High-resolution spatially interpolated FAO Penman-Monteith crop reference evapotranspiration maps of Sicily Island (Italy) and Jucar River system (Spain) using AgERA5 and ERA5-Land reanalysis datasets

High-resolution spatially interpolated FAO Penman-Monteith crop reference evapotranspiration maps of Sicily Island (Italy) and Jucar River system (Spain) using AgERA5 and ERA5-Land reanalysis datasets Alberto Garcia-Prats a,* , Juan Manuel Carricondo-Ant´ on a , Matteo Ippolito b , Dario De Caro b , Miguel Angel Jim´ enez-Bello a , Juan Manzano-Ju´ arez c , Manuel Pulido-Velazquez a a Research Institute of Water and Environmental Engineering (IIAMA), Universitat Polit` ecnica de Val` encia, Camino de vera s/n, Valencia 46022, Spain b Department of Engineering, Universit` a degli Studi di Palermo, Viale delle Scienze 12, Ed. 8, Palermo 90128, Italy c Centro Valenciano de Estudios sobre el Riego (CVER), Universitat Polit` ecnica de Val` encia, Camino de vera s/n, Valencia 46022, Spain ARTICLE INFO Keywords: Penman-Monteith Crop reference evapotranspiration ERA5L and AgERA5 Spatial interpolation ABSTRACT Study region: Jucar River System (Spain) and Sicily Island (Italy). Study focus: Penman-Monteith crop reference evapotranspiration (PM-ETo) is critical for irrigation planning and hydrological modeling. Its estimation typically requires dense agricultural weather networks with automated stations. Alternatively, reanalysis datasets like ERA5-Land and AgERA5 offer spatially comprehensive data, but their resolution is often insufficient. Spatial interpolation techniques are thus required to estimate PM-ETo at unsampled locations. This study applied the DRI (Dynamic Regression-Based Interpolation) algorithm to generate high-resolution (100 m) PM-ETo maps for both regions using three data sources: meteorological station records and ERA5-Land and AgERA5 reanalysis products. The performance of AgERA5 for PM-ETo estimation was also assessed. Additionally, PM-ETo interpolated maps from the three sources were compared. New hydrological insights for the region: AgERA5, a bias-corrected downscaling of ERA5, effectively removed bias in Sicily when compared to in situ data, but not in the Jucar system. Nonetheless, AgERA5 outperformed ERA5-Land in both regions for PM-ETo estimation. Following interpolation, the resulting maps retained the same biases identified in the original datasets and preserved the frequency distributions of ground-truth maps. This indicates that the interpolation method does not distort the underlying meteorological fields between stations. The proposed approach offers a valuable tool for practitioners and modelers, enabling the generation of high-resolution, accurate, and practical PM-ETo maps to support irrigation planning and hydrological applications. * Correspondig author. E-mail address: [email protected] (A. Garcia-Prats). Contents lists available at ScienceDirect Journal of Hydrology: Regional Studies journal homepage: www.elsevier.com/locate/ejrh https://doi.org/10.1016/j.ejrh.2025.102531 Received 28 January 2025; Received in revised form 7 May 2025; Accepted 12 June 2025 Journal of Hydrology: Regional Studies 60 (2025) 102531 Available online 17 June 2025 2214-5818/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ). 1. Introduction Agriculture is undoubtedly the most water-demanding activity globally, accounting for over 85 % of human water consumption (D’Odorico et al., 2020). As a result, irrigated agriculture plays a central role in any sustainable water resource management policy. The first step in this context is to determine crop water requirements accurately. It should be recognized that a significant effort has been made by the Food and Agriculture Organization of the United Nations (FAO) in this field during the last three decades. Today, irrigation recommendations worldwide follow the FAO-56 methodology (Allen et al., 1998), which is based on the Penman-Monteith crop reference evapotranspiration (PM-ETo) equation, as an expression of the daily evaporative demand of the atmosphere (Raes et al., 2009). For this purpose, hundreds of agricultural weather networks have been established worldwide, specifically set up to provide irrigation advice (Elliott et