to 2014.......................................................................................................................................45 1 Introduction.................................................................................................................................46 2. Data and methods.......................................................................................................................48 2.1 Meteorological driving data................................................................................................49 2.2 Snow energy and mass balance model................................................................................50 2.3 Validation procedure...........................................................................................................53 3 Results.........................................................................................................................................55 3.1 Validation............................................................................................................................55 3.2 Gridded snow dataset: applications and limitations............................................................58 4. Data availability.........................................................................................................................62 5. Conclusions................................................................................................................................62 References......................................................................................................................................64 Capítulo 3: Climatología de nieve de las montañas de la Península Ibérica mediante imagenes por satélite y simulaciones con mejoras de escala dinámicas...................................................................73 Snow climatology for the mountains in the Iberian Peninsula using satellite imagery and simulations with dynamically downscaled reanalysis data...........................................................................75 1 Introduction.................................................................................................................................76 2 Study area...................................................................................................................................78 3 Data and methods........................................................................................................................80 3.1 Remote sensing...................................................................................................................80 3.2 SD and SWE simulation.....................................................................................................81 4 Results and discussion................................................................................................................83 4.1 Probability of snow cover from MODIS.............................................................................83 4.2 Simulated snow SD and SWE.............................................................................................88 4.3 Intra-range variability from simulated snow data...............................................................93 4.4 Implications of the presented results...................................................................................94 5 Conclusions.................................................................................................................................95 References......................................................................................................................................96 Capítulo 4: Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica..............................................................................................................................109 Impact of North Atlantic Oscillation on the Snowpack in Iberian Peninsula Mountains.................111 1. Introduction..............................................................................................................................112 2. Study Area................................................................................................................................113 3. Methods....................................................................................................................................115 3.1. NAO Index and Temperature and Precipitation Patterns.................................................115 3.2. Snowpack Database and Statistical Analysis...................................................................115 4. Results and Discussion............................................................................................................116 4.1. Relationship between the DJFM-NAO and Temperature and Precipitation....................116 4.2. Relationship between the Winter NAO and Temperature and Precipitation....................117 4.3. NAO Spatial Influence on Peak SWE and Snow Season Duration.................................119 5. Conclusions..............................................................................................................................124 References....................................................................................................................................126 Capítulo 5: Sensibilidad del manto de nieve a la variabilidad de la temperatura, precipitación y radiación solar en un gradiente altitudinal en la Península Ibérica.....................................................135 Snowpack sensitivity to temperature, precipitation, and solar radiation variability over an elevational gradient in the Iberian mountains...............................................................................................137 1 Introducción..............................................................................................................................138 2 Study Area.................................................................................................................................140 3 Data and methods......................................................................................................................141 4 Results.......................................................................................................................................144 4.1 Selection of representative areas.......................................................................................144 4.2 Sensitivity of the snowpack duration and peak SWE to precipitation, short wave radiation, and temperature change...........................................................................................................146 4.3 Effect of temperature variability on the snowmelt rate.....................................................150 5 Discussion.................................................................................................................................152 6 Conclusions...............................................................................................................................155 References....................................................................................................................................157 Capítulo 6: Conclusiones..................................................................................................................167 Conclusions......................................................................................................................................171 Agradecimientos: Este trabajo no hubiera sido posible sin el apoyo de muchas personas, que de una manera u otra han contribuido a que esté ahora mismo escribiendo estas líneas. Primero, me gustaría agradecer el apoyo recibido por mis dos directores de tesis Nacho y Antonio. En todo momento he tenido la sensación de estar arropado por vuestra experiencia. Siempre habéis estado disponibles para mí, y habéis estado implicados al máximo en este proyecto. Nacho, las charlas sobre monte mientras hacíamos ciencia, y de ciencia mientras hacíamos monte, han sido una de las mayores fuentes de inspiración de estos cuatro años. Espero sigamos así mucho tiempo más, gracias. A mis compañeros IPErinos, especialmente al superpoblado despacho 3, hacéis que la gente se sienta en casa desde el primer día. Siempre se puede contar con vosotros para arreglar una línea de código, o la tarde de un viernes. Muchas gracias Chavi, Jesús, Fergus, Mocho, Makki, Ivan, Paco, Natalia, Clara, Dani, Ana, Elena solo por nombraros a algunos entre la mucha gente con la que he coincidido por el IPE en estos años. También me gustaría daros las gracias a vosotros Guille y Emilien que, a pesar de vuestra corta estancia en el IPE, hemos compartido mucho fuera del IPE, una buena parte del tiempo atados a la misma cuerda. A mis compañeros de mi estancia en el NCAR, especialmente a Sip, Luuk, Bao y Kris. Fue un poco frustrante aparecer en Colorado, el paraíso de la nieve polvo, en uno de los inviernos con menos nieve que se recuerdan. Pero a cambio, hicimos multitud de viajes juntos que se quedarán en mí memoria. Espero podamos seguir en contacto como hasta ahora, a pesar de los miles de kilómetros que nos separan. Buena parte de estos años los hemos pasado juntos como “extranjeros” en Zaragoza. Barbara, Feni y Lorenzo, muchas gracias por todas las mañanas, tardes y noches que hemos pasado juntos. Veremos si en la próxima ciudad en la que coincidamos, seguimos discutiendo sobre donde son peores las bravas. Por supuesto, tengo que agradecer el apoyo de mi familia, ya que de alguna manera también es en parte suyo este trabajo. Los tirones de orejas y las palmadas en la espalda son al final los que me han traído hasta aquí. A mis padres y mi hermana, muchas gracias. Itsaso, te mereces mucho más que estas líneas. Has conseguido convertir en llevadera la experiencia de escribir una tesis en medio de un confinamiento obligado de varias semanas. Tú me has dado la energía para culminar este trabajo, aportando como solo tú sabes. Este trabajo es en gran medida gracias a ti. Y por último, me gustaría dedicar este trabajo a sus máximas protagonistas, las montañas. Quien pudiera imaginar que seria de todos, y especialmente de mí, sin vosotras. A vosotras os dedico este trabajo. Resumen clustering algorithm was used to add similar climatologies over the different mountain ranges. We also studied how the values of snow probability and coefficient of variation change as a function of the elevation for each mountain range, as well as the intra-range variability. The results showed how the different behaviors of the snowpack are unevenly distributed along the elevational range of each mountain range. Snow values found above 1000m a.s.l. oscillate between areas with ephemeral or practically non-existent snow cover in contrast to areas with deep and long lasting snowpacks (198 days of snow presence and ~3m SD). The calculated snow indices suggest that, in terms of elevational bands, the Cantabrian Mountains and the Pyrenees have the largest and longest snow cover, followed by the Central and Iberian Systems. Sierra Nevada showed the shortest seasons and shallowest snow cover with the greatest inter-annual variability. The winter precipitation and temperature of the Iberian Peninsula is highly affected by the synoptic patterns associated with the North Atlantic Oscillation index (NAO) and therefore must necessarily influence the interannual variability of the snow cover. We have studied the influence of the NAO on the duration and magnitude of the snowpack in the Iberian Peninsula as well as the spatial patterns of this relationship. For this purpose, we calculated from the database described above the series of maximum accumulation and annual duration. Then we correlated these snow indexes with the averaged NAO value of the months of December, January, February and March (DJFM-NAO). The results show very high negative correlation in all the main mountain ranges of the Iberian Peninsula. This is particularly evident on the slopes exposed to the advections fostered by the negative phases of the DJFM-NAO index, mostly in a southwestern and western direction. To prove this, we have used a non-parametric statistical test (Wilcoxon-Mann-Whitney test) between the correlation values of the snowpack indices and the DJFM-NAO of both slopes. Thus, we prove that the opposite slopes respond differently in all the main mountain ranges with the exception of Sierra Nevada where no differences between slopes were found. Strong relationships were found between the correlation values of the areas exposed to the advection and the geographical length and elevation. The correlations were generally greater at high elevations and tending to be located in the western areas of the mountain ranges. These results could potentially be used on a long-term high spatial resolution prediction of the magnitude and duration of the snowpack. Our findings prove the importance of the existence of long series in detecting temporal and spatial trends in the duration and magnitude of the snowpack. Finally, the response of the snowpack to climate variability in different elevational bands has been studied. A factorial experiment was implemented, performing simulations of the snowpack by 6 Resumen means of FSM with progressive perturbations in the meteorological forcing simulated by WRF. The perturbation ranges of temperature were 0-4 ºC in 0.5 ºC increments, 0-40 Wm-2 of short wave radiation in 5 Wm-2 increments and ± 20% of precipitation in 5% increments, performing simulations within these disturbance ranges in all possible combinations. From the new SWE series generated we calculated the series of maximum annual SWE, duration of the snow season and average annual melting rate and used to estimate the percentage of change per increment of perturbation. The results have shown different sensitivities to climate variability but always with important losses of magnitude and duration of the snowpack at all elevation bands and study areas even in the situations of higher increase of precipitation. Sensitivity values of the melt rates were negative, suggesting longer and less intense melting periods, with obvious implications on the hydrology of areas influenced by the snow melt. 7 Resumen 8 Introducción Capítulo 1: Introducción 1.1 La importancia del manto de nieve Debido sus particulares características, el manto de nieve juega un papel crucial en una gran cantidad de procesos climáticos, ambientales y socio-económicos en numerosas zonas templadas y frías del planeta. El manto de nieve estacional cubre aproximadamente entre un mínimo de 2x106 km² y un máximo de 45x106 km2 del Hemisferio Norte (Fig. 1) en invierno y verano respectivamente. Esto supone una cobertura de cerca del 50% de tierras emergidas en el hemisferio Norte (Déry and Brown, 2007), controlando directamente el balance energético del planeta Tierra y por ende condicionando de manera drástica el clima global. A modo de ejemplo, Barnett et al., 1988 realizaron una de las primeras simulaciones numéricas para testar el efecto del manto de nieve estacional de Eurasia en el clima global. Sus resultados relacionaron la variabilidad interannual del manto de nieve a escala continental con los monzones, el clima de Norte América, así como otros aspectos muy relacionados con el fenómeno de El Niño-Oscilación del Sur. Más recientemente, estas hipótesis han seguido siendo testadas y corroboradas. Vavrus, 2007 intentó cuantificar el efecto global del manto de nieve en el clima del planeta mediante el uso de modelos atmosféricos globales, comparando una simulación control con otra simulación sintética sin manto de nieve. Sus resultados sugirieron drásticas consecuencias sobre el clima, afectando no solo las temperaturas del suelo y su humedad, si no a grandes patrones atmosféricos en latitudes medias y regiones polares. Algunas de las implicaciones más notables fueron un drástico calentamiento atmosférico que legó a superar los 10 ºC sobre la media durante el invierno en algunas zonas. El calentamiento global inducido por la falta de manto de nieve fue comparable a un tercio del inducido por doblar la cantidad de CO2. El aire contenido entre los cristales de hielo confiere al manto de nieve una gran capacidad aislante, resultando éste determinante para comprender la ecología de las zonas donde está presente. Las condiciones relativamente templadas del suelo bajo el manto de nieve, posibilitan una gran actividad biológica, incluyendo la actividad de plantas (Salisbury, 1985) y animales tanto vertebrados como invertebrados (Jones, 2001). Estas mismas características aislantes permiten la actividad microbiana siempre que se mantenga la temperatura del suelo por encima de aproximadamente -5 ºC (Schimel and Clein, 1996), lo cual tiene implicaciones en los ciclos de 11 Introducción elementos básicos del suelo, como por ejemplo el nitrógeno (Brooks and Williams, 1999; Jones, 1999 entre otros). Como recurso hidrológico, Barnett et al., 2005 estimaron que aproximadamente una sexta parte de la población mundial vive directamente en lugares en los que la fusión de nieve domina la hidrología de la región. Los mismos autores reconocen que esta estimación probablemente infravalora el grado de dependencia, ya que no incluye a las poblaciones que obtienen sus recursos hídricos fuera de las cuencas hidrológicas en las que viven. Más concretamente, Viviroli et al., 2007 estimaron que más de un 50% de las zonas montañosas del planeta juegan un papel esencial en las poblaciones aguas abajo de las cuencas de montaña, zonas en las que el ciclo hidrológico está profundamente influido por los procesos de acumulación y fusión nival. La importancia de la nieve como recurso hídrico, se vuelve especialmente crítica en las zonas montañosas con clima mediterráneo. La razón es que el clima mediterráneo está caracterizado por veranos secos, sucediendo la mayor parte de la precipitación anual durante los meses de invierno y 12 Figura 1: Promedio de cobertura de nieve del periodo 1999 - 2019 sobre el hemisferio norte calculado a partir el reanálisis ERA5-Land. Introducción en menor medida otoño y primavera. El conjunto del periodo alcanza con frecuencia el 80% del total anual de las precipitaciones (Fayad et al., 2017), con valores que pueden llegar hasta el 90%, como ha sido cuantificado para la cordillera de Sierra Nevada (Estados Unidos) (Jepsen et al., 2012). A la concentración de las precipitaciones en la estación fría, hay que sumarle el hecho de que las montañas concentran buena parte de las precipitaciones debido al efecto orográfico, de tal forma que las zonas elevadas acumulan un manto de nieve profundo y de considerable persistencia a lo largo de la primavera (Alonso-González et al., 2019). Por ello, el manto de nieve en el Mediterráneo tiene la capacidad de mitigar el efecto de la fuerte estacionalidad pluviométrica sobre el régimen fluvial de los ríos de montaña, con caudales elevados durante la primavera que, en combinación con infraestructuras para la gestión de los recursos hídricos (embalses, canales, etc), resulta crucial para satisfacer la elevada demanda de agua durante los meses de verano (López-Moreno and García- Ruiz, 2004a), cuando las actividades turísticas, la agricultura y el sector energético producen un máximo de consumo de agua. Debido a la influencia del manto de nieve en todos estos factores, resulta de gran interés conocer su variabilidad espacial y temporal, así como los factores que condicionan esta variabilidad. El manto de nieve presenta una gran variabilidad a diferentes escalas. A una escala local la distribución del manto de nieve está muy condicionada por las características topográficas del terreno, como pueden ser la pendiente, la curvatura o la exposición (López-Moreno et al., 2017; Revuelto et al., 2014), así como por los efectos de la redistribución por el viento (Winstral et al., 2002) o el bosque (Sanmiguel-Vallelado et al., 2020), jugando estos factores un importante papel en la formación de avalanchas de nieve (Schweizer et al., 2008). A escala regional, la variabilidad del manto de nieve está condicionada por la variabilidad climática. De esta manera los patrones atmosféricos tienen una gran influencia en la variabilidad interanual del espesor y duración del manto de nieve (Clark et al., 1999), ya que controlan los patrones espaciales y la variabilidad interanual de la precipitación y temperatura. 