al., November 14-16, 2000). These weather stations are equipped with sensors to calculate PM-ETo, which is used in irrigation recommendations (Brown, 2007). One of the earliest initiatives was the California Irrigation Management Information System (CIMIS) developed in 1982 by the Department of Water Resources of the State of California and the University of California, Davis (UC Davis), with 145 weather stations. This technique quickly spread across five continents, and today, almost every irrigated region has its own weather network, with data accessible and downloadable online. Examples include the Agro-climatic Information System for Irrigation (SIAR) in Spain, with 460 automated weather stations, and the Sicilian Agrometeorological Information System (SIAS) in Sicily (Italy), with 96 automated weather stations. Even when an irrigation advisory weather network is available in a region, the station density is often insufficient to generate specific recommendations for each farm. The PM-ETo data used may come from stations located many kilometers away, depending on the farm’s relative position to the nearest station (Martínez-Lüscher et al., 2022). This issue is especially significant in areas with varying topography, where geographical features create sharp gradients. In such cases, the only way to obtain site-specific PM-ETo values is by creating maps that reflect the spatial variation based on the landscape’s geographical features (Vicente-Serrano et al., 2007). This interpolation method is typically based on multiple linear regressions, where the meteorological variable being explained is related to geographical characteristics like latitude, longitude, elevation, continentality, slope, and aspect, which act as independent variables. Garcia-Prats et al. (2023) proposed a Daily Dynamic Regression-Based Interpolation method (DRI), showing that this method outperformed geostatistical approaches, as previously pointed out in Tomas-Burguera et al. (2018). At the end of each day, new regression coefficients are generated for each meteorological variable recorded, with different geographical features explaining the spatial variability depending on the day’s synoptic pattern. Garcia-Prats et al. (2023) quantified an average error of 10 % in PM-ETo (and irrigation requirements) when using the nearest weather station instead of a plot-specific PM-ETo to estimate the crop water needs in a Mediterranean basin in eastern Spain, with errors reaching up to 30 % in the worst cases. Accurate spatial representation of PM-ETo values is crucial not only for effective irrigation management but also for hydrological modeling at the basin scale. Given the importance of evapotranspiration in irrigation and hydrological processes, distributed -e.g., TETIS (Franc´ es et al., 2007; Gomis-Cebolla et al., 2023) and semi-distributed -e.g., SWAT (Soil & Water Assessment Tool) (Arnold et al., 1998)- widely used hydrological models depend on high-quality PM-ETo gridded maps as inputs to produce reliable outputs. Atmospheric reanalysis is a method that combines past observations with advanced dynamical-physical coupled numerical models to create a consistent time series of multiple weather variables. By assimilating observations from the atmosphere, land, and ocean into a forecast model, reanalysis produces a consistent time series estimate of past weather states at each time step (Vanella et al., 2022). These datasets, available on 3D grids at sub-daily intervals, provide a comprehensive and continuous weather-variable estimation over recent decades. Often referred to as ’maps without gaps’, the databases are served on regular grids, regardless of whether actual weather information was collected in a given region (Tarek et al., 2020). Among the various reanalysis products, the ERA5 dataset, produced by the Copernicus Climate Change Service (C3S) at the European Centre for Medium-Range Weather Forecasts (ECMWF), is gaining popularity. ERA5 provides hourly estimates of the variables needed to derive PM-ETo and is updated daily with a 5-day latency, covering the globe on a 31 km grid (Hersbach et al., 2020). The spatial resolution of ERA5 is insufficient for many applications, prompting the launch of two higher-resolution products: i) ERA5-Land and ii) AgERA5. ERA5-Land retains most of ERA5’s parameterizations, ensuring the use of state-of-the-art land surface modeling in numerical weather prediction (Mu˜ noz-Sabater et al., 2021). It offers the same temporal resolution and time lag as ERA5, but with an improved spatial resolution of 9 km, compared to ERA5’s 31 km. However, ERA5-Land does not address the systematic biases present in the original reanalysis product. Some studies in the literature have proposed bias correction methods for variables such as precipitation (Ascenso et al., 2024) (Cavalleri et al., 2024) (Gomis-Cebolla et al., 2023), air temperature (Niazkar et al., 2024) (Erlat and Güler, 2024) (Dhawan et al., 2024), global solar radiation (Pelosi and Chirico, 2021), or wind speed (Elshinnawy et al., 2024), to cite some examples. The second enhanced-resolution product, AgERA5 (Boogaard et al., 2020), has been bias-corrected by applying grid and variable-specific regression equations to an ERA5 dataset interpolated at 9 km. This product, with a 7-day time lag, is theoretically ready for direct use in models and other applications. To assess the utility of reanalysis products for filling data gaps in agricultural weather stations, providing data in regions without actual observations, or supporting agro-hydrological models with continuous spatiotemporal PM-ETo datasets, some efforts can be found in the literature comparing PM-ETo calculated from ERA5/ERA5-Land to ground-truth data from weather stations. Vanella et al. (Vanella et al., 2022) analyzed ERA5/ERA5-Land agrometeorological data from 2008 to 2020, comparing PM-ETo estimates with observational data from 66 sites across seven irrigation districts in Italy. They found a bias in PM-ETo, with ERA5/ERA5-Land underestimating values by 2–13 %, depending on the climatic region. Ippolito et al. (2024) quantified a 23 % underestimation across the entire island of Sicily using ERA5-Land and observations from 39 weather stations. Similarly, Gourgouletis et al. (Gourgouletis et al., 2023) reported a 35 % underestimation in Greece when comparing ERA5-Land to actual data from 3 weather-stations. Finally, Yu et al. (Yu et al., 2023) found a bias ranging from +7 % to −18 % when comparing ERA5 data to observations from 689 weather stations across seven climatic regions in China. It is worth emphasizing that all these studies compared A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 2 datasets recorded at weather stations during specific periods with corresponding datasets from the reanalysis product, taken from the pixel in which the weather station is located. In other matters, authors such as Negm et al. (Negm et al., 2017) suggested applying altitude corrections in their analyses; however, Vanella et al. (Vanella et al., 2022) concluded that climatic regions had a higher influence than topography. In any case, even when a good agreement is found between weather station data and reanalysis products in a Fig. 1. Case study regions’ location. Coordinates (latitude and longitude) in degrees. A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 3 region, the issue of using PM-ETo values for irrigation recommendations from several kilometers apart remains unresolved, because the spatial resolution of the reanalysis product, still at 9 km, may not accurately represent conditions at the plot level. Considering the current state of the art, this study aims at filling some research gaps found, contributing to a better comprehension Fig. 2. Automated weather station of SIAR and SIAS network’s locations, and ERA5-Land and AgERA5 grid over a digital elevation model in both regions. Coordinates (longitude and latitude) in degrees. A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 4 of PM-ETo mapping on a daily basis to be used in plot-specific irrigation recommendations and hydrological models. Our study was conducted on both sides of the Mediterranean Sea: on the one hand, in the Jucar River system, a Mediterranean region of eastern Spain, and on the other hand, the entire island of Sicily, in southern Italy; two regions where intensive irrigated agriculture and a pronounced relief in certain areas, provide an excellent context to address