1.2 Información disponible del manto de nieve A pesar de todas las implicaciones que el manto de nieve tiene en los procesos anteriormente explicados, existen grandes lagunas de conocimiento en su comportamiento a nivel regional. Dicha carencia se explica fundamentalmente por una falta generalizada de información cuantitativa sobre la duración y más aún el espesor y contenido en agua del manto de nieve, limitándose en la mayor 13 Introducción parte de los casos a registros de escasa consistencia espacial y referida a periodos de corta duración y con presencia de frecuentes lagunas. A la situación general anteriormente mencionada hay excepciones con registros de larga duración, densidad espacial y calidad en los datos. Un ejemplo fue la base datos diaria de espesor de nieve en 284 estaciones de la antigua Unión Soviética conocida como Historical Soviet Daily Snow Depth (HSDSD)(Armstrong, 2001), que cubrió el periodo 1881-1995, si bien existen importantes lagunas especialmente en el periodo anterior a 1966. En la costa oeste de los Estados Unidos se ha medido el espesor de nieve y su contenido en agua de forma regular (quincenal o mensualmente) desde las primeras décadas del siglo XX, siguiendo un buen número de transectos predefinidos. Esta red evolucionó a principios de los años ochenta hacia la actualmente conocida como red SNOTEL (SNOw TELemetry) (Schaefer and Paetzold, 2001), que cubre gran parte del oeste de Norte América mediante más de 800 estaciones automáticas que miden el espesor y su contenido en agua, así como otros parámetros meteorológicos como temperatura o precipitación. Sin embargo, esta red también posee problemas de representatividad espacial, pues las estaciones se localizan en zonas forestales y muchas veces situadas teniendo en cuenta la accesibilidad, quedando el piso alpino muy débilmente monitorizado (Molotch and Bales, 2006). En Europa occidental, no existen redes de monitorización del manto de nieve como las anteriormente descritas, en cuanto a su cobertura espacial y duración temporal, aunque sí es posible encontrar información sobre el manto de nieve, en algunas ocasiones cubriendo varias décadas de longitud, recogidas por diferentes agencias meteorológicas o hidrológicas estatales y más locales tanto para fines operacionales como para la investigación. Sin embargo muchos de estos se refieren con frecuencia a zonas de baja altitud donde se encuentran los principales núcleos de población. Así, zonas de alta montaña donde se acumula la mayor parte del recurso nieve quedan muy débilmente monitorizadas. Un buen ejemplo de red de monitorización en Europa occidental es la dirigida por el Instituto para la Investigación de la Nieve y las Avalanchas (Suiza, WSL Institute for Snow and Avalanche Research SLF) iniciada en la década de 1960 y con mediciones continuas de espesor de nieve y equivalente en agua (SWE). Los datos son obtenidos manualmente dos veces al mes en diferentes puntos de los Alpes Suizos (Jonas et al., 2009). Otra fuente de información interesante son las bases de datos nivometeorológicas de las diferentes zonas experimentales repartidas por el continente Europeo. En ellas se cuenta con información de larga duración del manto de nieve así como de multitud de variables meteorológicas. Algunos buenos ejemplos de zonas de investigación del manto de nieve pueden ser la zona de observación Suiza Weissfluhjoch, 14 Introducción que cuenta con datos desde 1936 (Marty and Meister, 2012), la estación francesa Col du Porte (Lejeune et al., 2019), que cuenta con información desde 1960 o la finlandesa Sodankyla (Essery et al., 2016). Un completo informe sobre los esfuerzos de los diferentes estados europeos para monitorizar el manto de nieve lo podemos encontrar en el “European Snow Booklet”(Haberkorn et al., n.d.). En el caso de la península ibérica, la monitorización del manto de nieve de forma regular y distribuida espacialmente comenzó en los años 80 (1986) en el Pirineo a través del programa ERHIN (Evaluación de los Recursos Hídricos Procedentes de la Innivación) (Arenillas et al., 2008; Pedrero, 1988) por parte del Ministerio de Medio Ambiente y Medio Rural y Marino. En el marco de este programa se instalaron alrededor de 100 jalones en los que se medía el espesor de nieve tres veces al año (enero, marzo y finales de abril o principios de mayo) y la densidad en alrededor de un tercio de las balizas (López-Moreno and Nogués-Bravo, 2005). Posteriormente el proyecto se expandió a otras cordilleras españolas como la Cordillera Cantábrica, Sistema Central y Sierra Nevada. A parte de las mediciones periódicas, un total de 29 telenivómetros se instalaron en las principales cordilleras de la península ibérica, con 19 más propuestos para instalar en una fase más avanzada sin determinar. Estos telenivómetros tienen capacidad para medir en tiempo real tanto el espesor como el SWE, y transmitir remotamente la información. Desgraciadamente, este ambicioso programa se ejecutó con un éxito moderado, pues la mayor parte de los telenivómetros nunca llegaron a funcionar de forma continua siendo los que dan servicio a la cuenca del Ebro los únicos de los que actualmente se puede obtener información adecuada. En cualquier caso, los datos procedentes de las balizas han permitido realizar diversos trabajos científicos que han ayudado a avanzar en el conocimiento de la dinámica del manto de nieve y su relación con la hidroclimatología de la península ibérica. Algunos ejemplos los podemos encontrar en estudios hidrológicos (Sanmiguel-Vallelado et al., 2017), estudios de la variabilidad de los eventos de nevadas (Navarro-Serrano and López-Moreno, 2017), estudios de tendencias (López-Moreno, 2005; Morán-Tejeda et al., 2013), validación in situ de información por satélite (Juan Collados-Lara et al., 2016), o incluso utilizar la información procedente de los jalones para reproducir el manto de nieve de manera distribuida (Collados-Lara et al., 2020, 2017; López-Moreno and Nogués-Bravo, 2006, 2005) entre otros. De forma semejante a la red ERHIN, pero en la vertiente francesa de los Pirineos, Meteofrance ha instalado la red NIVOSÊ, con 9 medidores automáticos de espesor del manto de nieve junto a estaciones automáticas de alta montaña. 15 Introducción muchos modelos atmosféricos globales (ECMWF, 2019), a múltiples capas (Niu et al., 2011), llegando algunos a tener una gran complejidad a costa de un gran esfuerzo computacional y siendo destinados principalmente a su uso en la simulación de situaciones de avalanchas (Vionnet et al., 2012). El mayor problema de los modelos de nieve de base física es disponer de información meteorológica suficiente para forzar las simulaciones. Estos modelos suelen requerir diferentes tipos de variables meteorológicas para resolver cada uno de los componentes del balance de energía, entre las cuales deberán de estar presentes como mínimo la temperatura del aire, la humedad atmosférica, viento, presión atmosférica, precipitación y radiación en onda corta. Habiendo visto anteriormente la dificultad de disponer de observaciones meteorológicas y nivológicas adecuadas en zonas de alta montaña (Raleigh et al., 2016), su aplicación en la mayor parte de las montañas en general, y en la península ibérica en particular, resulta muy compleja. La única alternativa viable para encontrar información meteorológica para forzar modelos de balance de masa y energía y poder realizar simulaciones de varios años del manto de nieve la encontramos en los datos procedentes de simulaciones atmosféricas, que para reproducir periodos pasados se denominan de reanálisis. Los reanálisis atmosféricos son bases de datos realizadas mediante modelos atmosféricos globales. Estos modelos son forzados por datos históricos procedentes de muy diversas fuentes de información incluyendo estaciones meteorológicas convencionales, información por satélite, globos sonda meteorológicos, estaciones meteorológicas o boyas marinas entre otras. Existen multitud de reanálisis diferentes que cubren diferentes periodos de tiempo con multitud de particularidades diferentes. Si bien todos ellos tienen muchas aplicaciones en el área de la climatología, prácticamente todos tienen en común la poca resolución espacial disponible, que a día de hoy puede llegará ser en el mejor de los casos de 0.25º (~ 30 km) siendo comunes resoluciones cercanas o superiores a 1º (~ 110 km). Esto imposibilita totalmente su uso para simular el manto de nieve en terreno complejo debido al enorme suavizado de la topografía que representa (Mass et al., 2002). No obstante, es posible incrementar su resolución mediante modelos atmosféricos regionales. Estos modelos funcionan de manera análoga a los modelos atmosféricos globales, con la particularidad de que es posible lanzarlos en zonas delimitadas del territorio con un modelo de elevaciones muy superior al del reanálisis original, y utilizando las condiciones iniciales y de contorno, de un modelo atmosférico global previo. El uso de este tipo de datos para realizar simulaciones físicas del balance de energía no es nuevo, por ejemplo van Pelt et al., 2016 utilizaron un procedimiento similar para Svalbard utilizando las salidas del modelo 22 Introducción climático regional HIRLAM como forzamiento del modelo de balance de masa y energía Snowmodel (Liston and Elder, 2006a). De manera análoga, Wu et al., 2016 generaron productos de SWE en las montañas de Altai, donde usaron las salidas del modelo climático WRF (Skamarock et al., 2008) como forzamiento de un modelo tipo grados día, ajustado mediante información por satélite. A pesar de las mejoras de resolución de los modelos atmosféricos regionales, las resoluciones alcanzadas cuando se simulan periodos de tiempo largos (más de 30 años) no son suficientes para resolver adecuadamente todos los campos meteorológicos de las zonas de terreno complejo con un coste computacional asumible. Existen diferentes maneras de sortear esta barrera computacional, mediante aproximaciones que, realizando algunas simplificaciones, consiguen mejorar todavía más la resolución de las simulaciones atmosféricas con un coste computacional asumible. Un ejemplo de estas soluciones alternativas es el modelo MICROMET (Liston and Elder, 2006b), el cual interpola espacialmente las variables meteorológicas de superficie de un modelo atmosférico (o una red de estaciones meteorológicas), y las corrige con la altura sobre un modelo digital de elevaciones, generando nuevos campos meteorológicos de mayor resolución. Estos productos pueden ser de mucha utilidad si se asume la incertidumbre introducida por la interpolación estadística o por otras simplificaciones como el uso de gradientes térmicos y pluviométricos, especialmente si se cuenta con datos observados para asimilar en el modelo. Otra técnica muy usada han sido las simulaciones semidistribuidas, en las que se reproducen los campos meteorológicos necesarios para forzar un modelo del manto de nieve en un punto con unas características de exposición, elevación o pendiente concretas (Revuelto et al., 2018), habiendo sido esta técnica previamente utilizada en los Pirineos para el estudio del efecto del cambio climático en el manto de nieve (López-Moreno et al., 2009). 1.3 El manto de nieve en la península ibérica A pesar de que no existe una climatología de nieve a nivel regional de toda la península ibérica, sí que existen trabajos previos que describen diversos aspectos de la dinámica del manto de nieve, especialmente en los Pirineos y Sierra Nevada. La península ibérica tiene una topografía muy compleja con varias cordilleras montañosas que pueden resumirse en cinco macizos donde la nieve posee un papel relevante desde un punto de visto hidrológico, ecológico y socioeconómico: Cordillera Cantábrica, Sistema Central, Sistema Ibérico, Pirineos y Sierra Nevada. Estas cordilleras rodean dos importantes mesetas que hacen de la península ibérica una de las regiones más elevadas 23 Introducción de Europa. A pesar de su baja latitud, nevadas ocasionales son comunes en la mayor parte del territorio, siendo muy frecuentes en varias ciudades importantes de la mitad norte de la península (Merino et al., 2014). Existen registros de nevadas incluso cerca del nivel del mar en la zona de Levante (Mora et al., 2016), Barcelona o Gerona (Aran et al., 2010). En cualquier caso, a pesar de que estos eventos suelen tener una repercusión muy elevada, ya que suceden en zonas de alta densidad de población, la realidad es que la península ibérica cuenta con un manto de nieve estacional únicamente en las zonas montañosas, dónde sí persiste hasta el final de la primavera y principios de verano, incluso en las zonas más meridionales y áridas como Sierra Nevada (Pimentel et al., 2017). Estas zonas montañosas ejercen un gran control sobre las zonas aguas abajo (López and Justribó, 2010) como ha sido confirmado en los Pirineos (López-Moreno and García-Ruiz 2004). De esta manera, el ciclo hidrológico está, en las épocas de fusión, mucho más relacionado con las acumulaciones de nieve existentes en cotas altas que con la precipitación en forma líquida, con importantes implicaciones también en las avenidas, las cuales son amplificadas a consecuencia de la fusión de nieve (Morán-Tejeda et al., 2019). Aparte de la importancia hidrológica anteriormente descrita, no hay que olvidar el impacto que el manto de nieve posee directamente sobre la economía de la península ibérica, y más concretamente en zonas muchas veces afectadas por la despoblación o la falta de motores económicos. El turismo relacionado de alguna manera con la nieve, y más concretamente el esquí alpino, ha demostrado ser una fuente de riqueza con potencial para recuperar la demografía de las zonas donde se consolida (Lasanta et al., 2007). En cualquier caso, las zonas rurales montañosa,s donde se asienta un manto de nieve estacional, están sufriendo una continuada despoblación, y es precisamente en ellas donde la nieve juega un papel como motor económico en forma de turismo deportivo, o de cualquier otro tipo. Esta es la razón por la que la gestión del territorio en las zonas rurales de la montaña Ibérica ha pasado en muchas ocasiones por la gestión de las actividades directamente relacionadas con la nieve por parte de los entes públicos y políticos (Gilaberte-Búrdalo et al., 2017). Otros sectores socioeconómicos afectados por la variabilidad espacio temporal de la nieve, tienen relación con la producción forestal, la cual está influenciada por los patrones de acumulación del manto de nieve (Sanmiguel-Vallelado et al., 2019) o la gestión del riesgo tanto en el transporte, el cual se ha demostrado aumenta considerablemente con las nevadas (Mills et al., 2011) o por avalanchas. Las avalanchas representan uno de los mayores riesgos en zonas montañosas, a las que se les asocia alrededor de 1900 muertes en Europa y Norte América en el periodo desde 2000/2001 hasta 2009/2010 (Schweizer et al., 2015). 24 Introducción La variabilidad intrannual e interanual de los eventos de nieve está muy asociada a los diferentes patrones sinópticos atmosféricos (Esteban et al., 2005; Navarro-Serrano and López-Moreno, 2017). La Oscilación del Atlántico Norte (NAO) es un índice de teleconexión que se define como la diferencia de presión atmosférica para el dipolo centrado en Islandia y las Islas Azores. En la cuenca Mediterránea, la NAO es el principal índice de teleconexión que determina los patrones espaciales y la magnitud de la precipitación y la temperatura (Fig.3) (Corte‐Real et al., 1995; Hurrell, 1995; J. I. López-Moreno et al., 2011b). Es por esto que el manto de nieve estacional estará necesariamente relacionado con la NAO, o más concretamente con el índice NAO promedio de la temporada de acumulación. López-Moreno and Vicente-Serrano (2007) y posteriormente Buisan et al.(2015) comprobaron una clara correlación negativa entre el índice NAO y la acumulación de nieve en el Pirineo Central, si bien con coeficientes muy variables en altura y localización geográfica. A una escala espacial mucho más amplia, López-Moreno et al. (2011) utilizaron datos climáticos en rejilla (0.5º de resolución) proporcionados por la Unidad de Investigación Climática de la universidad de East Anglia (CRU, Mitchell and Jones, 2005) junto con información observada del manto de nieve de los Pirineos y Alpes, para valorar la relación que existe entre el índice NAO y el manto de nieve a escala regional para las montañas que rodean la cuenca del Mediterráneo. Quedó demostrada la influencia de la NAO en el manto de nieve, pero con varias puntualizaciones. Por ejemplo, el efecto de la NAO de invierno en el manto de nieve tiende a perder relevancia cuanto más hacia el este se encuentra una determinada cordillera. Además de esto, el efecto de la NAO invernal en el manto presenta un claro gradiente altitudinal, a consecuencia de que en las cotas bajas la presencia de nieve está más marcada por el régimen térmico que por el pluviométrico. Esto se produce a consecuencia de que los patrones sinópticos asociados a las fases de NAO negativa favorecen los flujos de viento procedentes del sur y sur oeste, por lo que propician advecciones de humedad, pero con condiciones térmicas templadas. 25 Introducción Esta relación entre el índice NAO y el manto de nieve tiene grandes implicaciones. A día de hoy, las predicciones estacionales a largo plazo mediante simulaciones atmosféricas presentan una gran incertidumbre a consecuencia de la no-linealidad de las ecuaciones involucradas en cualquier modelo atmosférico numérico moderno. Esto hace muy difícil poder predecir con exactitud la cantidad de nieve que estará presente en el futuro a medio plazo de manera directa. En cambio, la predicción de los grandes patrones sinópticos atmosféricos mediante el uso de indicadores empíricos (Scaife et al., 2014; Wang et al., 2017), con una antelación temporal entre uno y varios años, sí ha mejorado notablemente (Dunstone et al., 2016). De esta manera, potencialmente sería posible obtener una aproximación a la cantidad de nieve disponible con mucha antelación, con todas las implicaciones que ello tiene para la gestión de los recursos hídricos y la economía. A pesar de la certidumbre que se tenía en la influencia de la NAO invernal en el manto de nieve, la falta de datos o, como en el caso del trabajo de López-Moreno et al., 2011, la falta de resolución, no han permitido un estudio en profundidad de cómo esta influencia se distribuye espacialmente, información fundamental para el uso operacional de la NAO invernal como predictor del manto de nieve. 26 Figura 3: Patrones de correlación entre el indice NAO y la precipitación y temperatura en el Hemisferio Norte. Modificado de NOAA/ National Weather Service. Introducción 1.4 Impactos de la variabilidad climática en el manto de nieve de la península ibérica No es solo crucial conocer las características actuales del manto de nieve, si no entender como estas características pueden evolucionar en el futuro. Debido a su naturaleza, el manto de nieve responde de manera muy clara a la variabilidad climática. En el actual proceso de calentamiento en el que nos encontramos, son de esperar importantes cambios en el manto de nieve que potencialmente tendrán efectos sobre un alto porcentaje de la población mundial (Barnett et al., 2005). A parte de las claras implicaciones que el cambio climático tendrá en el manto de nieve en términos de duración y magnitud, recientes estudios han demostrado como el manto de nieve puede responder de manera muy compleja a los procesos de calentamiento. Por ejemplo, Musselman et al., 2017 observaron que bajo condiciones más cálidas las tasas de fusión tienden a ralentizarse, hipótesis que ha sido confirmada para el conjunto del hemisferio norte (Wu et al., 2018). Esta respuesta implica periodos de fusión más largos, con importantes impactos en la gestión de los recursos hídricos entre otros. Otro ejemplo de cómo de compleja es la respuesta del manto de nieve a los procesos de calentamiento lo encontramos en Harpold and Brooks, (2018). En este trabajo se demuestra que los procesos de fusión asociados al calentamiento están controlados en realidad por la humedad atmosférica que controla el flujo de masa en forma de sublimación y por lo tanto una importante parte del balance de energía. Diferentes repuestas del manto de nieve a la variabilidad climática se pueden producir incluso dentro de áreas catalogadas como climáticamente iguales. López-Moreno et al. (2017), demostraron como la sensibilidad del manto de nieve al calentamiento puede ser muy diferente incluso en zonas catalogadas como clima mediterráneo, con diferencias de una disminución de entre unos 6 días ºC -1 de duración y un 9% ºC -1 de máximo de acumulación, hasta 28 días ºC -1 y un 19% ºC -1 de máximo de acumulación en las zonas más sensibles. Todas estas respuestas tan diferenciadas se producen a consecuencia de los diferentes pesos de los componentes de los respectivos balances de masa y energía de cada lugar. A demás del incremento de temperatura asociado al cambio climático, existen otras fuentes de variabilidad climática que influyen directamente en el balance de masa y energía de la nieve. Por ejemplo, existe una gran incertidumbre sobre como las precipitaciones pueden evolucionar en las próximas décadas (Monjo et al., 2016), a pesar de que la disminución de la nieve a causa del cambio climático puede estar muy relacionada con la disminución de las precipitaciones (Irannezhad et al., 2016). Por otro lado, hay una tendencia muy clara en el incremento de la 27 Introducción radiación en onda corta sobre la península ibérica, cuando se analizan datos observados. Este incremento se achaca al efecto mixto de una atmósfera con cada vez con menos aerosoles, y un descenso de la nubosidad sobre la península (Vicente-Serrano et al., 2017), efecto que ha sido también observado para la totalidad de Europa (Sanchez-Lorenzo et al., 2017). Todos estos factores representan una gran fuente de incertidumbre sobre como el manto de nieve podría evolucionar en las próximas décadas, a la que habría que sumar el efecto de la altitud, ya que se ha probado que la sensibilidad del manto de nieve es mayor en los lugares cerca de la isoterma 0 ºC (Pierce and Cayan, 2013), y que resulta crucial en una zona de topografía tan compleja como es la península ibérica. 1.5 Objetivos Todo lo expuesto en los apartados anteriores, pone de manifiesto la necesidad de entender en profundidad la forma en que el manto de nieve se comporta a nivel regional, dado el importante papel que este juega en múltiples aspectos de la península ibérica. En este contexto se desarrolla la presente tesis doctoral, la cual tiene como objetivo principal conocer el comportamiento del manto de nieve a nivel regional, así como su respuesta a la variabilidad climática. Expuestas previamente las dificultades y las herramientas disponibles para completar este objetivo, así como un resumen del conocimiento actual que se tiene sobre el manto de nieve en la península ibérica, nos planteamos los siguientes objetivos secundarios, que se han materializado en diferentes etapas del trabajo, habiendo sido cada uno de ellos objeto de una publicación científica: 1- Generar una base de datos del manto de nieve semidistribuida a resolución diaria consistente con las observaciones disponibles que cubran un periodo mínimo de 30 años, permitiendo así realizar estudios climáticos robustos. 2- Estudiar las características del manto de nieve a nivel regional en las principales cordilleras de la península ibérica, realizando la primera climatología de nieve del mismo territorio. 3- Estudiar el efecto que los patrones sinópticos asociados a las diferentes fases de la Oscilación del Atlántico Norte (NAO) durante los meses de invierno tienen sobre los patrones espaciales de magnitud y duración del manto de nieve en la península ibérica. 4- Estudiar las diferencias en los valores de sensibilidad del manto de nieve respecto a la variabilidad de la temperatura, precipitación y radiación en onda corta a lo largo de un gradiente altitudinal sobre la península ibérica. 28 Introducción Cada uno de los próximos capítulos tiene como objetivo completar respectivamente cada uno de estos puntos, correspondiéndose con las cuatro etapas de la tesis doctoral. El último capítulo de la tesis doctoral está constituido por unas conclusiones generales de todo el trabajo, donde también se tratarán posibles líneas de investigación futuras. 29 Referencias: Alonso-González, E., López-Moreno, J.I., Navarro-Serrano, F., Sanmiguel-Vallelado, A., Revuelto, J., Domínguez-Castro, F., Ceballos, A., 2020. Snow climatology for the mountains in the Iberian Peninsula using satellite imagery and simulations with dynamically downscaled reanalysis data. Int. J. Climatol. 40, 477–491. https://doi.org/10.1002/joc.6223 Aran, M., Rigo, T., Bech, J., 2010. Analysis of the hazardous low-altitude snowfall, 8th March 2010, in Catalonia. … Sept. 1-4, 2010 … 12, 2010. Arenillas, M., Cobos, G., Navarro, J., 2008. Datos sobre la nieve y los glaciares en las cordilleras españolas. El programa ERHIN (1984--2008). Ed. Minist. Medio Ambient. y Medio Rural y Mar. Madrid. Armstrong, R., 2001. Historical Soviet daily snow depth version 2 (HSDSD). Natl. Snow Ice Data Center, Boulder, CO. CD-ROM. Barnett, T.P., Adam, J.C., Lettenmaier, D.P., 2005. Potential impacts of a warming climate on water availability in snow-dominated regions. Nature 438, 303–309. https://doi.org/10.1038/nature04141 Barnett, T.P., Dümenil, L., Schlese, U., Roeckner, E., 1988. The effect of eurasian snow cover on global climate. Science (80-. ). 239, 504–507. https://doi.org/10.1126/science.239.4839.504 Brooks, P.D., Williams, M.W., 1999. Snowpack controls on nitrogen cycling and export in seasonally snow-covered catchments. Hydrol. Process. 13, 2177–2190. Buisan, S.T., Saz, M.A., López-Moreno, J.I., 2015. Spatial and temporal variability of winter snow and precipitation days in the western and central Spanish Pyrenees. Int. J. Climatol. 35, 259–274. https://doi.org/10.1002/joc.3978 Ceballos-Barbancho, A., Llorente-Pinto, J.M., Alonso-González, E., López-Moreno, J.I., 2018. Dinámica del manto de nieve en una pequeña cuenca de montaña mediterránea: El caso del río tormes (cuenca del duero, españa)1. Rev. Geogr. Norte Gd. 2018, 9–34. https://doi.org/10.4067/s0718-34022018000300009 Clark, M.P., Serreze, M.C., Robinson, D.A., 1999. Atmospheric controls on Eurasian snow extent. Int. J. Climatol. 19, 27–40. https://doi.org/10.1002/(SICI)1097-0088(199901)19:1<27::AIDJOC346>3.0.CO;2-N Collados-Lara, A.J., Pardo-Igúzquiza, E., Pulido-Velazquez, D., 2020. Optimal design of snow stake networks to estimate snow depth in an alpine mountain range. Hydrol. Process. 34, 82–95. https://doi.org/10.1002/hyp.13574 Collados-Lara, A.J., Pardo-Igúzquiza, E., Pulido-Velazquez, D., 2017. Spatiotemporal estimation of snow depth using point data from snow stakes, digital terrain models, and satellite data. Hydrol. Process. 31, 1966–1982. https://doi.org/10.1002/hyp.11165 Corte‐Real, J., Zhang, X., Wang, X., 1995. Large‐scale circulation regimes and surface climatic anomalies over the Mediterranean. Int. J. Climatol. 15, 1135–1150. https://doi.org/10.1002/joc.3370151006 Déry, S.J., Brown, R.D., 2007. Recent Northern Hemisphere snow cover extent trends and implications for the snow-albedo feedback. Geophys. Res. Lett. 34. https://doi.org/10.1029/2007GL031474 Dietz, A.J., Kuenzer, C., Gessner, U., Dech, S., 2012. Remote sensing of snow - a review of available methods. Int. J. Remote Sens. 33, 4094–4134. https://doi.org/10.1080/01431161.2011.640964 Dunstone, N., Smith, D., Scaife, A., Hermanson, L., Eade, R., Robinson, N., Andrews, M., Knight, J., 2016. Skilful predictions of the winter North Atlantic Oscillation one year ahead. Nat. Geosci. 9, 809–814. https://doi.org/10.1038/ngeo2824 ECMWF, 2019. PART IV: PHYSICAL PROCESSES, in: IFS Documentation CY46R1, IFS Documentation. ECMWF, p. 4. Essery, R., Kontu, A., Lemmetyinen, J., Dumont, M., Ménard, C.B., 2016. A 7-year dataset for driving and evaluating snow models at an Arctic site (Sodankylä, Finland). Geosci. Instrumentation, Methods Data Syst. 5, 219–227. https://doi.org/10.5194/gi-5-219-2016 Esteban, P., Jones, P.D., Martín-Vide, J., Mases, M., 2005. Atmospheric circulation patterns related to heavy snowfall days in Andorra, Pyrenees. Int. J. Climatol. 25, 319–329. https://doi.org/10.1002/joc.1103 Fayad, A., Gascoin, S., Faour, G., López-Moreno, J.I., Drapeau, L., Page, M. Le, Escadafal, R., 2017. Snow hydrology in Mediterranean mountain regions: A review. J. Hydrol. 551, 374–396. https://doi.org/10.1016/j.jhydrol.2017.05.063 Gascoin, S., Hagolle, O., Huc, M., Jarlan, L., Dejoux, J.F., Szczypta, C., Marti, R., Sánchez, R., 2015. A snow cover climatology for the Pyrenees from MODIS snow products. Hydrol. Earth Syst. Sci. 19, 2337–2351. https://doi.org/10.5194/hess-19-2337-2015 Gilaberte-Búrdalo, M., López-Moreno, J.I., Morán-Tejeda, E., Jerez, S., Alonso-González, E., López-Martín, F., Pino-Otín, M.R., 2017. Assessment of ski condition reliability in the Spanish and Sanmiguel-Vallelado, A., Morán-Tejeda, E., Alonso-González, E., López-Moreno, J.I., 2017. Effect of snow on mountain river regimes: an example from the Pyrenees. Front. Earth Sci. 11, 515–530. https://doi.org/10.1007/s11707-016-0630-z Sanmiguel‐Vallelado, A., López‐Moreno, J.I., Morán‐Tejeda, E., Alonso‐González, E., Navarro‐ Serrano, F.M., Rico, I., Camarero, J.J., 2020. Variable effects of forest canopies on snow processes in a valley of the central Spanish Pyrenees. Hydrol. Process. https://doi.org/10.1002/hyp.13721 Scaife, A.A., Arribas, A., Blockley, E., Brookshaw, A., Clark, R.T., Dunstone, N., Eade, R., Fereday, D., Folland, C.K., Gordon, M., Hermanson, L., Knight, J.R., Lea, D.J., MacLachlan, C., Maidens, A., Martin, M., Peterson, A.K., Smith, D., Vellinga, M., Wallace, E., Waters, J., Williams, A., 2014. Skillful long-range prediction of European and North American winters. Geophys. Res. Lett. 41, 2514–2519. https://doi.org/10.1002/2014GL059637 Schaefer, G.L., Paetzold, R.F., 2001. SNOTEL (SNOwpack TELemetry) and SCAN (soil climate analysis network), in: Proc. Intl. Workshop on Automated Wea. Stations for Appl. in Agr. and Water Resour. Mgmt. Schimel, J.P., Clein, J.S., 1996. Microbial response to freeze-thaw cycles in tundra and taiga soils. Soil Biol. Biochem. 28, 1061–1066. Schweizer, J., Bartelt, P., van Herwijnen, A., 2015. Snow Avalanches, in: Snow and Ice-Related Hazards, Risks, and Disasters. Elsevier, pp. 395–436. https://doi.org/10.1016/B978-0-12-394849- 6.00012-3 Schweizer, J., Kronholm, K., Jamieson, J.B., Birkeland, K.W., 2008. Review of spatial variability of snowpack properties and its importance for avalanche formation. Cold Reg. Sci. Technol. 51, 253– 272. https://doi.org/10.1016/j.coldregions.2007.04.009 Shaw, T.E., Gascoin, S., Mendoza, P.A., Pellicciotti, F., McPhee, J., 2020. Snow Depth Patterns in a High Mountain Andean Catchment from Satellite Optical Tristereoscopic Remote Sensing. Water Resour. Res. 56, e2019WR024880. Skamarock, W.C., Klemp, J.B., Dudhia, J.B., Gill, D.O., Barker, D.M., Duda, M.G., Huang, X.-Y., Wang, W., Powers, J.G., 2008. A description of the Advanced Research WRF Version 3, NCAR Technical Note TN-475+STR. Tech. Rep. 113. https://doi.org/10.5065/D68S4MVH Takala, M., Luojus, K., Pulliainen, J., Derksen, C., Lemmetyinen, J., Kärnä, J.P., Koskinen, J., Bojkov, B., 2011. Estimating northern hemisphere snow water equivalent for climate research through assimilation of space-borne radiometer data and ground-based measurements. Remote Sens. Environ. 115, 3517–3529. https://doi.org/10.1016/j.rse.2011.08.014 Thompson, J.A., 2016. A modis-derived snow climatology (2000-2014) for the Australian Alps. Clim. Res. 68, 25–38. https://doi.org/10.3354/cr01379 van Pelt, W.J.J., Kohler, J., Liston, G.E., Hagen, J.O., Luks, B., Reijmer, C.H., Pohjola, V.A., 2016. Multidecadal climate and seasonal snow conditions in Svalbard. J. Geophys. Res. Earth Surf. 121, 2100–2117. https://doi.org/10.1002/2016JF003999 Vavrus, S., 2007. The role of terrestrial snow cover in the climate system. Clim. Dyn. 29, 73–88. https://doi.org/10.1007/s00382-007-0226-0 Vicente-Serrano, S.M., Rodríguez-Camino, E., Domínguez-Castro, F., El Kenawy, A., Azorín- Molina, C., 2017. An updated review on recent trends in observational surface atmospheric variables and their extremes over Spain. Cuad. Investig. Geogr. 43, 209–232. https://doi.org/10.18172/cig.3134 Vionnet, V., Brun, E., Morin, S., Boone, A., Faroux, S., Le Moigne, P., Martin, E., Willemet, J.M., 2012. The detailed snowpack scheme Crocus and its implementation in SURFEX v7.2. Geosci. Model Dev. 5, 773–791. https://doi.org/10.5194/gmd-5-773-2012 Viviroli, D., Dürr, H.H., Messerli, B., Meybeck, M., Weingartner, R., 2007. Mountains of the world, water towers for humanity: Typology, mapping, and global significance. Water Resour. Res. 43. https://doi.org/10.1029/2006WR005653 Wang, L., Ting, M., Kushner, P.J., 2017. A robust empirical seasonal prediction of winter NAO and surface climate. Sci. Rep. 7, 279. https://doi.org/10.1038/s41598-017-00353-y Winstral, A., Elder, K., Davis, R.E., 2002. Spatial snow modeling of wind-redistributed snow using terrain-based parameters. J. Hydrometeorol. 3, 524–538. https://doi.org/10.1175/1525- 7541(2002)003<0524:SSMOWR>2.0.CO;2 Wu, X., Che, T., Li, X., Wang, N., Yang, X., 2018. Slower Snowmelt in Spring Along With Climate Warming Across the Northern Hemisphere. Geophys. Res. Lett. 45, 12,331-12,339. https://doi.org/10.1029/2018GL079511 Wu, X., Shen, Y., Wang, N., Pan, X., Zhang, W., He, J., Wang, G., 2016. Coupling the WRF model with a temperature index model based on remote sensing for snowmelt simulations in a river basin in the Altay Mountains, north-west China. Hydrol. Process. 30, 3967–3977. https://doi.org/10.1002/ hyp.10924 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Capítulo 2: Rejillas de datos diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Resumen: En este trabajo presentamos observaciones de nieve y una base de datos diaria en rejilla que se generó a partir de un reanálisis al que previamente se le mejoro la resolución en la Península Ibérica. La península Ibérica presenta un manto de nieve estacional de larga duración en sus diferentes cordilleras, y en la mayor parte de su territorio se producen nevadas invernales. Sin embargo, solo hay observaciones directas limitadas de la profundidad de la nieve (SD) y del equivalente en agua de la nieve (SWE), lo que dificulta el análisis de la dinámica de la nieve y de los patrones espacio-temporales de las nevadas. Utilizamos los datos meteorológicos de los reanálisis como forzamiento de un modelo de balance de energía de la nieve de base física para simular el SWE y el SD en la Península Ibérica desde 1980 hasta 2014. Más concretamente, el reanálisis ERA-interim se redujo a una resolución de 10 km utilizando el modelo de Weather research and Forecast (WRF). Los resultados de WRF se utilizaron directamente, o como entrada a otros submodelos, para obtener los datos necesarios para lanzar el modelo Factorial Snow Model (FSM). Para mejorar la resolución de las salidas de WRF, utilizamos gradientes térmicos y ajustes higrobarométricos para simular series de nieve en bandas de elevación de 100 m por cada celda de 10 km en la Península Ibérica. Las series de nieve fueron validadas utilizando datos del sensor satelital MODIS y observaciones terrestres. La simulación de nieve reprodujo con precisión la variabilidad interanual del manto nivoso así como la variabilidad espacial de la acumulación y la fusión, incluso en terreno complejo. Así pues, el conjunto de datos presentado puede ser útil para muchas aplicaciones, entre ellas la ordenación del territorio, los estudios hidrometeorológicos, la fenología de la flora y la fauna, el turismo invernal y la gestión de riesgos. Los datos presentados aquí pueden descargarse gratuitamente de Zenodo (https://doi.org/10.5281/zenodo.854618). A continuación, se describe detalladamente el procedimiento, la validación de los datos, la evaluación de la incertidumbre y las posibles aplicaciones y limitaciones de la base de datos 43 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Cita completa: Alonso-González, E., López-Moreno, J. I., Gascoin, S., García-Valdecasas Ojeda, M., Sanmiguel- Vallelado, A., Navarro-Serrano, F., Revuelto, J., Ceballos, A., Esteban-Parra, M. J., and Essery, R.: Daily gridded datasets of snow depth and snow water equivalent for the Iberian Peninsula from 1980 to 2014, Earth Syst. Sci. Data, 10, 303–315, https://doi.org/10.5194/essd-10-303-2018, 2018. 44 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Daily gridded datasets of snow depth and snow water equivalent for the Iberian Peninsula from 1980 to 2014 Esteban Alonso-González1, J. Ignacio López-Moreno1, Simon Gascoin2, Matilde García-Valdecasas Ojeda3, Alba Sanmiguel-Vallelado1, Francisco Navarro-Serrano1, Jesús Revuelto4, Antonio Ceballos5, María Jesús Esteban-Parra3, and Richard Essery6 1. Instituto Pirenaico de Ecología, Consejo Superior de Investigaciones Científicas (IPE-CSIC), Zaragoza, Spain. 2. Centre d’Etudes Spatiales de la Biosphère (CESBIO), UPS/CNRS/IRD/CNES, Toulouse, France. 3. Departamento de Física Aplicada, Facultad de Ciencias, Universidad de Granada, Granada, Spain. 4. Météo-France – CNRS, CNRM (UMR3589), Centre d’Etudes de la Neige, Grenoble, France. 5. Dept. Geografía, Universidad de Salamanca, Salamanca, Spain 6School of GeoSciences. 6. University of Edinburgh, Edinburgh, UK. Correspondence: Esteban Alonso-González ([email protected]c.com) Received: 11 September 2017 – Discussion started: 20 October 2017 Revised: 28 December 2017 – Accepted: 8 January 2018 – Published: 20 February 2018 Abstract. We present snow observations and a validated daily gridded snowpack dataset that was simulated from downscaled reanalysis of data for the Iberian Peninsula. The Iberian Peninsula has long-lasting seasonal snowpacks in its different mountain ranges, and winter snowfall occurs in most of its area. However, there are only limited direct observations of snow depth (SD) and snow water equivalent (SWE), making it difficult to analyze snow dynamics and the spatiotemporal patterns of snowfall. We used meteorological data from downscaled reanalyses as input of a physically based snow energy balance model to simulate SWE and SD over the Iberian Peninsula from 1980 to 2014. More specifically, the ERA-Interim reanalysis was downscaled to 10 km resolution using the Weather Research and Forecasting (WRF) model. The WRF outputs were used directly, or as input to other submodels, to obtain data needed to drive the Factorial Snow Model (FSM). We used lapse rate coefficients and hygrobarometric adjustments to simulate snow series at 100 m elevations bands for each 10 km grid cell in the Iberian Peninsula. The snow series were validated using data from MODIS satellite sensor and ground observations. The overall simulated snow series accurately reproduced the interannual variability of snowpack and the spatial variability of snow accumulation and melting, even in very complex topographic terrains. Thus, the presented dataset may be useful for many applications, including land management, hydrometeorological 45 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. studies, phenology of flora and fauna, winter tourism, and risk management. The data presented here are freely available for download from Zenodo (https://doi.org/10.5281/zenodo.854618). This paper fully describes the work flow, data validation, uncertainty assessment, and possible applications and limitations of the database. 