the following research questions (RQ): RQ1: Currently, no studies have evaluated the suitability of the AgERA5 bias-corrected reanalysis product for PM-ETo calculation compared to ground-truth observations from agricultural weather stations. Does AgERA5 resolve the bias issues identified for ERA5/ ERA5-Land in the Mediterranean region, as reported in previous studies? RQ2: Can the DRI interpolation method, as proposed by Garcia-Prats et al. (2023), be applied to ERA5-Land and AgERA5 datasets to generate high-resolution PM-ETo maps to be used in plot-specific irrigation recommendations, or as input in distributed and semidistributed hydrological models? RQ3: If the answer to RQ2 is yes, what is the pixel-by-pixel agreement between PM-ETo high-resolution maps derived from actual observational data from agricultural weather networks and those generated using reanalysis products? The answer to each question is a contribution of this paper to the advancement of knowledge in this field. Checking the effectiveness of the AgERA5 bias correction in order to know whether the product can be used directly or additional corrections will be necessary before using this database in hydrological models and irrigation management is the first one. The ability of the DRI method to downscale reanalysis products to be used in high-resolution PM-ETo maps is the second one. Finally, we checked the agreement between those high-resolution maps to the one produced using actual registered data. 2. Material and methods 2.1. Case study regions 2.1.1. The Jucar River system (JRs) Located in eastern Spain (Fig. 1), the Jucar River system (JRs) closely aligns with the boundaries of the Jucar River basin. According to the river authority (Confederaci´ on Hidrogr´ afica del Jucar (CHJ), 2022), the basin experiences a typical Mediterranean climate, with an average annual precipitation of around 500 mm, ranging from 300 to 780 mm. The rainfall pattern includes intense rains in autumn (with October being the wettest month at almost 60 mm) and spring, along with dry summers (July is the driest month, with about 13 mm). Average temperatures range from 11◦C in winter to 26.6◦C in summer. According to the latest version of the Koppen climatic classification, the region is characterized by a hot summer Mediterranean climate (Csa) (Kottek et al., 2006). The average PM-ETo for the period 1980–2018 was 992.7 mm. The JRs cover an area of 2226,093 ha, of which 210,000 ha are irrigated. Citrus trees dominate the irrigated area, accounting for nearly 40 %, while the remaining 60 % is devoted to cereals, vineyards, fruit trees, horticultural crops, and olives Confederaci´ on Hidrogr´ afica del Jucar (CHJ) (2022). 2.1.2. Sicily Island Sicily region is the largest Italian island (Fig. 1). The main mountain range extends along the longitudinal direction from the Fig. 3. Automated weather station of SIAS (left) and SIAR (right) networks. A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 5 northern to the eastern side. According to the latest version of the Koppen climatic classification, the region is characterized by a hot summer Mediterranean climate (Csa)(Kottek et al., 2006). The precipitation distribution across the island shows clear spatial-temporal variability; specifically, in the southeast area, the mean annual precipitation is about 360 mm, while a mean value of 1900 mm is reached in the north-east (Treppiedi et al., 2021). The average temperature during the winter ranges from 4◦C in the inland areas to 9.7◦in the coastal zones, whereas, in summer from 28◦C to 30◦C, respectively (Di Nunno and Granata, 2023). The mean annual value of PM-ETo is 900 mm. The area of the Sicily region is approximately 2570,000 ha, of which about 820,000 ha are irrigated. According to the regional agricultural soil, the main cultivated crops are cereals (38 %), while regarding the tree crops, olives are the most common (11 %) but also vineyards (9 %) and citrus (5,7 %) are cultivated (CREA, 2024). 