1 Introduction Seasonal snowpack exerts an important control on the hydrology and economy of many mountainous and cold regions worldwide (Barnett et al., 2005). Snow variability also affects different ecological processes, such as species composition, distribution, and phenology (Keller et al., 2000; Wipf et al., 2009). For example, snowpack on Mediterranean mountains is a crucial source of water during the dry season (Fayad et al., 2017; García-Ruiz et al., 2011; Viviroli et al., 2007). Long-term data are required to analyze the spatiotemporal dynamics of snowpack, to assess the importance of snow as a resource, and to understand the effect of climatic fluctuations. However, there are only limited in situ observations of snowpack for most mountain regions (Raleigh et al., 2016). Currently, remote-sensing techniques can only reliably provide information about snow cover based on observations in the visible spectrum (Dietz et al., 2012). Current spaceborne sensors do not provide accurate data on snow water equivalent (SWE) and/or snow depth (SD) in mountainous regions (Dozier et al., 2016). Microwave imaging has a coarse resolution (grid cell size: 25 km), so does not characterize snowpack variability in the Mediterranean mountains, which have a high spatial heterogeneity not captured with this resolution. There are also spatial and temporal limitations when attempting to estimate snowpack using closerange remote-sensing techniques such as LIDAR (Revuelto et al., 2016). There are limited in situ snow observations and meteorological data at high elevations in the Iberian Peninsula. Although the number of monitored sites has increased in recent years, there are no longterm series and there is in-sufficient characterization of snowpack dynamics at a regional scale. However, snowpack in the Iberian Peninsula is an important hydrological and also economical resource. An area of 19456.4 km2 in the Iberian Peninsula lies above 1500 m a.s.l., mostly in the five most important mountain ranges (Pyrenees, Cantabrian Mountains, Central System, Iberian Range, and Sierra Nevada). At this elevation, snow-pack occurs for at least 4 months of the year (López-Moreno et al., 2011) making it a critical resource for water management in the largest 46 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. hydrological basins (Morán-Tejeda et al., 2014). Snowpack influences the interannual variability of water resources (López-Moreno and García-Ruiz, 2004) and the timing of the winter low flows and spring peak flows (Sanmiguel-Vallelado et al., 2017). Moreover, winter tourism (mainly skiing) has been increasingly important to the economy of mountain valleys in recent decades, and the large interannual fluctuations of snowpack in the different mountain regions of the Iberian Peninsula affect the economic viability of tourism (Gilaberte-Búrdalo et al., 2014, 2017). The importance of snow to the environment and economy of the Iberian Peninsula, and the lack of data on snowpack in this region, motivated us to use meteorological outputs from downscaled reanalysis data to simulate snowpack at different elevations in the Iberian Peninsula. Atmospheric reanalyses, based on data assimilation and modeling (Saha et al., 2010), can provide important information about the temporal evolution of the atmosphere. Meteorological variables obtained from reanalysis data can be used as inputs for models of snow mass and energy balance which can be applied to describe the behavior of the snowpack over large areas (Brun et al., 2013; Krogh et al., 2015; Wegmann et al., 2017). However, the coarse resolution (cell size: around tens of kilo-meters) implies these simulations may have insufficient spatial resolution for characterizing the topographical complexity of mountain areas (Mass et al., 2002). To overcome this limitation, regional climate models (RCMs) are often used to obtain better representations of surface climatology, be-cause they downscale physically reanalysis products (García-Valdecasas Ojeda et al., 2017; Kryza et al., 2017; Warrach-Sagi et al., 2013). Previous studies have used RCMs to study SD and SWE dynamics at finer resolutions (grid cell size: 5 to 11 km) when they are driven with reanalyses, and the resolution increases further (grid cell size: 1 km) when using forecasted data (Bellaire et al., 2011; van Pelt et al., 2016; Quéno et al., 2016; Wu et al., 2016). van Pelt et al. (2016) used the High Resolution Limited Area Model (HIRLAM) in Svalbard (Norway), with forcing by ERA-40 and ERA-Interim reanalysis, and then used the meteorological simulation as driving data for SnowModel (Liston and Elder, 2006a). Their results support the usefulness of the methodology extracting snowpack trends from these data. Wu et al. (2016) used a similar procedure to describe the behavior of snowpack over the Altai Mountains in China. They coupled outputs from the Weather Research and Forecasting (WRF) model (Skamarock et al., 2008) driven by NCEP/NCAR reanalysis with a temperature index model (based on remote sensing), and their results had low error values. To increase the spatial resolution of the WRF out-puts, they used the MICROMET model (Liston and Elder, 2006b), a submodel of SnowModel in which WRF outputs are interpolated to a new grid, and then corrected physically according to topography. 47 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. First, we compared MODIS data with the SD and SWE time series (10 km resolution). MODIS snow maps were generated using the same workflow for each mountain range in the study area (Pyrenees, Cantabrian Mountains, Central System, Iberian Range, and Sierra Nevada). We downloaded all the available MOD10A1 and MYD10A1 products (version 5) from the National Snow and Ice Data Center (Hall et al., 2006). The original granules were mosaicked and reprojected from the sinusoidal system to the Universal Trans-verse Mercator (UTM) reference system. Then, we ran a gapfilling algorithm, using the binary snow product to avoid data losses due to cloud cover (Gascoin et al., 2015). This provided gap-free daily maps showing the presence and absence of snow in each mountain range from 2000 to 2014. From these maps, the probability of snow was calculated as follows: P(Snow)=Ns Nx100 where P(Snow) is the probability of snow (%), Ns is the number of days with snow, and N is the total number of days of the period. Snow probability maps were also calculated from the FSM snow cover maps. In this work, we chose a threshold of 0.11 m for SD and a threshold of 40 mm for SWE (Gascoin et al., 2015) in the FSM time series. This allowed us to generate snow cover maps from FSM outputs. Then, we aggregated the MODIS pixels (500 m) to the simulation grid ( 10 km), with averaging of the values of MODIS pixels to make them comparable. We also used data from 11 telenivometers, which measure subhourly SWE and SD using gamma ray attenuation and acoustic sensors. These data were provided by the ERHIN program (Estimación de Recursos Hídricos Proce-dentes de la Nieve) of the Hydrological Ebro River Basin Authority (Navarro-Serrano and López-Moreno, 2017). Ten telenivometers were located in the Pyrenees, and one in the Cantabrian Mountains. A complete description of the telenivometers and their locations can be found at www.saihhebro.com. We also used a SD sensor in the Central System mountain range (Durán et al., 2017), which is from the National Meteorological Agency of Spain (AEMET). We projected the meteorological variables from the WRF simulation to elevations of the different telenivometers for simulations. Figure 3 shows a comparison of the modeled and observed SD time series at these 10 sites. It must be noted that it is challenging to validate gridded products from ground-based data (Snauffer et al., 2016). Snowpack can have large variability over small distances (López-Moreno et al., 2015; 54 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Meromy et al., 2013). This implies that punctual measurements may not be representative of the 10 km resolution data, even when comparing a simulation at the same elevation as the telenivometer. In addition, snow measurements always include biases from the different measuring devices (Kinar and Pomeroy, 2015). Thus, we focused on the temporal patterns of snowpack during the season. More specifically, we compared the accumulation patterns during the season, assuming that accumulation and melting rates were similar in the simulated and observational data, but that SD and SWE likely differ between the telenivometer and the simulation. Thus, we first compared different percentiles of SD and SWE in the telenivometer and the simulated time series. Then, using each percentile as a threshold for snow presence, we converted the series into binary data, allowing use of the kappa test (Cohen, 1960) for each percentile. The kappa coefficient ranges from 1 and < 0, but it is difficult to assign an agreement criterion based on kappa value. Thus, we used the thresholds proposed by Landis and Koch (1977), which basically agree with values proposed by Fleiss et al. (1969; <0.00: poor; 0.00–0.20: slight; 0.21–0.4: fair; 0.41–0.60: moderate; 0.61–0.80: substantial; and 0.81–1.00: almost perfect). We examined percentile values between 10 and 90 %, as more representative of snow accumulation during the sea-son. 3 Results 3.1 Validation Our analysis of the probability of snow presence from MODIS and FSM shows that the outputs had good correlations (Fig. 4). This analysis compared the probability of snow at each pixel ( 10 km 10 km) from MODIS and FSM outputs for the SWE and SD time series from September 2000 to November 2014. The mean coefficient (R2) was 0.76, and a mean absolute error was 6.3 %. This analysis also shows the correlations for each mountain range, and the distribution of errors for SWE and SD (simulated less observed). These results also show there are no significant differences in the errors of P(Snow) for the different mountain ranges. However, the correlation was not strong for the Sierra Nevada range, probably due to its limited snow cover, although this remained inside the variability of the scatter plot. 55 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Validation of these results with telenivometers indicated kappa values for thresholds in the 10th to 90th percentiles of each season (Fig. 5). The kappa values were mostly above 0.6, although accuracy declined for the highest percentiles. 56 Figure 3: Comparison between modeled (red) and observed (black) SD time series for each telenivometer and the Cotos SD sensor. Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. The kappa coefficient does not account for the displacement magnitude of the different percentiles, and a difference of a few days at the time of peak accumulation may cause a sharp decrease in the kappa value. This is the reason for the loss of accuracy at the highest percentiles. Thus, we further analyzed these data to determine the time of the year when snowpack exceeded the 90th, 75th, and 50th percentiles at each telenivometer in the observed (OBS) and simulated (SIM) series (Fig. 5c). This analysis shows that, despite small temporal shifts, the simulated snow series accurately represents the temporal patterns when different snow percentiles are exceeded. The biggest shift in the position of the 90th and 75th percentiles was during the 2011/2012 season. This season was extremely dry on the Iberian Peninsula, and there were very few snowfall events (Fig. 3). Thus, a small bias in the simulation of a single event during this time could lead to a large error in prediction of the magnitude and timing of SD and SWE maxima. 57 Figure 4: Correlation between the long-term (2000–2015) mean probability of snow depth (a) and snow water equivalent (b) from MODIS data and from FSM output. Box plot insets show the frequency distributions of errors (%), with the central red lines indicating average errors, boxes indicating the 25th and 75th percentiles, bars indicating the 10th and 90th percentiles, and dots indicating the 5th and 95th percentiles. Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. 3.2 Gridded snow dataset: applications and limitations The final products of the models are daily gridded datasets (resolution: 0.088º, ~10 km) of SD and SWE at elevations from 500 to 2900 m a.s.l. (100 m intervals) from 1980 to 2014. The datasets (ncdf4 format) cover the entire Iberian Peninsula, including the north side of the Pyrenees in France. Each dataset contains information of the entire Iberian Peninsula and a mask that covers pixels that do not present areas at the elevations of the simulation estimated from a 250 m resolution DEM. 58 Figure 5: Kappa values derived from comparison of observed and simulated series for different percentiles of snow depth (a) and snow water equivalent (b), and periods of the year (blue) when snowpack exceeds the 90th, 75th, and 50th percentiles (c). In (c), each pair of bands shows the times when the different percentiles in the observed (OBS) and simulated (SIM) series at each telenivometer exceeded the indicated percentile. Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. This snow database provides new opportunities for studies of snow in the Iberian Peninsula. In particular, the temporal resolution and the duration of the series show significant improvements over previous observational data. Also, the geographic data generated on SD and SWE provides the opportunity to obtain more snow and hydrologically relevant information than available from remote sensing alone. It is also possible to develop different snow products at different elevations, allowing for comparison of different elevations and different regions. For example, Fig. 6 shows the long-term average interannual maximum SD and SWE at three different elevations. 59 Figure 6: Long-term (1980–2014) average maximum SWE and SD grids at 1500, 2000, and 2500 m a.s.l. Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Figure 7 shows examples of other snow variables that can be derived from the database: average number of snowfall events and percentage of days with snow cover at three elevations. These analyses are particularly useful for the development of different snow climatologies for the whole Iberian Peninsula, or for specific areas, in studies that rely on ecological data (e.g., phenology or distribution of plants and animals, forest growth), studies that require hydrological parameters for different catchments, and studies that determine risk maps for snow-related events. It is also possible to extract daily time series for different areas or elevations at each pixel. For example, Fig. 8 compares SWE series at three elevations in the pixel at the highest peak of the Pyrenees (Aneto Peak, 3404 m a.s.l.). Thus, these series allows for the study of different annual snow accumulation and melting patterns at a specific location and how elevation influences snow 60 Figure 7: Long-term (1980–2014) average number of snowfall events and percentage of snow presence at 1500, 2000, and 2500 m a.s.l. Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. evolution. Similarly, it enables one to study the existence of temporal trends or the occurrence of extreme snowfall and melting events. The database contains uncertainties that are not easy to quantify, due to the limited amount of observational data. Biases may be due to uncertainty of the boundary conditions from the ERA- Interim reanalysis (Chaudhuri et al., 2013) since errors from the WRF downscaling model are difficult to quantify in mountain areas (Gutmann et al., 2012), and uncertainties that typically result from simulations of snow mass and energy balance from meteorological data (Essery et al., 1999, 2013; Magnusson et al., 2015). The use of the standard air temperature lapse rate could also be a source of uncertainty. Although other studies have observed a decrease in the lapse rate during winter months, this effect is result of thermic inversions that are not considered to be due to the spatial resolution of the simulation. Despite these limitations, we had very satisfactory results when testing the duration and the interannual variability of the snowpack against MODIS and telenivometer data, which provided reliable observations during several snow seasons. This way, the database presents a reliable validation for more than a third of the time period generated. When using this database, it is important to consider that it was based on the assumption of flat topography within each 10 km 10 km pixel. Therefore, this dataset is not suitable for studies of snow variability due to terrain aspect, slope, and snow redistribution processes, such as avalanches and wind transport. 61 Figure 8: Comparison of SWE time series at 1500, 2000, and 2500 m a.s.l. at Aneto Peak. Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. 4. Data availability The data presented here are freely available for download from Zenodo (https://doi.org/10.5281/zenodo.854618). SD and SWE datasets are in ncdf4 format, with one file for each elevation band. The observational information used to validate the main data is also available for download. All telenivometer data are in CSV format. Daily snow cover (derived from MODIS) is provided as five multiband GeoTiff files (one file for each mountain range, each band is a date), and a CSV file indicates the date of each band. The FSM code is freely available from https://github.com/ RichardEssery/FSM (Essery, 2015). 5. Conclusions We presented a new daily gridded database of SD and SWE for the Iberian Peninsula from 1980 to 2014 period at a resolution of 0.088 ( 10 km). The database consists of 50 ncdf4 files for SD and SWE from 500 to 2900 m a.s.l., and an-other 2 files of WRF simulation DEMs, summing more than 652 000 maps. A mask label of “no data” is included if the grid is not found at the elevation of the simulated elevation band. The scarcity of snow observations in the Iberian Peninsula made it necessary to couple a dynamic downscaling of ERA-Interim reanalysis using the WRF model by use of a snow energy and mass balance model (FSM). Input data of FSM provided directly, or estimated from WRF outputs, were avail-able for the average elevation of each 10 km 10 km pixel, and these data were transformed to achieve an elevation off-set at 100 m intervals. Despite some uncertainties, the database is consistent with available observational data. More specifically, validation with MODIS data indicated an error of 6.07 % and an R2 of 0.76 from the analysis of the mean presence of snow. The database also provides good representation of the temporal patterns of the telenivometers, with kappa values generally over 0.6, and above 0.4 for all analyzed percentiles. This database will be an important resource for studies of many different hydrological, environmental, and economic processes in Mediterranean areas. Thus, we expect the database presented here will be useful for future snow-related studies at regional scales on the Iberian Peninsula, and for a broad community of researchers and land managers working in areas where snowfall occurs. 62 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Acknowledgements. Esteban Alonso-González is supported by the Spanish Ministry of Economy and Competitiveness (BES-2015-071466). This study was funded by the Spanish Ministry of Economy and Competitiveness projects CGL2014-52599-P 10 (Estudio del manto de nieve en la montaña española y su respuesta a la variabilidad y cambio climatico) and CGL2017-82216-R (HIDROIBERNIEVE) and (with additional support from the European Community funds, FEDER) CGL2013-48539-R (Impactos del cambio climático en los recursos hídricos de la cuenca del Duero a alta resolución). Also, the Regional Government of Andalusia has funded this research with the project P11-RNM-7941 (Impactos del Cambio Climático en la cuenca del Guadalquivir, LICUA). The authors would like to express thanks to Hydrological Ebro River Basin Authority (CHE) for providing telenivometer data. Development of FSM is supported by NERC grant NE/P011926/1. Cotos snow data were provided by Consejería de Medio Ambiente, Administración Local y Ordenación del Territorio de la Comunidad de Madrid, from the meteorological network of Parque Natural de Peñalara. The authors sincerely thank Jan Magnusson for his help on the first steps of the use of FSM code. 63 Rejillas diarias de espesor de nieve y equivalente en agua de nieve de la Península Ibérica desde 1980 hasta 2014. Viviroli, D., Dürr, H. H., Messerli, B., Meybeck, M., and Wein-gartner, R.: Mountains of the world, water towers for humanity: Typology, mapping, and global significance, Water Resour. Res., 43, W07447, https://doi.org/10.1029/2006WR005653, 2007. Walcek, C. J.: Cloud Cover and Its Relationship to Relative Humidity during a Springtime Midlatitude Cyclone, Mon. Weather Rev., 122, 1021–1035, https://doi.org/10.1175/1520- 0493(1994)122<1021:CCAIRT>2.0.CO;2, 1994. Warrach-Sagi, K., Schwitalla, T., Wulfmeyer, V., and Bauer, H.-S.: Evaluation of a climate simulation in Europe based on the WRF– NOAH model system: precipitation in Germany, Clim. Dynam., 41, 755–774, https://doi.org/10.1007/s00382-013-1727-7, 2013. Wegmann, M., Orsolini, Y., Dutra, E., Bulygina, O., Sterin, A., and Brönnimann, S.: Eurasian snow depth in long-term climate reanalyses, The Cryosphere, 11, 923–935, https://doi.org/10.5194/tc-11- 923-2017, 2017. Wipf, S., Stoeckli, V., and Bebi, P.: Winter climate change in alpine tundra: plant responses to changes in snow depth and snowmelt timing, Climatic Change, 94, 105–121, https://doi.org/10.1007/s10584-009-9546-x, 2009. Wrzesien, M. L., Durand, M. T., Pavelsky, T. M., Howat, I. M., Margulis, S. A., and Huning, L. S.: Comparison of Methods to Estimate Snow Water Equivalent at the Mountain Range Scale: A Case Study of the California Sierra Nevada, J. Hydrometeo-rol., 18, 1101–1119, https://doi.org/10.1175/JHM-D-16-0246.1, 2017. Wu, X., Shen, Y., Wang, N., Pan, X., Zhang, W., He, J., and Wang, G.: Coupling the WRF model with a temperature index model based on remote sensing for snowmelt simulations in a river basin in the Altay Mountains, north-west China, Hydrol. Pro-cess., 30, 3967–3977, https://doi.org/10.1002/hyp.10924, 2016. 70 Climatología de nieve de las montañas de la Península Ibérica Capítulo 3: Climatología de nieve de las montañas de la Península Ibérica mediante imagenes por satélite y simulaciones con mejoras de escala dinámicas. Resumen: La presencia de un manto de nieve estacional determina la hidrología, geomorfología y ecología de amplias zonas de la Península Ibérica, con grandes implicaciones en la economía, el transporte y la gestión de riesgos. Por esta razón, es necesario contar con información fiable desde un punto de vista operacional y científico. Este es el caso de la Península Ibérica, donde la falta de datos ha impedido el estudio de la duración del manto de nieve, magnitud así como su variabilidad interanual. En este trabajo presentamos la primera climatología del manto de nieve de toda la península Ibérica. La escasez de observaciones in situ se ha solucionado mediante el uso de una nueva base de datos generada mediante información procedente del sensor satelital MODIS (2000- 2014) y el modelo de base física Factorial Snow Model (FSM), forzado por las salidas del modelo climático regional Weather Research and Forecast (WRF) sobre la Península Ibérica cubriendo el periodo 1980-2014. El manto de nieve de las principales áreas montañosas (Pirineos, Cordillera Cantábrica, Sistema Central, Sistema Ibérico y Sierra Nevada) es descrito, estimado de la base de datos generada. La información del manto de nieve ha sido procesada usando un algoritmo de agrupamiento k-means, buscando similitudes en el manto de nieve a diferentes alturas. Los resultados han encontrado cuatro tipos diferentes de manto de nieve en términos de profundidad, duración y variabilidad interanual sobre las diferentes bandas altitudinales demostrando la variabilidad del manto de nieve de la península Ibérica. Los análisis han revelado zonas con nevadas efímeras en contraste con áreas con duraciones medias de 198 días de nieve al año y 3m de profundidad. Los coeficientes de variación del manto de nieve oscilaron entre 35,2 y 162,4%. Todos los índices analizados indican que la Cordillera Cantábrica y los Pirineos tienen los mantos de nieve más profundos y de mayor duración, seguidos por el Sistema Central e Ibérico. Sierra Nevada mostró el manto de nieve de menor duración y menso profundo con una mayor variabilidad interanual. 73 Climatología de nieve de las montañas de la Península Ibérica Cita completa: Alonso‐González, E., López‐Moreno, J. I., Navarro‐Serrano, F., Sanmiguel‐Vallelado, A., Revuelto, J., Domínguez‐Castro, F., & Ceballos, A. (2020). Snow climatology for the mountains in the Iberian Peninsula using satellite imagery and simulations with dynamically downscaled reanalysis data. Int J Climatol. 2020; 40: 477– 491. https://doi.org/10.1002/joc.6223 74 Climatología de nieve de las montañas de la Península Ibérica Snow climatology for the mountains in the Iberian Peninsula using satellite imagery and simulations with dynamically downscaled reanalysis data Esteban Alonso-González1, Juan Ignacio López-Moreno1, Francisco Navarro-Serrano1, Alba Sanmiguel-Vallelado1, Jesús Revuelto2, Fernando Domínguez-Castro1, Antonio Ceballos3 1. Instituto Pirenaico de Ecología, Consejo Superior de Investigaciones Científicas (IPE-CSIC), Zaragoza, Spain 2. Météo-France – CNRS, CNRM (UMR3589), Centre d'Etudes de la Neige, Grenoble, France 3. Dept. Geografía, Universidad de Salamanca, Salamanca, Spain Correspondence: Esteban Alonso-González ([email protected]c.com) Received: 03 Diciembre 2018 Accepted: 02 July 2019 Abstract. The presence of a seasonal snowpack determines the hydrology, geomorphology and ecology of wide parts of the Iberian Peninsula, with strong implications for the economy, transport and risk management. Thus, reliable information on snow is necessary from a scientific and operational point of view. This is the case of the Iberian Peninsula where, lack of observation has impeded proper analysis of snow-pack duration, magnitude and interannual variability. In this study, we present the first snow climatology of the entire Iberian Peninsula. The scarcity of in situ observations has been overcome, using a newly developed remote sensing snow data-base from MODIS satellite sensors for the period 2000–2014 and a physically based snow model (Factorial Snow Model—FSM), driven by a regional atmospheric model (Weather Research and Forecast model—WRF) over the Iberian Peninsula for the period 1980–2014. The snowpack of the main mountain areas (Pyrenees, Cantabrian, Central, Iberian range and Sierra Nevada) are described, estimated from the generated databases. The information has been processed using a k-means cluster algorithm, looking for similarities in snow indices at different elevation bands. Results show four different types of snowpack in terms of depth, duration and interannual variability, lying over different elevation bands in the different ranges, proving the variability of the snowpack over Iberia. Analyses reveal areas characterized by ephemeral snowpacks, while in some sectors snowpack lasts, on average, 198 days per year with 3.02 m of peak snow depth. The coefficient of variation of interannual peak snow depth oscillated between 35.2 and 162.4%. All the analysed indices show that at common elevations the Cantabrian range and the Pyrenees host the deepest and longest 75 Climatología de nieve de las montañas de la Península Ibérica lasting snowpacks, followed by the Central and Iberian ranges. The Sierra Nevada exhibits the shortest, shallowest snowpack and more year-to-year variability. 1 Introduction Snowpack has significant implications for many environmental processes at high latitudes and in most of the mountainous areas of the world. Thus, snow directly affects geomorphological processes (Carrera-Gómez and Valcárcel, 2018; González Trueba and Serrano, 2010), nutrient transport (Jones, 1999), plants and animal phenology (García-Hernández et al., 2016; Slatyer et al., 2017) and the surface energy balance (Lund et al., 2017). It also may become a risk source associated with heavy snowfall, avalanches, floods triggered by rain on snow events (Garvelmann et al., 2015; Würzer and Jonas, 2018), snow melt induced land-slides, cause transport problems or damage to infrastructure (García et al., 2009; Chueca Cía et al., 2014; García-Hernández et al., 2018). These events cause economic losses and even loss of life (Stethem et al., 2003; Höller, 2007; Pfeifer et al., 2018). Furthermore, seasonal snowpack exerts a critical role in various surface and groundwater hydrological processes (Viviroli et al., 2007), especially in those areas, such as the Mediterranean region, where the wet sea-son coincides with the snow accumulation period (Fayad et al., 2017). Additionally, snow, as it relates to tourism, has become a main economic support of mountain areas world-wide (Elsasser and Bürki, 2002). Thus, appropriate management of risks and natural resources of mountainous areas requires spatially comparable information on the duration, magnitude and temporal variability of the snowpack. The orography of the Iberian Peninsula is complex, with several mountainous areas and two important elevated pla-teaus that make it one of the highest average elevation regions in Europe (Casas-Sainz and de Vicente, 2009). Snowfall can occur over most of its territory, even at the lowest elevations (Arndt et al., 2010; Bech et al., 2013). As an example, from 1947 to 2009, 16 snowfall events were recorded in Barcelona (13 m a.s.l.) and 18 in Gerona (69 m a.s.l.) areas (Aran et al., 2010). Few snow events are recorded at the coastal elevations of the Valencian Community during the snow season (Mora et al., 2016). Snowfall is not rare over the main cities of northern Spain during the winter, causing associated transportation risks and traffic collapse (Merino et al., 2014). Despite these occasional events at low elevation, a continuous seasonal snowpack appears only in the mountains, generally higher than 1,300–1,500 m a.s.l. (López-Moreno et al., 2009). 76 Climatología de nieve de las montañas de la Península Ibérica Mountainous areas host the headwaters of the main rivers of the Iberian Peninsula. Thus, snow exerts a strong hydrological influence on downstream sectors (López and Justribó, 2010), controlling the interannual variability of river flows; López-Moreno and García-Ruiz, 2005) and reshaping the annual hygrograms. This has obvious implications on water management (Morán- Tejeda et al., 2014; Sanmiguel-Vallelado et al., 2017), which is expected to become more challenging in the future (Viviroli et al., 2011) due to climatological and landscape changes (García- Ruiz et al., 2011; Vicente-Serrano et al., 2014). The economy of many regions in the Iberian Peninsula has become dependent on the interannual variability of snow, which makes these regions highly vulnerable to climate change, since snow duration in this area is largely connected to climate warming (Gilaberte-Búrdalo et al., 2014, 2017). There are some studies describing the behavior of the snowpack over different mountain ranges of Iberia. More specifically, most of these studies are focused on the Pyrenees and, to a lesser degree, in the Sierra Nevada (i.e., Herrero et al., 2011; Revuelto et al., 2014; Gascoin et al., 2015; López- Moreno et al., 2015; Quéno et al., 2016; Navarro-Serrano and López-Moreno, 2017) with just some mention of other mountain ranges in Iberia where snow plays an important role (i.e., Palacios et al., 2003; Serrano Cañadas et al., 2016; Soteres García et al., 2016; Ceballos-Barbancho et al., 2018). Most of the snow studies are limited by the limited snow data available. Topographical complexity and the spatial and temporal variability of the snowpack at different scales make it challenging to characterize the snowpack at regional scale from field measurements (López-Moreno et al., 2011a, 2013), which are, in turn, very scarce in the Iberian mountains. Daily snow observations in Spain rely on observatories from the National Meteorological Agency (AEMET), but information is generally restricted to the presence or absence of snow, less often providing snow depth (SD) (Buisan et al., 2015). In the last years, AEMET and some hydrological administrations have installed automatic stations at higher elevations. Despite the interest in this new information, available series are too short and too sparsely distributed over terrain to be used for generating sound regional information (Revuelto et al., 2017). Remote sensing data can be useful for retrieving information on snow presence over large areas, but in most cases this information is limited to the snow cover area. Obtaining SD and Snow Water Equivalent (SWE) data from remote sensing techniques is an active research topic. However, this information is not always available, and there are not yet products able to retrieve information on snow depth or SWE under complex topography and at a spatial and, more importantly, temporal 77 Climatología de nieve de las montañas de la Península Ibérica scale useful for scientific and managerial purposes. The use of numerical mesoatmospheric models coupled online (Wrzesien et al., 2018) or offline (van Pelt et al., 2016; Wu et al., 2016) with snow modelling systems has proven consistent, replicating interannual and intraannual snowpack variability, allowing the development of consistent snow climatologies (Wrzesien et al., 2017; Revuelto et al., 2018). Recently, Alonso-González et al. (2018) generated the first snow database for the whole of Iberia. The database was made by combining remote sensing data (Moderate-Resolution Imaging Spectroradiometer, MODIS) for the period 2000–2014 with simulations from a snow energy balance model (Factorial Snow Model, FSM) driven by the out-puts of a mesoatmospheric climate model (Weather Research and Forecast, WRF) at 0.088 (~10 km) resolution for the period 1980– 2014. This was further rescaled at different elevation bands to estimate SD and SWE. The objectives here are: (a) to illustrate and quantify the main differences of the snow phenomena across the Iberian Peninsula; and (b) to characterize the snow cover duration, magnitude of the snow pack and its interannual variability exhibited in the last decades. 2 Study area The Iberian Peninsula has a complex topography, dominated by two large central plateaus surrounded for several mountain ranges that exceed 2,000 m a.s.l. These mountain areas are well represented by five principal mountain ranges (Pyrenees, Cantabrian Mountains, Central, Iberian and Sierra Nevada; see Table 1 for a detailed description of position and elevation of each range) where seasonal snowpack develops every year. The Sierra Nevada (Mulhacén peak, 3,478 m a.s.l.) and the Pyrenees (Aneto peak, 3,404 m a.s.l.) mountain ranges reach the higher elevations of the Peninsula. 78 Climatología de nieve de las montañas de la Península Ibérica Table 1: Maximunm elevation, location and area at different elevation ranges of the main mountain ranges of Iberia. Surface (Km2) Study area Max. elevation 1000 - 1500 1500 – 2000 2000 – 2500 >2500 Cantabrian 2648 11397 3211 143 <1 Central 2592 9772 2116 218 <1 Iberian 2314 7198 918 26 0 Sierra Nevada 3479 1510 902 388 190 Pyrenees 3404 9318 5560 3543 836 The distribution of the mountain ranges runs generally from west to east, localized at different latitudes and at different distances to the sea (Figure 1). This particular distribution and the contrasted air masses that affect the Iberian Peninsula (transition from Atlantic to Mediterranean patterns) generate different climatological areas over Iberia (Lopez-Bustins et al., 2008). The climatological particularities of each mountain range affect the energy balance of the snowpack in different ways, generating different behaviours of the snowpack and different responses to climate warming (López-Moreno et al., 2017a). All these different conditions also ensure the snowpack will exhibit very different duration and thicknesses at the different elevation bands when the mountain ranges are compared. In addition, the hypsometry of each mountain range (Figure 1) differs considerably in what may have noticeable impacts on the importance of the snow phenomena. Thus, the Pyrenees and Sierra Nevada exhibits much larger areas above 1,500–2,000 m where seasonal and long-lasting snowpacks generally exists. 79 Climatología de nieve de las montañas de la Península Ibérica Table 2: Mean values of P(Snow) (%) and averaged interannual coefficient of variation CV (%) for each mountain range at common elevation bands (m a.s.l.). 1000 m a.s.l. 1500 m as.l. 2000 m as.l. 2500 m as.l. 2900 m as.l. P(Snow) CV P(Snow) CV P(Snow) CV P(Snow) CV P(Snow) CV Cantabrian 4.3 60.6 19.0 33.1 37.0 13.7 50.1 6.2 - - Central 1.5 32.1 7.0 43.3 27.8 19.8 - - - - Iberian 2.2 44.5 13.7 26.9 35.1 11.1 - - - - Sierra Nevada 0.9 47.9 3.7 53.1 13.8 34.2 33.5 18.0 50.9 6.9 Pyrenees 5.8 64.3 17.6 48.0 38.3 21.5 51.7 9.6 64.0 4.74 The inter-comparison at 2,000 m a.s.l. elevation band highlights again the great spatial variability of the Pyrenees mentioned above, the longer persistence of the snowpack in the Pyrenees, Cantabrian range and Iberian Range, and the shorter persistence of the snowpack in the Central range and, in particular, in the Sierra Nevada. The differences between the Pyrenees and the Sierra Nevada are also very marked at 2900 m a.s.l (Figure 4f), especially at the beginning and end of the snow season, when P(Snow) values have at least a 1-month lag, showing a much later snowpack in the Pyrenees. 86 Climatología de nieve de las montañas de la Península Ibérica There is a general trend in all mountain ranges toward lower CV (interannual variability) as elevation increases (Figure 5). This is particularly evident around 1,000 m a.s.l., related to the occurrence of frequent snowfall but also to the ephemeral nature of the snowpack (Grünewald et al., 2014). As shown for P(Snow), the Pyrenees, Cantabrian and Iberian range show similar distribution and mean CV values, with a noticeable increase in the variability for the Central range that becomes similar to the other ranges at 2,000 m a.s.l. The Sierra Nevada is the mountain range that shows the greatest values of CV at all elevations above 1,500 m a.s.l. (i.e., ~16% more than in the Pyrenees at 2,000 m a.s.l.). Indeed, at the highest elevations in the Sierra Nevada, CV is still higher than that observed in the other mountains at 2,000 m a.s.l., indicating that the interannual variability of P(Snow) is related to the climatic dynamics of each mountain range. In the particular case of the Sierra Nevada, the very strong dependency of winter precipitation on the occurrence of negative phases of NAO (López-Moreno et al., 2011c) explains the very large interannual variability found at this mountain range. Although the effect of NAO has obvious implications in the variability of different snow indexes, its complexity exceeds the objectives of this work, making necessary further research in this topic. 87 Figure 4: (a-e) Monthly distribution of values of P (Snow) from MODIS products for mountain ranges at 2,000 m a.s.l.; (f) distribution of values of P (Snow) from MODIS products for the Pyrenees and the Sierra Nevada at 2900 m a.s.l. (Pyrenees in purple and Sierra Nevada in green). Climatología de nieve de las montañas de la Península Ibérica 4.2 Simulated snow SD and SWE There is a general trend in all the mountain ranges of delay in the peak SWE date with elevation (Table 3), due to delay in the onset of melt and spring precipitation falling as snowin the highest portions of the mountain ranges (Navarro-Serrano et al., 2018). Peak SWE shows the highest values in the Cantabrian Range for the maximum comparable elevation band (2,400 m a.s.l.) with 1,687 mm, followed by the Pyrenees with 713 mm. The Central range shows values of 475 mm and the Sierra Nevada shows only 164 mm. At the highest elevations (over 2,900 m a.s.l.), the Pyrenees and the Sierra Nevada show noticeable differences in SWE peak values, with 1,313 and 656 mm, respectively. 88 Figure 5: (a-e) Distribution of values of CV per elevations over the main mountain ranges of Iberia from MODIS data; (f) Intercomparison of mean values of CV for each mountain range. The colours of the scatterplots indicates the relative density of points, ranging from blue (lower density) to red (higher density). Climatología de nieve de las montañas de la Península Ibérica Table 3: Average peak SWE and average date of maximum SWE, for different representative elevation bands for the main mountain ranges of the Iberian Peninsula. Elevation Pyrenees S. Nevada Cantabrian Central Iberian Avg. Max. SWE (mm) 1400 99.4 ~0 77.5 ~0 ~0 1900 347.5 8.3 471.5 119.9 150.1 2400 713.2 184.7 1686.9 * * 2900 1313.4 655.0 * * * Date max (date) 1400 08-Mar 0 01-Mar 0 0 1900 08-Mar 30-Jan 01-Mar 22-Feb 23-Feb 2400 18-Apr 01-Mar Apr-26 06-Mar * 2900 14-Apr 13-Apr * * * To properly interpret the results in terms of snow abundance at each mountain range, it is important to consider the hypsography of the ranges (see Section 2. Study area, Figure 1 and Table 1). As an example, the Cantabrian range shows the highest values for average SWE peak (Figure 6), but it has 143 km2 lying between 2,000 and 2,500 m a.s.l, compared with 3,543 km2 for the Pyrenees. This means that even though there are certain areas in the Cantabrian range with maximum values of SWE peak for the whole Iberian Peninsula, the total amount of snow is much larger in the Pyrenees. As a consequence of the extension and climatological variability of the Pyrenees at 2400 m a.s.l., this value of averaged SWE peak (713.2 mm) hides a great dispersion over the mountain range (from a maximum of 1895.9 mm to a minimum 94.54 mm). 