2.2. Meteorological data 2.2.1. “SIAR” agricultural weather network The irrigated areas of the JRs are monitored by the Agro-climatic Information System for Irrigation (SIAR - www.siar.es), an automated weather network created in 1998 to promote water savings through accurate irrigation recommendations. SIAR covers the main irrigated regions of Spain, operating 460 automated weather stations, 63 of which are located within or near the JRs, as shown in Fig. 2. SIAR was initiated by the Spanish Ministry of Agriculture, Fisheries, and Food (MAPA) and is maintained—and sometimes expanded with additional stations—by regional governments. In the case of the JRs, weather stations are managed by the regional governments of the Valencian Community and Castile-La Mancha. Fig. 3 shows a typical SIAR automated weather station, equipped with a bucket rain gauge, anemometer, wind direction sensor at a standard 2-meter height, temperature and humidity probes, and a pyranometer. Each station also includes a data logger and GSM communication module for automatic data transmission several times a day. Meteorological data is recorded every 30 min and then aggregated into hourly and daily data, which is made available via an API. Preventive maintenance is performed every six months, with an annual calibration of the station’s sensors conducted in a laboratory. Corrective maintenance is carried out to address any detected anomalies. 2.2.2. “SIAS” agricultural weather network Sicily is covered by a large monitoring network of 96 automated weather stations, operated by the SIAS (Servizio Informativo Agrometeorologico Siciliano - www.sias.regione.sicilia.it/). In this study, 39 stations were selected to represent the different environmental conditions, as shown in Fig. 2; specific details regarding the weather stations’ position, elevation, and available records are reported in Ippolito et al.(2024). SIAS stations acquire the same agro-climatic variables as SIAR, at an hourly time step. SIAS weather stations are periodically maintained by technical experts. Moreover, Quality Assurance (QA) and Quality Control (QC) procedures are carried out before the data is published (De Caro et al., 2023). 2.2.2. Reanalysis data collection and characteristics In this work, two reanalysis products were employed: i) ERA5-Land (Mu˜ noz-Sabater et al., 2021) and ii) AgERA5 (Boogaard et al., 2020). Both are derived from the ECMWF’s 5th-generation global reanalysis product, ERA5. ERA5-Land retains the same temporal resolution and time lag as ERA5 but features an improved spatial resolution of 0.1◦(9 km). AgERA5, in contrast, has undergone bias correction by aggregating data to daily time steps in the local time zone and adjusting for finer topography using a 0.1◦(9 km) spatial resolution grid. Additionally, variable-specific regression equations were applied to the ERA5 dataset and trained on ECMWF’s operational high-resolution atmospheric model (HRES). Despite a slightly longer day-time lag, AgERA5 is theoretically ready for direct use in models and other applications. The main characteristics are summarized in Table 1. Both datasets were downloaded from the New Climate Data Store (CDS) operated by the Copernicus Climate Change Service. ERA5Land can be retrieved from: https://cds.climate.copernicus.eu/datasets/reanalysis-era5-land?tab=overview, while AgERA5 in: https://cds.climate.copernicus.eu/datasets/sis-agrometeorological-indicators?tab=overview. Table 2 summarizes the statistical characteristics of the four above-mentioned datasets: SIAR, SIAS, ERA5-Land, and AgERA5. 2.3. Modeling framework 2.3.1. PM-ETo estimation The Penman-Monteith equation, as outlined in the FAO-56 guidelines (Allen et al., 1998), was adopted to calculate the crop reference evapotranspiration (PM-ETo) for both datasets: the agricultural weather networks (SIAR and SIAS) and the reanalysis products (ERA5-Land and AgERA5). Table 1 Main characteristics of the reanalysis products ERA5-Land and AgERA5. Property/Model ERA5-Land AgERA5 Data type Gridded Gridded Projection Latitude-Longitude Latitude-Longitude Horizontal coverage Global Global Horizontal resolution 0.1◦x0.1◦(9 km) 0.1◦x0.1◦(9 km) Temporal coverage 1979 to present 1979 to present Temporal resolution Hourly Daily Time lag 5 day 7 day A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 6 ETo =0.408 ⋅Δ⋅(Rn−G) + γ⋅900 Tmean+273 ⋅U2⋅(es−ea) Δ+γ⋅(1+0.34 ⋅U2)(1) where ETo =Crop reference evapotranspiration, in mm⋅d −1 ; Δ =slope vapour