89 Figure 6: Averaged SWE series at different elevation bands for each mountain range of the Iberian Peninsula in the hydrological year (from 1 October to 30 September) based on FSM simulations. Climatología de nieve de las montañas de la Península Ibérica From the k-means cluster analysis, four different snow-pack behaviours over Iberian Peninsula were found with different averaged values of elevation (Z), mean snow season duration (Duration), mean maximum SD (Maximum), and maximum SD coefficient of variability (SDCV) (Table 4). These four behaviours can be summarized as: Cluster 1: No snow is detected. Cluster 2: Transition mid-mountain areas where insignificant amounts of snow can be found. The slightly higher values of peak SD and Duration in conjunction with a high value of SDCV, suggests that occasional snowfalls can occur. Cluster 3: Mountain environments with significant amounts of snow most seasons. The SDCV added to the peak SD values suggests that years with shallow snowpacks can occur. Cluster 4: High mountain environments where greats amount of snow can be found every year. The SDCV is lower than previous clusters, which means that, together with a much higher Maximum SD, we found important amounts of snow lying in these areas in every season. The cluster groups show geographical consistency along the Iberian Peninsula (Figure 7) but with significant differences in the distribution of P(Snow) values above 1,000 m a.s.l. (Figure 8). Cluster 1 is restricted to the low elevated portions of the most Mediterranean areas, with the exception of the western limits of the Central range Cluster 2 contains most of the areas below 1,000 m a.s.l. of the Iberian Peninsula, with the exception of the Sierra Nevada, where cluster 1 is dominant at the same eleva-tions, and in the southwestern Pyrenees and northeastern Central range. The distribution of cluster 3 shows an evident decline with latitude. The Cantabrian range and the Pyrenees show significant percentages belonging to this cluster (25 and 32% of its area over 1,000 m a.s.l.), at an average elevation lower than other mountain ranges (~1,550 m a.s.l.). Table 4: Averaged values of elevation, snow probability, maximum SD and CV for each cluster group. Cluster group Z ma.s.l. Duration (%) Maximum (m) SDCV (%) 1 1111 0.1 0.1 162.4 2 1281 21.0 0.2 81.2 3 1667 24.3 1.1 49.7 4 2197 54.1 3.0 35.3 90 Climatología de nieve de las montañas de la Península Ibérica The 4 cluster is found mostly in the highest elevations in the Iberian Peninsula, namely in the Pyrenees and the Sierra Nevada. In the Iberian range, it is not possible to find this cluster at any elevation. In the Central range, it is possible to find it only in small areas at the highest elevations in the southwest (Figure 7). In the Cantabrian range, it is possible to find areas associated with the 4th cluster at the lowest elevation of the Iberian Peninsula. Thus, in the northern areas of the Cantabrian range, it is possible to find cluster 4 from 1,450 m a.s.l. to the maximum elevation (Torre Cerredo peak 2,648 m a.s.l.) (Figure 8). The reason for this small percentage of surface (2.2%) covered by cluster 4 in the Cantabrian range is the hypsography of the range (Figure 1), as the southern part of the range shows a great surface over 1,000 m a.s.l. Cluster 4 is only found at the highest elevations of the Sierra Nevada (above 2,500 m a.s.l.) with little variability in the values of elevation. The Pyrenees show the highest percentage of area associated with cluster 4 (17.27%). This cluster is 91 Figure 7: Spatial distribution of cluster groups over 1,000 m a.s.l. Climatología de nieve de las montañas de la Península Ibérica found above 1,300 m a.s.l. in the northwestern sectors, increasing to 2,900 m a.s.l. in the southeastern sectors. The different distribution of elevation values at each cluster for each mountain range highlights the complexity of snowpack characteristics over Iberia. This distribution is strongly related with latitude but also with other effects as north to south precipitation patterns that have strong implications even in the Central range that is relatively far from the sea (Durán et al., 2013). The Pyrenees shows the highest scatter of elevation values for each cluster group, due to the high spatial variability in the precipitation patterns (Lemus-Canovas et al., 2019), that varies along the longitudinal axe (see Section. 4.3). The higher influence of the Atlantic winds in the Cantabrian range (as in the north-western part of the Pyrenees), explains the lower values of elevation at each cluster compared with other mountain ranges (Navarro-Serrano and López-Moreno, 2017). The 92 Figure 8: Distribution of elevation values (over 1,000 m a.s.l.) for each mountain range per cluster group. The values at the top of the boxplots show percentage of the surface of the range included in each cluster. Climatología de nieve de las montañas de la Península Ibérica Sierra Nevada shows the higher values of elevation for each cluster (Figure 8), likely due to its more southerly latitude. 4.3 Intra-range variability from simulated snow data The snowpack behaviour inside each mountain range differs over the Iberian Peninsula (Figure 9). The Cantabrian range shows less variability than the Pyrenees but with a notice-able north-to-south gradient, from its northern part with higher P(Snow) values to the southern part with lower P(Snow) values. This gradient is the result of the strong rain shadow effect that occurs in the lee side of this range (south face) (Ortega Villazán and Morales Rodríguez, 2015). In the Pyrenees, the longitudinal and latitudinal P(Snow) gradients are much larger; despite the existence of a north (higher values) to south (lower values) transition. This is consistent with the results of other analysis in this study, as a consequence of the large climatological differences found in this sec-tor of the Iberian Peninsula (Buisan et al., 2015; Navarro-Serrano and López-Moreno, 2017). The strongest gradient can be found in an oblique line from the southeast, where the snow probability values are much lower, to the northwest, which is affected by the Mediterranean- Atlantic gradient (east to west). There were strong north–south precipitation differences induced by the blocking effect of the range to the very common advections from the north and northwest during winter months (Navarro-Serrano and López-Moreno, 2017). 93 Figure 9: Spatial patterns of mean P(snow) based on FSM simulations for Cantabrian range (a), Pyrenees (b), central range (c) at 2,000 m a.s.l. and for Sierra Nevada (d) at 2500 m a.s.l. Climatología de nieve de las montañas de la Península Ibérica Over the Central range and the Sierra Nevada, the gradients are lower than in the Pyrenees or even the Cantabrian range. The main reason for this is their reduced size and lack of high values of P(Snow). For the Central range, it is possible to find a longitudinal gradient from the northeast, with the highest values in the southwest. The latter is one of the wettest spots in the Iberian Peninsula (García et al., 2017). Despite the reduced size of the Sierra Nevada, P(Snow) values are higher in the northern part of the range due to the precipitation and incoming radiation gradients (Huete-Morales et al., 2018). 4.4 Implications of the presented results Numerical modelling techniques and remote sensing data have some uncertainties that should be considered for proper interpretation of the presented results. MODIS products have limitations in snow and cloud detection (Wang et al., 2016). Forest existence below the subalpine belts are another source of uncertainty for snow detection (Simic et al., 2003). Further-more, gap-filling products used here could introduce uncertainties, as the process makes some necessary assumptions and a regression tree in its last step (Gascoin et al., 2015). The estimation of uncertainty on the distribution of precipitation over complex terrain in numerical mesoatmospheric models as WRF is still an active research topic (i.e., Jing et al., 2017; Gerber et al., 2018). Moreover, the methodology proposed by Alonso-González et al. (2018) makes some assumptions, such as constant temperature lapse-rates, which is a common practice in the distribution of meteorological data (Liston and Elder, 2006). As a consequence of the high computational cost of numerical weather simulations (Gutmann et al., 2016), final resolution of the outputs was not enough to resolve the local particularities of the meteorological variables involved in the energy and mass balance of the snowpack. The subpixel snowpack redistribution due to topographic factors such as slope, curvature or wind sheltering (López-Moreno et al., 2017b) was not considered in this study. The smoothing of the topography can also cause some problems in accurately reproducing interactions between precipitation and terrain complexity due to the constant precipitation along the elevation assumption, as noted by Alonso-González et al. (2018). Despite these uncertainties, the same study illustrates the consistency of the database with in situ observations for both reproducing thickness and duration of the snowpack, but especially for accurately reproducing the temporal (seasonal and interannual) patterns of the snowpack, which gives the data set reliability for developing regional scale studies (Alonso-González et al., 2018). 94 Climatología de nieve de las montañas de la Península Ibérica 5 Conclusions This work has shown the high variability of snowpack characteristics over the different mountain ranges of Iberia. Pyrenees and Cantabrian range store the deepest and long lasting snowpack in the Iberian Peninsula having similar averaged snowpack behaviours. Nevertheless, Pyrenees has the greatest snowpack spatial variability of all the Iberian ranges. This is as a consequence of the high climatological variability induced by the long longitudinal axis and the strong climatological gradients from Atlantic to Mediterranean conditions and north–south slopes. On the contrary Sierra Nevada reveals the shallowest and ephemeral snow-pack, where the accumulation and duration is similar to the Pyrenees and Cantabrian ranges with a downward offset of at least 500 m a.s.l. The Iberian and Central ranges have shown intermediate accumulation and duration patterns but their snowpack tend to be closer to the observed in the Pyrenees and Cantabrian ranges compared to Sierra Nevada. Specifically compared to Cantabrian and Pyrenees, Iberian range shows similar duration values whit lower values of accumulation and Central range shows lower values of accumulation and duration. The differences between mountain ranges cannot be only explained by the latitudinal gradient, other drivers as continentality and precipitation patterns play also an important role. Acknowledgements. Esteban Alonso-González is granted with a pre-doctoral FPI grant by the Spanish Ministry of Economy and Competitive-ness (BES-2015-071466). This study was funded by the Spanish Ministry of Economy and Competitiveness projects CGL2014-52599-P 10 (“Estudio del manto de nieve en la montaña española y su respuesta a la variabilidad y cambio climatico) and CGL2017-82216-R (HIDROIBERNIEVE). Navarro-Serrano, F. and Sanmiguel-Vallelado, A. are granted with a pre-doctoral FPU grant (Spanish Ministry of Education, Culture and Sports). J. Revuelto benefited from a grant within the above-cited PAPI project and is now supported by a Post-doctoral Fellowship of the AXA research fund (le Post-Doctorant Jesús Revuelto est bénéficiaire d'une bourse postdoctorale du Fonds AXA pour la Recherchem Ref: CNRM 3.2.01/17). All the snow simu-lated and remote sensing data is freely downloadable at https://zenodo.org/record/854619. 95 Climatología de nieve de las montañas de la Península Ibérica Morán-Tejeda, E., Lorenzo-Lacruz, J., López-Moreno, J.I., Rahman, K. and Beniston, M. (2014) Streamflow timing of mountain rivers in Spain: recent changes and future projections. Journal of Hydrology, 517, 1114–1127. https://doi.org/10.1016/j.jhydrol.2014.06.053. Navarro-Serrano, F. and López-Moreno, J.I. (2017) Spatio-temporal analysis of snowfall events in the Spanish pyrenees and their rela-tionship to atmospheric circulation | Análisis espacio-temporal de los eventos de nevadas en el pirineo Español y su relación con la circulación atmosférica. Cuadernos de Investigacion Geografica, 43(1), 233–254. https://doi.org/10.18172/cig.3042. Navarro-Serrano, F., López-Moreno, J.I., Azorin-Molina, C., Alonso-González, E., Tomás- Burguera, M., Sanmiguel-Vallelado, A., Revuelto, J. and Vicente-Serrano, S.M. (2018) Estimation of near-surface air temperature lapse rates over continental Spain and its mountain areas. International Journal of Climatology, 38, 3233–3249. https://doi.org/10.1002/joc.5497. Ortega Villazán, M.T. and Morales Rodríguez, C.G. (2015) El clima de la Cordillera Cantábrica castellano-leonesa: diversidad, contrastes y cambios. Investigaciones Geográficas, 0(63), 45–67. https://doi. org/10.14198/INGEO2015.63.04. Palacios, D., de Andrés, N. and Luengo, E. (2003) Distribution and effectiveness of nivation in Mediterranean mountains: Peñalara (Spain). Geomorphology, 54(3–4), 157–178. https://doi.org/10. 1016/S0169-555X(02)00340-9. Pfeifer, C., Höller, P. and Zeileis, A. (2018) Spatial and temporal analy-sis of fatal off-piste and backcountry avalanche accidents in Austria with a comparison of results in Switzerland, France, Italy and the US. Natural Hazards and Earth System Sciences, 185194, 571–582. https://doi.org/10.5194/nhess-18-571-2018. Quéno, L., Vionnet, V., Dombrowski-Etchevers, I., Lafaysse, M., Dumont, M. and Karbou, F. (2016) Snowpack modelling in the Pyrenees driven by kilometric-resolution meteorological forecasts. The Cryosphere, 10(4), 1571–1589. https://doi.org/10.5194/tc-10- 1571-2016. Revuelto, J., Azorin-Molina, C., Alonso-González, E., Sanmiguel-Vallelado, A., Navarro-Serrano, F., Rico, I. and Ignacio López-Moreno, J. (2017) Meteorological and snow distribution data in the Izas experimental catchment (Spanish Pyrenees) from 2011 to 2017. Earth System Science Data, 9(2), 993–1005. https://doi.org/ 10.5194/essd-9-993-2017. Reuelto, J., Lecourt, G., Lafaysse, M., Zin, I., Charrois, L., Vionnet, V., Dumont, M., Rabatel, A., Six, D., Condom, T., Morin, S., Viani, A. and Sirguey, P. (2018) Multi-criteria evalua-tion of 102 Climatología de nieve de las montañas de la Península Ibérica snowpack simulations in complex alpine terrain using satel-lite and in situ observations. Remote Sensing, 10(8), 1171. https:// doi.org/10.3390/rs10081171. Revuelto, J., López-Moreno, J.I., Azorin-Molina, C. and Vicente-Serrano, S.M. (2014) Topographic control of snowpack distribution in a small catchment in the central Spanish Pyrenees: intra- and inter-annual persistence. Copernicus, 8(5), 1989–2006. https://doi. org/10.5194/tc-8-1989-2014. Sanmiguel-Vallelado, A., Morán-Tejeda, E., Alonso-González, E. and López-Moreno, J.I. (2017) Effect of snow on mountain river regimes: an example from the Pyrenees. Frontiers of earth Science, 11, 515–530. https://doi.org/10.1007/s11707-016-0630-z. Serrano Cañadas, E., Gómez Lende, M. and Pisabarro Pérez, A. (2016) Nieve y riesgo de aludes en la montaña cantábrica: el alud de car-daño de arriba, alto carrión (Palencia). Polígonos. Revista de Geografía, 0(28), 239. https://doi.org/10.18002/pol.v0i28.4295. Simic A, Fernandes R, Brown R, Romanov P, Park W, Hall DK. (2003) Validation of MODIS, VEGETATION, and GOES+SSM/I snow cover products over Canada based on surface snow depth observations. IGARSS 2003. 2003 IEEE International Geoscience and Remote Sensing Symposium. Proceedings (IEEE Cat. No.03CH37477). IEEE, 836–838. https://doi.org/10.1109/IGARSS. 2003.1293936. Skamarock WC, Klemp JB, Dudhia J, Gill DO, Barker DM, Dudha MG, Huang X, Wang W, Powers Y. (2008) A Description of the Advanced Research WRF Version 3. NCAR Technical Note NCAR/TN-475+STR. https://doi.org/10.5065/D68S4MVH. Slatyer, R.A., Nash, M.A. and Hoffmann, A.A. (2017) Measuring the effects of reduced snow cover on Australia's alpine arthropods. Aus-tral Ecology, 42(7), 844–857. https://doi.org/10.1111/aec.12507. Soteres García, R.L., Pedraza Gilsanz, J. and Carrasco González, R.M. (2016) Cartografía de susceptibilidad y estimación del máximo alcance de aludes en el circo de gredos (sistema central ibérico). Polígonos. Revista de Geografía, 0(28), 265. https://doi.org/10. 18002/pol.v0i28.4296. Stethem, C., Jamieson, B., Schaerer, P., Liverman, D., Germain, D. and Walker, S. (2003) Snow avalanche Hazard in Canada – a review. Natural Hazards, 28(2/3), 487–515. https://doi.org/10.1023/ A: 1022998512227. 103 Climatología de nieve de las montañas de la Península Ibérica van Pelt, W.J.J., Kohler, J., Liston, G.E., Hagen, J.O., Luks, B., Reijmer, C.H. and Pohjola, V.A. (2016) Multidecadal climate and seasonal snow conditions in Svalbard. Journal of Geophysical Research: Earth Surface, 121(11), 2100–2117. https://doi.org/10. 1002/2016JF003999. Vicente-Serrano, S.M., Lopez-Moreno, J.-I., Beguería, S., Lorenzo-Lacruz, J., Sanchez-Lorenzo, A., García-Ruiz, J.M., Azorin-Molina, C., Morán-Tejeda, E., Revuelto, J., Trigo, R., Coelho, F. and Espejo, F. (2014) Evidence of increasing drought severity cau-sed by temperature rise in southern Europe. Environmental Research Letters, 9(4), 044001. https://doi.org/10.1088/1748-9326/ 9/4/044001. Viviroli, D., Archer, D.R., Buytaert, W., Fowler, H.J., Greenwood, G. B., Hamlet, A.F., Huang, Y., Koboltschnig, G., Litaor, M.I., López-Moreno, J.I., Lorentz, S., Schädler, B., Schreier, H., Schwaiger, K., Vuille, M. and Woods, R. (2011) Climate change and mountain water resources: overview and recommendations for research, man-agement and policy. Hydrology and Earth System Sciences, 15(2), 471–504. https://doi.org/10.5194/hess-15-471-2011. Viviroli, D., Dürr, H.H., Messerli, B., Meybeck, M. and Weingartner, R. (2007) Mountains of the world, water towers for humanity: typol-ogy, mapping, and global significance. Water Resources Research, 43(7), W07447. https://doi.org/10.1029/2006WR005653. Wang, T., Fetzer, E.J., Wong, S., Kahn, B.H. and Yue, Q. (2016) Vali-dation of MODIS cloud mask and multilayer flag using CloudSat-CALIPSO cloud profiles and a cross-reference of their cloud classi-fications. Journal of Geophysical Research: Atmospheres, 121(19), 11,620–11,635. https://doi.org/10.1002/2016JD025239. Wrzesien, M.L., Durand, M.T., Pavelsky, T.M., Howat, I.M., Margulis, S.A., Huning, L.S., Wrzesien, M.L., Durand, M.T., Pavelsky, T.M., Howat, I.M., Margulis, S.A. and Huning, L.S. (2017) Comparison of methods to estimate snow water equivalent at the mountain range scale: a case study of the California Sierra Nevada. Journal of Hydrometeorology, 18(4), 1101–1119. https:// doi.org/10.1175/JHM-D-16-0246.1. Wrzesien, M.L., Durand, M.T., Pavelsky, T.M., Kapnick, S.B., Zhang, Y., Guo, J. and Shum, C.K. (2018) A new estimate of north American Mountain snow accumulation from regional climate model simulations. Geophysical Research Letters, 45(3), 1423–1432. https://doi.org/10.1002/2017GL076664. 104 Climatología de nieve de las montañas de la Península Ibérica Wu, X., Shen, Y., Wang, N., Pan, X., Zhang, W., He, J. and Wang, G. (2016) Coupling the WRF model with a temperature index model based on remote sensing for snowmelt simulations in a river basin in the Altay Mountains, north-West China. Hydrological Processes, 30(21), 3967–3977. https://doi.org/10.1002/hyp.10924. Würzer, S. and Jonas, T. (2018) Spatio-temporal aspects of snowpack runoff formation during rain on snow. Hydrological Processes, 32 (23), 3434–3445. https://doi.org/10.1002/hyp.13240. 105 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica Capítulo 4: Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica Resumen: La Oscilación del Atlántico Norte (NAO) es considerada el principal factor atmosférico que explica el clima invernal y la evolución de la nieve en gran parte del hemisferio norte. Sin embargo, la ausencia series largas del manto de nieve en las regiones montañosas de la península Ibérica ha impedido una evaluación completa del impacto de la NAO a escala regional en esta zona. En este estudio, evaluamos la relación entre la NAO de los meses de invierno (DJFM-NAO) y el manto de nieve de la Península Ibérica. Para ello, hemos simulado la temperatura, precipitación y nieve para el período 1979-2014 mediante reducción dinámica de la resolución de los datos del reanálisis ERA-Interim, correlacionando nuestras simulaciones con el índice DJFM-NAO en las cinco principales cadenas montañosas de la Península Ibérica (Cordillera Cantábrica, Sistema Central, Sistema Ibérico, Pirineos y Sierra Nevada). Los resultados confirmaron que los valores negativos del DJFM-NAO generalmente ocurren durante condiciones húmedas y templadas en la mayor parte de la Península Ibérica. Debido a la dirección de las masas de aire húmedo inducidas por las fases negativas de la NAO, esta tiene una gran influencia en la duración de la nieve y en el máximo anual de SWE en la mayoría de las cordilleras del estudio, sobre todo en las laderas al sur del eje principal de las cordilleras. En cambio, el impacto de la variabilidad de la NAO es limitado en las laderas orientadas al norte. Los valores negativos (positivos) de DJFM-NAO se asociaron a una duración más larga (más corta) y a picos más altos (más bajos) de SWE en todas las montañas analizadas en el estudio. Encontramos una marcada variabilidad en las correlaciones del índice DJFM-NAO con los índices de nieve dentro de cada cadena montañosa, incluso cuando sólo se consideraron las laderas orientadas al sur. Las correlaciones encontradas fueron más elevadas en las mayores elevaciones de las cordilleras, pero la longitud geográfica también explicaba la variabilidad dentro de las cordilleras en la mayoría de las montañas estudiadas. 109 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica Cita completa: Alonso-González, E.; López-Moreno, J.I.; Navarro-Serrano, F.M.; Revuelto, J. Impact of North Atlantic Oscillation on the Snowpack in Iberian Peninsula Mountains. Water 2020, 12(1), 105; https://doi.org/10.3390/w12010105 110 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica Impact of North Atlantic Oscillation on the Snowpack in Iberian Peninsula Mountains Esteban Alonso-González , Juan I. López-Moreno, Francisco M. Navarro-Serrano and Jesús Revuelto Instituto Pirenaico de Ecología, CSIC. Campus de Aula Dei, Av. Montañana 1005, 50059 Zaragoza, Correspondence: Esteban Alonso-González ([email protected]c.com) Received: 11 November 2019 Accepted: 20 December 2019 Abstract: The North Atlantic Oscillation (NAO) is considered to be the main atmospheric factor explaining the winter climate and snow evolution over much of the Northern Hemisphere. However, the absence of long-term snow data in mountain regions has prevented full assessment of the impact of the NAO at the regional scales, where data are limited. In this study, we assessed the relationship between the NAO of the winter months (DJFM-NAO) and the snowpack of the Iberian Peninsula. We simulated temperature, precipitation, and snow data for the period 1979–2014 by dynamic downscaling of ERA-Interim reanalysis data, and correlated this with the DJFM-NAO for the five main mountain ranges of the Iberian Peninsula (Cantabrian Range, Central Range, Iberian Range, the Pyrenees, and the Sierra Nevada). The results confirmed that negative DJFM-NAO values generally occur during wet and mild conditions over most of the Iberian Peninsula. Due to the direction of the wet air masses, the NAO has a large influence on snow duration and the annual peak snow water equivalent (peak SWE) in most of the mountain ranges in the study, mostly on the slopes south of the main axis of the ranges. In contrast, the impact of NAO variability is limited on north-facing slopes. Negative (positive) DJFM-NAO values were associated with longer (shorter) duration and higher (lower) peak SWEs in all mountains analyzed in the study. We found marked variability in correlations of the DJFM-NAO with snow indices within each mountain range, even when only the south-facing slopes were considered. The correlations were stronger for higher elevations in the mountain ranges, but geographical longitude also explained the intrarange variability in the majority of the studied mountains. 111 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica The DJFM-NAO and temperature were generally positively correlated. During negative DJFMNAO phases the occurrence of more frequent cyclonic circulations brings lower temperatures than during anticyclonic periods, but the most frequent wind directions during negative DJFM-NAO phases has a clear Atlantic influence, which generally brings mild temperatures during snowfall events. This causes a marked increase of the elevation where precipitation changes from liquid to solid phase compared to other synoptic situations over Iberia (López-Moreno, 2005; Navarro- Serrano and López-Moreno, 2017). The correlation between monthly accumulated precipitation and the NAO index was negative for the entire Iberian Peninsula, as previously reported by other authors over central Portugal (Trigo et al., 2005) or in the northwestern part of Spain (Fernández-González et al., 2012), as well as in the main mountain ranges of the Iberian Peninsula (Juan I. López-Moreno et al., 2011)(Figure 2); although, for small Mediterranean and Cantabrian areas, the correlations were negligible (Ríos- Cornejo et al., 2015; Rodrigo et al., 2000). Southwest areas showed the greatest negative correlations (values of approximately 0.8), which is consistent with the findings of Munoz-Díaz and Rodrigo (2004), where they found different areas of correlation of the precipitation with the NAO index over the Iberian Peninsula and, more specifically, Queralt et al. (2009) about extreme precipitation events associated with DJFM-NAO phases. There was differentiation between the highly correlated southern slopes and the less correlated northern slopes, as previously reported for the Spanish Pyrenees by Vada et al. (2013) and Buisan et al. (2015), and for the Alps by Stefanicki et al. (1998). This differentiation was especially evident for the mountain ranges located at higher elevations (i.e., the Cantabrian Range and the Pyrenees). The differences between the southward and northward slopes of the mountain ranges were clear for the Pyrenees (north approximately 0.1; south approximately 0.7), the Cantabrian Range (north approximately 0.3; south approximately 0.6), and the Iberian Range (north approximately 0.2; south approximately 0.6). The differences were less marked for the Central Range (north approximately 0.4; south approximately 0.5), and for the Sierra Nevada were almost negligible. This effect is caused likely by the geographical position of Sierra Nevada and Central Range, and its limited potential to be an effective topographical barrier against the westerly and southwesterly wind directions associated to negative DJFM-NAO phases. 118 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica 4.3. NAO Spatial Influence on Peak SWE and Snow Season Duration Figure 3 shows that the DJFM-NAO has a clear correlation with the interannual variability of the snow duration on the southern slopes of the Pyrenees and the Cantabrian, Central, and Iberian ranges. The correlation of the DJFM-NAO with snow duration was high and statistically significant at the highest elevations of the Sierra Nevada, but with no statistically significant differences between the north- and south-facing slopes. However, the other mountain areas showed statistical significance (p-value < 0.05; Wilcoxon–Mann–Whitney test) between north- and south-facing slopes. This finding is consistent with the spatial patterns shown in Section 3.2, between the NAO and the interannual variability of winter temperature and precipitation. The spatial patterns and magnitude of the correlations for peak SWE were very similar to those for snow duration (Supplementary Figure S1). However, there were differences between these snow variables. Figure 4 shows the absolute difference in the Pearson’s correlation coefficients of the DJFM-NAO with the snow duration and the peak SWE for each pixel for the south-facing slopes, 119 Figure 3: Pearson's R values between de DJFM-NAO and the snowpack duration. The lines represent the boundaties of the main catchments in the Iberian Peninsula. Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica where high correlation values were found (except for the Sierra Nevada, for which all pixels were considered). The greatest differences were found for the higher elevations in the Pyrenees and the Sierra Nevada, where the correlations for peak SWE were higher than those for snow duration. These two mountain ranges exhibit large areas above 2000 m a.s.l, in contrast with the other mountain ranges. We hypothesize that, at high elevations, the correlation of the DJFM-NAO with the temperature and precipitation leads to high levels of snow accumulation during the DJFM-NAO phases, as a consequence of the driving mechanism that controls the DJFM-NAO; however, this is not reflected in the duration of the snow cover. This is mainly because snow duration also depends on solid precipitation occurring during April and May, which is not related to the DJFM-NAO. In addition, the very late and rapid melting that occurs in these high elevation areas (Musselman et al., 2017) is mostly due to the marked increase of the incoming solar radiation at the end of the snow season period. Figure 4 shows that there was large spatial variability in the correlation coefficients between the DJFM-NAO and the snow indices within each of the analyzed mountain ranges. The influence of elevation on the distribution of the DJFM-NAO correlation values between snow duration and the peak SWE (on south-facing slopes, except in the Sierra Nevada, where both slopes were 120 Figure 4: Absolute differences in the Pearson's R statistic for correlations of the DJFM-NAO with the snow duration and the peak SWE. The colors in the scatterplots indicate the relative density of points, ranging from blue (lower relative density) to red (higher relative density). Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica considered) is shown in Figure 5, and the comparable influence of geographical longitude is shown in Figure 6. 121 Figure 5: Influence of elevation on the correlation of the DJFM-NAO with snow duration and the peak SWE. Red lines indicate no statistical significance of the correlations of the random samples (p-value > 0.05), while green lines show the statistically significance correlation). Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica For the Cantabrian Range, most of the correlations of the DJFM-NAO with snow duration ranged from 0.25 to 0.75, and with peak SWE ranged from 0 to 0.75. This variability showed no relationship to elevation, but a clear link with longitude was evident, with an increase in correlation westward. For the Pyrenees, most correlations ranged from 0 to 0.75 for snow duration and peak SWE. While the response of snow duration to elevation was unclear, there was a strong negative relationship for peak SWE. The increase in the correlation of the DJFM-NAO on snow magnitude at high elevations has previously been reported by López-Moreno and Vicente-Serrano (2007), where they found statistically significant correlations of the DJFM-NAO with snowpack observations above 1700 m a.s.l., being consistent with our results. These findings suggest that the frequent westerly and southwesterly air fluxes during negative DJFM-NAO years bring mild temperatures during snowfall events, which may cause rain precipitation at low and mid elevations, but heavy snowfalls over 1800–2000 m a.s.l. The absence of a vertical gradient on the relationship of the DJFM-NAO on snow duration is explained by the presence of snow cover and snowfalls until late spring (end of May to early June); these are also highly dependent on spring precipitation, which is unrelated to the DJFM-NAO. Figure 6 does not show statistically significant correlations with longitude but reveals a clear pattern for both indices, characterized by the highest correlations 122 Figure 6: Influence of geographical longitude on the correlation of the DJFM-NAO with snow duration and the peak SWE. Red lines indicate no statistical significance of the correlations of the random samples (p-value > 0.05), while green lines show the statistically significance correlation). Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica (values reaching 0.75) for the Central Pyrenees, and a decrease westward and eastward. This effect is likely a consequence of the small surface with elevation above 1800 m a.s.l. in the western part compared to the Central Pyrenees, causing low correlation between DJFM-NAO and snow indices (Figure 5). The also-low correlations at the easternmost part are explained because southwestern air fluxes, which are frequent during negative DJFM-NAO phases and generally affect only the Western and Central Pyrenees. Thus, DJFM-NAO are not clearly related with both temperature and precipitation in this sector (Figure 2). In the Central Range, the correlation with the DJFM-NAO ranged from 0 to 0.75 for snow duration and 0.1 to 0.6 for peak SWE. This variability showed a clear negative trend with elevation for both indices. There was no strong relationship with longitude for the correlations between the snow indices and the DJFM-NAO for the Central Range. For the Iberian Range, most of the correlations with the DJFM-NAO ranged from 0.1 to 0.6 for snow duration, and 0 to 0.75 for peak SWE. The reduced area at high elevations and their limited geographical extent made it difficult to distinguish clear spatial patterns in the relationship of the DJFM-NAO and the snow indices. The Sierra Nevada showed the largest variability in the correlation values for both indices, ranging from slightly positive values to 0.75. For both indices the impact of the DJFM-NAO was clearer at higher elevations and eastward in this southern mountain range. This study shows that the relationship of the DJFM-NAO and the snowpack can be seen in longterm trends of snow cover over wide areas of the Iberian Peninsula. Thus, very different trends in snow accumulation have been found in the Pyrenees when different periods have been analyzed. López-Moreno (2005) found a statistically significant decrease in snow accumulation in the central Spanish Pyrenees for the period 1950–1999, but Buisan et al. (2015) found no significant trends for the period 1985–2014. Such disagreements were explained by strong decadal fluctuations in the snow data, which correlated with the DJFM-NAO. The marked spatial differences observed in the correlation between the DJFM-NAO and snow indices may explain the marked contrast in snow trends over very short distances, as has been observed in the Pyrenees (Morán-Tejeda et al., 2017; Revuelto et al., 2013). The presence of areas showing strong correlations between the snowpack and the DJFM-NAO, independent of the study period (as demonstrated by Buisan et al. 2015 for the Pyrenees), provides the promise of improving the seasonal forecasting of snowpack based on short- and medium-term projections for the DJFM-NAO from climate models (Dunstone et al., 2016). This could improve optimization of water management and benefit the skiing industry, both of 123 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica which are highly affected by interannual fluctuations in snow accumulation and the duration of the snowpack. 5. Conclusions The DJFM-NAO is related to the atmospheric circulation patterns over the Iberian Peninsula during the snow-accumulation period. The NAO is related to contrasts in precipitation and temperature over the Iberian Peninsula, and hence related to the snow-accumulation and melting processes. The mountain ranges analyzed in this study are topographic barriers to the main westerly and southwesterly air fluxes related to the DJFM-NAO and explain their variable impacts on the snowpack. Thus, the snowpack on north-facing slopes along the main mountain range axes did not show statistically significant correlations in most of its surface, but there were significant correlations in south-facing slopes. The Sierra Nevada was the only mountain range where this north south difference was not observed. For all Iberian mountain areas, the statistically significant correlations between the DJFM-NAO and snow indices were negative, and, in some cases, R values exceeded 0.7. Marked spatial differences were found, even for those slopes where the DJFM-NAO generally affects the temporal evolution of the snow. In most of the mountain areas, the strength of the relationship between the DJFM-NAO and peak SWE increased with elevation. This also occurred for snow duration, except in the Pyrenees, where spring snowfall can have a strong influence on the snow duration at high elevations, and snowmelt occurs later and more rapidly. This explains why the snowpack duration and peak SWE are not always related, and explains the weaker correlations between the DJFM-NAO and snow duration compared with peak SWE. In several mountain areas, we found spatial differences that were related to geographical longitude. This was evident for the Cantabrian Range and the Sierra Nevada, where the correlation values decreased from west to east. For the Pyrenees, the maximum correlations were found for the central valleys and decreased toward its western and eastern edges. The results of this study have clear implications for understanding the effect of the length of the study period on the detection of robust long-term trends in snow data, and also spatial differences in trend analyses over short distances. 124 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica Supplementary Materials.The following are available online at https://www.mdpi.com/2073- 4441/12/1/105/s1, Figure S1: Pearson’s R values between the DJFM-NAO and the peak SWE. The lines represent the boundaries of the main catchments in the Iberian Peninsula. Acknowledgements. Esteban Alonso-González is the recipient of a pre-doctoral FPI grant by the Spanish Ministry of Economy and Competitiveness (BES-2015-071466). This study was funded by the Spanish Ministry of Economy and Competitiveness project CGL2017-82216-R (HIDROIBERNIEVE). Francisco Navarro-Serrano is the recipient of a pre-doctoral FPU grant (Spanish Ministry of Education, Culture and Sports). All the simulation data are freely downloadable at https://zenodo.org/record/854619. 125 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica References Alonso-González, E., Ignacio López-Moreno, J., Gascoin, S., García-Valdecasas Ojeda, M., Sanmiguel-Vallelado, A., Navarro-Serrano, F., Revuelto, J., Ceballos, A., Esteban-Parra, M.J., Essery, R., 2018. Daily gridded datasets of snow depth and snow water equivalent for the Iberian Peninsula from 1980 to 2014. Earth Syst. Sci. Data 10, 303–315. https://doi.org/10.5194/essd-10- 303-2018 Alonso-González, E., López-Moreno, J.I., Navarro-Serrano, F., Sanmiguel-Vallelado, A., Revuelto, J., Domínguez-Castro, F., Ceballos, A., 2020. Snow climatology for the mountains in the Iberian Peninsula using satellite imagery and simulations with dynamically downscaled reanalysis data. Int. J. Climatol. 40, 477–491. https://doi.org/10.1002/joc.6223 Bednorz, E., 2002. Snow cover in western Poland and macro-scale circulation conditions. Int. J. Climatol. 22, 533–541. https://doi.org/10.1002/joc.752 Bednorz, E., Wibig, J., 2016. Spatial distribution and synoptic conditions of snow accumulation in the Russian Arctic. Polar Res. 35, 25916. https://doi.org/10.3402/polar.v35.25916 Beniston, M., 2003. Climatic change in mountain regions: A review of possible impacts. Clim. Change 59, 5–31. https://doi.org/10.1023/A:1024458411589 Beniston, M., 1997. Variations of snow depth and duration in the Swiss Alps over the last 50 years: Links to changes in large-scale climatic forcings, in: Climatic Change. Springer, pp. 281–300. https://doi.org/10.1007/978-94-015-8905-5_3 Berrisford, P., Dee, D., Poli, P., Brugge, R., Fielding, K., Fuentes, M., Kallberg, P., Kobayashi, S., Uppala, S., Simmons, A., 2009. The ERA-Interim Archive Version 2.0. ERA Rep. Ser. Brown, R.D., Petkova, N., 2007. Snow cover variability in Bulgarian mountainous regions, 1931- 2000. Int. J. Climatol. 27, 1215–1229. https://doi.org/10.1002/joc.1468 Buisan, S.T., López-Moreno, J.I., Saz, M.A., Kochendorfer, J., 2016. Impact of weather type variability on winter precipitation, temperature and annual snowpack in the Spanish Pyrenees. Clim. Res. 69, 79–92. https://doi.org/10.3354/cr01391 126 Impacto de la Oscilación del Atlántico Norte en el manto de nieve de las montañas de la Península Ibérica Buisan, S.T., Saz, M.A., López-Moreno, J.I., 2015. Spatial and temporal variability of winter snow and precipitation days in the western and central Spanish Pyrenees. Int. J. Climatol. 35, 259–274. https://doi.org/10.1002/joc.3978 Cohen, J., Fletcher, C., 2007. Improved skill of northern hemisphere winter surface temperature predictions based on land-atmosphere fall anomalies. J. Clim. 20, 4118–4132. https://doi.org/10.1175/JCLI4241.1 Dee, D.P., Uppala, S.M., Simmons, A.J., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U., Balmaseda, M.A., Balsamo, G., Bauer, P., Bechtold, P., Beljaars, A.C.M., van de Berg, L., Bidlot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer, A.J., Haimberger, L., Healy, S.B., Hersbach, H., Hólm, E. V., Isaksen, L., Kållberg, P., Köhler, M., Matricardi, M., Mcnally, A.P., Monge-Sanz, B.M., Morcrette, J.J., Park, B.K., Peubey, C., de Rosnay, P., Tavolato, C., Thépaut, J.N., Vitart, F., 2011. The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Q. J. R. Meteorol. Soc. 137, 553–597. https://doi.org/10.1002/qj.828 del Río, S., Fraile, R., Herrero, L., Penas, A., 2007. Analysis of recent trends in mean maximum and minimum temperatures in a region of the NW of Spain (Castilla y León). Theor. Appl. Climatol. 