pressure curve, in kPa ºC −1 ; R n =net radiation at the crop surface, in MJ m −2 d −1 ; G =soil heat flux density, in MJ m −2 d −1 ; U 2 =wind speed at 2 m height, in m s −1 ; γ =psychrometric constant, in kPa ºC −1 ; (e s – e a ) =saturation vapour pressure deficit, in kPa; and T mean =mean daily air temperature, in ºC. Reanalysis products do not provide air relative humidity. However, e s and e a calculation in the weather stations is based on these variables. Relative humidity for the reanalysis products was obtained after (Allen et al., 1998) as: RH =ea e0(T)(2) Being: ea=e0(Tdew) = 0.618 ⋅e(17.27⋅Tdew Tdew+237.3)(3) e0(T) = 0.618 ⋅e(17.27⋅Tmed Tmed+237.3)(4) Where e0(T) = saturation vapour pressure at T med or T dew respectively, in kPa. T dew =dew point temperature, in ºC; T med is obtained as (T max +T min )/2, in ºC, being T max and T min the maximum and minimum daily temperature respectively, in ºC. Finally, wind speed in the reanalysis products is modeled at 10 m above ground, whereas the standard for weather stations is to Table 2 Statistical characteristics of the reanalysis products ERA5-Land and AgERA5 compared to actual meteorological registered SIAR/SIAS data. T max =air maximum temperature; T min =air minimum temperature; RH =air relative humidity; U 2 =wind speed; Rs =global solar radiation. JRs-SIAR T max (ºC) T min (ºC) U 2 (m s −1 ) Rs (MJ m −2 d −1 ) RH (%) Average 23.69 10.81 1.27 15.86 68.32 Standard dev. 7.99 6.84 0.85 8.55 15.91 Maximum 45.05 28.4 9.51 32.38 100.00 Minimum 0.00 −12.65 0.00 0.00 0.00 Median 23.61 10.89 1.05 14.25 69.77 JRs-ERA5Land 23.69 10.81 1.27 15.86 68.32 Average 21.02 10.24 1.79 17.37 59.28 Standard dev. 10.91 7.43 0.96 9.83 27.24 Maximum 43.49 29.51 9.62 31.63 99.15 Minimum 0.09 −7.07 0.17 0.74 11.20 Median 20.09 9.97 1.62 15.88 60.72 Average 21.02 10.24 1.79 17.37 59.28 JRs-AgERA5 10.91 7.43 0.96 9.83 27.24 Average 20.90 10.71 2.27 17.23 59.26 Standard dev. 9.74 7.40 1.15 9.13 23.49 Maximum 43.62 28.21 12.61 31.56 100.00 Minimum −2.13 −7.37 0.26 0.41 11.77 Median 20.13 10.79 2.03 15.75 60.60 Average 20.90 10.71 2.27 17.23 59.26 Sicily-SIAS T max (ºC) T min (ºC) U 2 (m s −1 ) Rs (MJ m −2 d −1 ) RH (%) Average 22.50 11.90 1.90 17.50 65.10 Standard dev. 8.30 6.70 1.10 8.00 17.30 Maximum 43.30 31.40 11.30 34.87 100.00 Minimum −2.60 −8.30 0.37 0.49 0.00 Median 21.80 11.60 1.55 17.06 67.50 Sicily-ERA5Land 22.50 11.90 1.90 17.50 65.10 Average 20.34 12.40 1.86 17.29 72.65 Standard dev. 7.60 4.93 0.77 6.64 25.72 Maximum 42.24 27.80 11.24 30.14 98.25 Minimum −1.13 −5.68 0.21 1.63 25.85 Median 19.71 12.16 1.54 16.68 74.93 Average 20.34 12.40 1.86 17.29 72.65 Sicily-AgERA5 Average 20.83 15.28 2.68 17.86 72.35 Standard dev. 10.04 7.77 1.48 8.97 33.14 Maximum 41.77 29.98 11.99 30.46 100.00 Minimum −0.60 −8.13 0.19 0.17 23.65 Median 20.39 15.21 2.18 17.62 73.82 Average 20.83 15.28 2.68 17.86 72.35 A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 7 measure at 2 m. To align the datasets, the 10-meter wind speed from the reanalysis products was converted to 2-meter wind speed using the logarithmic wind profile equation proposed by 2.3.2. Daily dynamic regression-based interpolation procedure (DRI) Maximum and minimum temperature, relative humidity, and wind speed coming for both agricultural weather networks (SIARSIAS) and reanalysis products (ERA5-Land, AgERA5) were spatially interpolated using the DRI procedure after Garcia-Prats et al. (2023). The value of the variable in the space among weather stations (in the reanalysis products, the center of the pixel was treated as a fictitious weather station), was calculated using the equation (Ninyerola et al., 2000): Z(x) = a0+a1⋅x1+a2⋅x2+…+an⋅xn(5) where Z(x) =predicted value of the air temperature, air humidity, or wind speed, in ºC, % or m s −1 respectively; a 0 to a n =regression coefficients, and x 1 to x n =values of the independent variables (geographical features) at the spatial point x. The geographical features acting as independent variables in the multiple regression interpolation considered were longitude (LON), in degrees; latitude (LAT), in degrees; distance to the nearest sea with clear influence in the weather synoptic (MEDIT) -the Mediterranean Sea in our case-, in meters (m); annual