90, 1–12. https://doi.org/10.1007/s00704-006-0278-9 Dunstone, N., Smith, D., Scaife, A., Hermanson, L., Eade, R., Robinson, N., Andrews, M., Knight, J., 2016. Skilful predictions of the winter North Atlantic Oscillation one year ahead. Nat. Geosci. 9, 809–814. https://doi.org/10.1038/ngeo2824 Durán, L., Rodríguez-Fonseca, B., Yagüe, C., Sánchez, E., 2015. Water vapour flux patterns and precipitation at Sierra de Guadarrama mountain range (Spain). Int. J. Climatol. 35, 1593–1610. https://doi.org/10.1002/joc.4079 Essery, R., 2015. A factorial snowpack model (FSM 1.0). Geosci. Model Dev. 8, 3867–3876. https:// doi.org/10.5194/gmd-8-3867-2015 Esteban-Parra, M.J., Pozo-Vázquez, D., Rodrigo, F.S., Castro-Díez, Y., 2003. Temperature and Precipitation Variability and Trends in Northern Spain in the Context of the Iberian Peninsula Climate, in: Mediterranean Climate. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 259–276. https://doi.org/10.1007/978-3-642-55657-9_15 127 Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica Capítulo 5: Sensibilidad del manto de nieve a la variabilidad de la temperatura, precipitación y radiación solar en un gradiente altitudinal en la Península Ibérica Resumen: En el siguiente trabajo se ha investigado la sensibilidad del manto de nieve al aumento de la temperatura y a la radiación de onda corta, y al cambio de las precipitaciones a lo largo de un gradiente altitudinal (1500 - 2500 m s.n.m.) sobre las principales cadenas montañosas de la Península Ibérica (Cordillera Cantábrica, Sistema Central, Sistema Ibérico, Pirineos y Sierra Nevada). Hemos utilizado la salida de un modelo mesoatmosférico (WRF) como forzamiento en un modelo de base física de balance de masa y energía de la nieve (FSM2). Se aplicó un algoritmo de agrupamiento a los datos de entrada del modelo FSM2 para identificar un total de 12 celdas que resumen la variabilidad climática de las cadenas montañosas. Las salidas de WRF se proyectaron a diferentes bandas altitudinales utilizando una serie de fórmulas psicrométricas y radiativas así como gradientes de temperatura del aire. Con estos datos, hemos realizado un experimento factorial generando series meteorológicas sintéticas que implicaban una alteración gradual de la temperatura (aumentos de 0 - 4 ºC), radiación de onda corta (aumentos de 0-40 Wm-2) y precipitaciones (variaciones de ± 20%) en todas sus posibles combinaciones y usando estas series como forzamiento de FSM2. Los resultados muestran diferentes sensibilidades del manto de nieve en las diversas zonas montañosas como consecuencia de las diferentes particiones de los balances de masa y energía. Los resultados mostraron un impacto generalmente negativo del calentamiento climático en la magnitud, duración y tasas de fusión de la capa de nieve en todas las bandas de elevación, incluso bajo los escenarios de mayor precipitación. El efecto medio del calentamiento sobre la duración del manto de nieve osciló entre -23% por ºC a 1500 m s.n.m. y -13% por ºC a 2500 m s.n.m., sobre el máximo SWE acumulado osciló entre -20% por ºC a 1500 m s.n.m. y -15% por ºC a 2500 m s.n.m., y sobre las tasas de fusión osciló entre -9% y -6% por ºC. El efecto del aumento de la radiación de onda corta sobre el manto de nieve varió desde aproximadamente -2% por 10 Wm-2 a 1500 m a.s.l. hasta -1% por 10 Wm-2 a 2500 m a.s.l. tanto para la duración del manto de nieve como para los los máximos de SWE. El efecto sobre el manto de nieve causado por los cambios en 135 Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica la precipitación se redujo gradualmente con el aumento de la elevación, especialmente en las zonas más frías. La respuesta de las tasas de fusión al calentamiento fue negativa en la mayoría de las áreas en todas las elevaciones, lo que sugiere periodos de fusión menos intensos pero más largos. Cita completa: Alonso-González, E., López-Moreno, J. I., Navarro-Serrano, F., Sanmiguel-Vallelado, A., Aznárez- Balta, M., Revuelto, J., Ceballos, A.(2020). Snowpack sensitivity to temperature, precipitation, and solar radiation variability over an elevational gradient in the Iberian mountains. Atmospheric Research. Pp: 104973. https://doi.org/10.1016/j.atmosres.2020.104973 136 Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica Snowpack sensitivity to temperature, precipitation, and solar radiation variability over an elevational gradient in the Iberian mountains Esteban Alonso-González1, Juan Ignacio López-Moreno1, Francisco Navarro-Serrano1, Alba Sanmiguel-Vallelado1,,M. Aznárez-Balta1, Jesús Revuelto2, Fernando Domínguez-Castro1, Ceballos, A2. 1. Instituto Pirenaico de Ecología, Consejo Superior de Investigaciones Científicas (IPE-CSIC), Zaragoza, Spain 2. Dept. Geografía, Universidad de Salamanca, Salamanca, Spain Correspondence: Esteban Alonso-González ([email protected]c.com) Received: 16 January 2020 Accepted: 26 March 2019 Abstract. In this study we investigated the sensitivity of the snowpack to increased temperature and short-wave radiation, and precipitation change along an elevation gradient (1500 - 2500 m a.s.l.) over the main mountain ranges of the Iberian Peninsula (Cantabrian Range, Central Range, Iberian Range, Pyrenees, and the Sierra Nevada). The output of a meso-atmospheric model (WRF) was used as forcing data in a physically-based energy and mass balance snowpack model (FSM2). A cluster analyses was applied to the input data of the FSM2 model to identify a total of 12 cells that summarized the climatic variability of the mountain ranges. The WRF output was then rescaled to various elevation bands using an array of psychrometric and radiative formulae and air temperature lapse rates. A factorial experiment was performed to generate synthetic meteorological series involving gradual alteration of the temperature (0 – 4 ºC increases), short-wave radiation (0-40 Wm-2 increases), and precipitation (variations of ± 20%) to force the FSM2. We found differing sensitivities across the various mountainous areas as a consequence of differences in their energy and mass balances. The results showed a generally negative impact of climate warming on the magnitude, duration, and melt rates of the snowpack over all elevation bands, even under scenarios of greater precipitation. The average effect of warming on the duration of the snowpack ranged from -23% per ºC at 1500 m a.s.l. to -13% per ºC at 2500 m a.s.l., on the peak snow water equivalent ranged from -20% per ºC at 1500 m a.s.l. to -15% per ºC at 2500 m a.s.l., and on melt rates ranged from -9% to -6% per ºC. The effect of increasing short-wave radiation on the snowpack ranged from approximately -2% per 5 Wm-2 at 1500 m a.s.l. to -1% per 5 Wm-2 at 2500 m a.s.l. for both the snowpack duration and peak SWE indices. The effect on the snowpack caused by 137 Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica precipitation changes reduced gradually with increasing elevation, especially in the colder areas. The response of the melt rates to warming was negative in most of the areas at all elevations, suggesting less intense but longer melt seasons. 1 Introducción As a consequence of its physical properties (low thermal conductivity, high water storage capacity, and high albedo), the snowpack is a key element in many ecological, hydrological, and atmospheric processes in cold and mountainous areas. As in other Mediterranean areas (Fayad et al., 2017), in the Iberian Peninsula a large amount of the total annual precipitation falls during winter (López-Moreno et al., 2011, 2008), resulting in extensive snow-covered areas each year (Alonso-González et al., 2019a). Thus, the snowpack has a major influence on seasonal river flows (García-Ruiz et al. 2011), and enables water availability to be matched with the high demand during the warm and dry season (López-Moreno and García- Ruiz, 2004; Sanmiguel-Vallelado et al., 2017). Furthermore, winter tourism represents a key economic resource for mountainous areas of the Iberian Peninsula, where there are 64 ski resorts including in Spain (31), France (29), Andorra (3), and Portugal (1). However, various scenarios suggest that the snowpack and winter tourism are very vulnerable to climate change (Gilaberte- Búrdalo et al., 2017, 2014; Pons et al., 2015). The magnitude and duration of the snowpack responds rapidly to changes in temperature and other climate parameters (Barnett et al., 2005; Connolly et al., 2019), with deep implications over such processes. However, the response of the seasonal snowpack to climate variability is complex. Beyond the obvious implications for reduction in the magnitude and duration of the snowpack, recent studies have shown the potential for warming to decelerate melting rates (Musselman et al., 2017), and also the influence of humidity in controlling melting events (Harpold and Brooks, 2018). These indicate very contrasting sensitivities to warming processes, even among mountainous areas identified as Mediterranean (López-Moreno et al., 2017). Furthermore, potentially large withinrange variability in the response of the snowpack to climate change can occur because of elevation differences and local climatological variability, as has been reported for the Pyrenees (López- Moreno et al., 2009) or the Cascades (Sproles et al., 2013). Differing energy and mass balances in the snowpack, caused by different climatological characteristics, explain this contrasting response. Thus, physically-based models are commonly used to study and reproduce the heat and mass fluxes 138 Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica of the snowpack under varying conditions. Accurate meteorological forcing data are essential to correctly represent the snow processes in physically-based snowpack models (Côté et al., 2017; Raleigh et al., 2016; Slater et al., 2013). Unfortunately, there is a generalized lack of meteorological observational data for the mountainous areas of Iberia, and the available series are too short, sparse, or incomplete. Although data from meteorological observations are available for some mountain locations (Polo et al., 2019; Revuelto et al., 2017), meso-atmospheric simulations are the only available tool for obtaining long-term forcing data over the Iberian mountains. In addition to future warmer conditions, the Iberian Peninsula shows a significant positive trend in downward shortwave radiation, caused by the combined effect of a decrease in cloudiness and atmospheric aerosols (Vicente-Serrano et al., 2017). Radiative heat flux is an important factor in the energetic flux of the snowpack (Marsh et al., 2012), but how its variability affects the snowpack has been little investigated, even though it is known that this flux can control the spatiotemporal sensitivity of the snowpack, depending on slope and aspect (López-Moreno et al., 2013). Furthermore, there is large uncertainty in the projections for future precipitation over the Iberian Peninsula (Monjo et al., 2016), which has clear implications for predicting the evolution of the snowpack under changing climate conditions. Because of the dependence of the Iberian Peninsula on water stored as snow in mountain areas (López-Moreno and García-Ruiz, 2004), it is important to take into account the effect of elevation in investigating the sensitivity of the snowpack, as this will markedly affect the susceptibility of the snowpack to increasing temperature (López-Moreno et al., 2009; Marty et al., 2017; Sospedra- Alfonso et al., 2015). These uncertainties can be considered from a factorial point of view (Rasouli et al., 2015) by studying how variability in various factors influences the snowpack response, either independently or jointly.. In this study we investigated the sensitivity of the snowpack in various Iberian mountain ranges to changing temperature, incoming solar radiation, and precipitation. We used a pre-existing mesoatmospheric simulation as forcing data in a physically-based energy and mass balance snowpack model. Altering the forcing meteorological variables enabled us to simulate the snowpack under different climatological scenarios across an elevational gradient. 139 Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica 2 Study Area The Iberian Peninsula is located in the south-western part of Europe between latitudes 36 ºN and 43.5 ºN, and covers an area of 596,740 km2. It is characterized by high topographical complexity, and includes five main mountain ranges (the Cantabrian, Iberian, and Central ranges, and the Pyrenees and Sierra Nevada). These are each aligned roughly east-west, with maximum elevations of approximately 2500 m a.s.l. in the Cantabrian, Iberian, and Central ranges, and exceeding 3000 m a.s.l. in the Pyrenees and Sierra Nevada (Fig. 1). Most of the mainland areas of Iberia are on the Central Plateau, which is crossed by the Central Range, with the Cantabrian Range to the north, the Iberian Range to the east, and the Sierra Nevada 140 Figure 1: Main mountain ranges of the Iberian Peninsula; (A) Cantabrian range, (B) Central range, (C) Iberian range, (D) Pyrenees, (E) Sierra Nevada. Digital Elevation Model provided by the Spanish Geographical Survey. Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica to the south, in addition to many other secondary middle mountain ranges. The Pyrenees is the largest mountain range, in the north-eastern part of Iberian Peninsula, and is separated from the Central Plateau by the Ebro basin. As a consequence of this topographical complexity, the wide latitudinal range, and differing exposures to Atlantic and Mediterranean air masses, the Iberian Peninsula exhibits large climatological variability. This is reflected in the rich diversity of snowpack behaviors at different elevations in each mountain area, and within each mountain range (Alonso- González et al., 2019a). 3 Data and methods We used a pre-existing meso-atmospheric simulation for Iberia (García-Valdecasas Ojeda et al., 2017) as meteorological forcing data in a physically-based snowpack model, the Flexible Snow Model (FSM2) which is the second version of the Factorial Snow Model (Essery, 2015). The FSM model can be freely downloadable from https://github.com/RichardEssery/FSM2. The mesoatmospheric simulation was developed using the Weather Research and Forecast model (WRF) (Skamarock et al., 2008), driven by the ERA-Interim reanalysis dataset (Berrisford et al., 2011; Dee et al., 2011). The simulation spans the period 1980-2014 using a spatial resolution of 0.088º (~10 km at the latitude of the Iberian Peninsula). Previously, the meteorological surface data was rescaled to the common elevation band of 2000 m a.s.l. to make all the cells comparable independently of its elevation, using the methodology described by Alonso-González et al. (2018). Reescaling was performed using an array of psychrometric, barometric, and radiative formulae (Harder and Pomeroy, 2014; Liston and Elder, 2006). The 2000 m a.s.l. band was selected as it is a representative elevation band in all the involved mountain ranges (Alonso-González et al., 2019a). Thus, 2-m temperature, atmospheric pressure, longwave radiation, relative humidity and precipitation phase is corrected while wind speed, total precipitation, and shortwave radiation were not adjusted with the elevation difference. To reduce the computational cost of using all the cells of the simulation, we computed a spatial cluster analysis using the K-means algorithm (Hartigan and Wong, 1979) over the rescaled meteorological fields to enable selection of those pixels for each mountain range representing the most contrasting climatic characteristics. Firstly, we extracted those cells from the 3-hourly WRF simulation representing the main mountains of the Iberian Peninsula at 2000 m a.s.l., and obtained the temporally averaged values for the main variables involved in the snow energy and mass balance (2-m surface temperature, precipitation, short-wave radiation, and long-wave radiation) for 141 Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica winter (December, January, February) and spring (March, April, May). We then computed a principal components analysis (PCA) for the calculated variables, to reduce the number of dimensions. The cluster analysis was computed for the PCA components that overall explained > 85% of the variability of the meteorological data for each range independently. To determine the most appropriated cluster partitioning schema we computed an array of 30 indices for determining the number of clusters of each mountain range. We follow the majority criteria, where the most appropriate number of clusters is the most common number among the 30 indices (Charrad et al., 2014). The meteorological forcing data used in this study were extracted from the cell nearest to each cluster centroid, previously rescaled to the common elevation bands of 1500, 2000 and 2500 m a.s.l. using the same methodology explained before. The 2-m surface temperature, short-wave incoming radiation, and precipitation variables for each centroid were altered independently and in all possible combinations over the selected ranges to simulate differing climatological scenarios. Temperature was progressively increased by 0 to +4 ºC using 0.5 ºC incremental steps, while short-wave incoming radiation was increased over the range 0-40 Wm-2, using 5 Wm-2 steps. The selected ranges fall within the range of expected changes at the end of the 21st century, estimated from the observed trends for the period 1985-2010 (Sanchez- Lorenzo et al., 2013; Vicente-Serrano et al., 2017). Changes in precipitation were simulated incrementally by 5% steps over the range ± 20%, which encompassed the uncertainty projected by climatological models for the Mediterranean area (Knutti and Sedláček, 2013). These meteorological fields were altered using the framework proposed by Alonso-González et al. (2018), where the forcing meteorological fields are interrelated. For this study a constant relative humidity was maintained inside the framework, independently of the warming scenario (López-Moreno et al., 2017; Rasouli et al., 2014). All the new synthetic forcing series summarize a total of 1215 forcing datasets per centroid considering the three elevation bands. The topographical effects like slope or aspect are not considered in this study. We used the newly synthetically generated meteorological data as forcing in the FSM2 (Essery, 2015), for simulating daily snow water equivalent (SWE) series. We estimated the mean peak SWE and mean snow season duration by averaging the long-term values of annual maximum SWE values, and the annual number of days with SWE exceeding 10 mm. The mean melt rates were calculated by averaging the long-term annual ratio between peak SWE and the length of the melting season . We then calculated the sensitivity of each snow index, from the averaged change in each index influenced by the forcing variability. As a consequence of the generalized lack of 142 Sensibilidad del manto de nieve a la variabilidad climática en un gradiente altitudinal en la Península Ibérica observational snow data available to calibrate FSM2, we chose the model configuration with more physically based parametrizations, as the alternative configurations of the FSM were mostly a simplification of the snowpack processes like constant density or the estimation of the albedo as function of the surface temperature. This configuration has proved to be consistent with high mountain observations in previous studies in the Iberian Peninsula, and reproduced the inter- and intra-annual snowpack patterns (Alonso-González et al., 2018). Thus, albedo decreased as snow aged with time, and increased with snowfall. The compaction rate was calculated based on the overburden and thermal metamorphism. The turbulent exchange coefficient was corrected based on the bulk Richardson number. The thermal conductivity was calculated based on snow density. Finally, FSM2 configuration accounted for retention and refreezing of water inside the snowpack. We estimated the average peak SWE, snow season duration, and melt rates from the generated SWE series. The sensitivity of the snowpack duration, peak SWE, and snow melt rate was calculated for each forcing perturbation. The sensitivity was calculated as the average of the relative changes for each snow index to each forcing perturbation step (Fig. 2). 143 Figure 2: Example showing the relative sensitivities of snow duration and peak SWE for one of the centroids. The blue lines represent the variability caused uncertainty caused by a ± 20% perturbation in precipitation (at 10% intervals).