average incoming solar radiation (RAD), in MJ m −2 d −1 ; annual average incoming solar radiation within the raadius x i (RADx), where x is a radius of 2.5, 5, and 25 km, in MJ m −2 d −1 ; elevation (ELEV), in meters (m); Mean elevation within x i (ELEVx), where x is a radius of 2.5, 5, and 10 km, in m; terrain slope (SLOPE), in %; and terrain aspect (ASPECT), in 0–360 degrees, with respect to north (0 degrees). The linear relationship between geographical features and predicted air temperature, relative air humidity, and wind speed was explored using a forward stepwise multiple regression procedure. Each day, a separate regression equation was generated, including only independent variables with a p-value < 0.01 to prevent collinearity issues. Homoscedasticity (checked via residual plots) and the absence of residual autocorrelation (using the Durbin-Watson statistic) were also verified. Once the relevant geographical variables and coefficients for the multiple regression equations were identified, a continuous map was created by combining raster layers containing each geographic variable’s data and applying the algebraic formula. Naturally, the registered and modeled air temperature, relative humidity, and wind speed values do not exactly match at weather station locations. Using kriging techniques, residuals were interpolated across the study area and added to the predicted values. In this case, the splines with tension method (Mit´ aˇ sov´ a and Mit´ aˇ s, 1993) was used. This approach ensures that the registered and predicted values at weather stations used to train the model align, significantly improving validation at other locations, as noted by (Ninyerola et al., 2000) and Agnew and Palutikof (Agnew and Palutikof, 2000). Using this described methodology, both datasets, actual observations from the SIAR/SIAS weather network, and ERA5-Land/ AgERA5 reanalysis products were interpolated. The spatial resolution for all interpolated maps was 100 m. 2.3.3. Radiation interpolation procedure Solar radiation plays a crucial role in the evapotranspiration (ET) process by providing the energy needed to transform water from liquid to vapor. The spatial heterogeneity of incoming solar energy, influenced by complex topography (elevation, slope, aspect, and shadows), creates strong local gradients that significantly impact ET dynamics, as noted by Tovar-Pescador et al. (Tovar-Pescador et al., 2006). This variability cannot be adequately captured by assuming flat terrain. A widely used approach to account for this is the upward-looking hemispherical viewshed algorithm, which is a virtual equivalent of hemispherical canopy photography (Kodysh et al., 2013). This method provides three key outputs: i) global, ii) direct, and iii) diffuse radiation. Regardless of the specific version of the viewshed algorithm used, two parameters control its operation: i) atmospheric transmissivity and ii) diffuse radiation rate. Transmissivity, which ranges from 0 (no transmission) to 1 (complete transmission), represents the fraction of solar radiation that passes through the atmosphere and reaches the Earth’s surface. Typical values range from 0.6 to 0.7 for very clear skies and around 0.5 for generally clear conditions. The diffuse radiation rate, or the proportion of global radiation that is scattered, ranges from about 0.2 for very clear skies to 0.7 for very cloudy conditions (ArcGIS Help, 2022; Kodysh et al., 2013). These parameters, however, are highly variable, changing with atmospheric conditions such as cloud cover, precipitation, dust, and aerosols, all of which influence the amount of diffuse and direct solar radiation reaching a given surface. To account for this temporal and spatial variability while balancing computational feasibility, the following approach was applied: For 2022, 365 potential radiation (PR) grid maps (one for each day of the year) were generated using the upward-looking hemispherical viewshed algorithm, assuming clear sky conditions with a transmissivity of 0.6 and a diffuse radiation rate of 0.2. For each day requiring a radiation map, the ratio of actual radiation (Rs) recorded at weather stations -or reanalysis productto the day-specific PR at the station’s location was calculated. These Rs/PR ratios were then spatially interpolated using the splines-with-tension method (Mit´ aˇ sov´ a and Mit´ aˇ s, 1993) to create a grid map of Rs/PR. Finally, a daily Rs grid map was produced by multiplying this ratio grid by the corresponding PR grid. In this study, Solar Analyst in the ArcGIS environment was used to generate the 365 PR maps and the splines-with-tension method in GRASS GIS’s v.surf.rst tool (Mit´ aˇ sov´ a and Mit´ aˇ s, 1993) was used to interpolate the Rs/PR ratios. Using this described methodology, both datasets, actual observations from the SIAR/SIAS weather network, and ERA5-Land/ AgERA5 reanalysis product were interpolated, and the PM-ETo maps were derived from them. The spatial resolution of the radiation and PM-ETo interpolated maps was 100 m. A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 8 2.4. Statistical analysis and model performance A triple task was developed depending on the origin of the dataset. On the one hand, the accuracy of the interpolated maps (DRI method after Garcia-Prats et al. (2023)) coming from observed weather data registered at the agricultural weather networks SIAR and SIAS was assessed by comparing modeled vs actual data. This comparison has to be done in different weather stations from the ones employed in the interpolation procedure. For this purpose, approximately 20 % of the stations (14 out of 63 in the JRs and 6 out of 39 in Sicily) were randomly selected to be employed only for the validation tests. As explained before, the rest of the stations (49 in SIAR and 33 in SIAS) were employed exclusively in the interpolation procedures. For each weather variable (air temperature, air relative humidity, wind speed, and global solar radiation), the statistics contained in Table 3 were calculated (Legates and McCabe, 1999). The same statistical metrics were employed to assess the PM-ETo maps. NSE ranges between 1 and −∞. NSE =1 means a perfect match between model outputs and observed values, and NSE <0 means that the average of the observations would be a better predictor than the model Besides Table 3 statistics, the Intraclass Correlation Coefficient (ICC) was calculated. ICC comes from the clinical field to establish the reliability of some measurements taken using different instruments or assessed by different raters (McGraw and Wong, 1996; Koo and Li, 2016). In our case, the measurement result is the PM-ETo in every pixel, and the rater is the origin of the data, observations from the SIAR/SIAS weather network, or reanalysis product ERA5-Land/AgERA5. From a mathematical point of view, reliability is defined as the ratio of true variance to the sum of true variance plus error variance. However, in practice, this task can be performed in 10 different ways (McGraw and Wong, 1996; Koo and Li, 2016). The “Two-way mixed effects, absolute agreement, single rater/measurement” was adopted in this work. After organizing the PM-ETo values in a table with two columns, one per origin of data (observation/reanalysis) and one raw per station or pixel, ICC was obtained using the following equation: ICC =MSR−MSE MSR+ (k−1)⋅MSE+k n⋅(MSC−MSE)(6) Where: MS R =mean square for rows; MS E =mean square for error; MS C =mean square for columns; n =number of subjects (pixels); k=number of raters (data origins). According to Koo and Li (Koo and Li, 2016), an ICC index above 0.9 means an excellent agreement, between 0.75 and 0.9 is good, between 0.5 and 0.75 is moderate, and for ICC<0.5, the agreement among maps can be considered poor. Finally, Kling-Gupta Efficiency (KGE) (Gupta et al., 2009) was calculated using the original equation: KGE =1− (r−1)2+ ( α −1)2+ (β−1)2 √(7) Where r =Pearson correlation coefficient between the observed and predicted values, dimensionless; α =variability of the prediction error, calculated as the ratio of the standard deviations of the predicted and observed values, dimensionless; β =bias term obtained as the ratio of the means of the predicted and observed values, dimensionless. The authors derived KGE by breaking down NSE into its correlation, bias, and variability components. As a result, KGE can take on the same values as NSE, with the same interpretation as explained above (Xie et al., 2024). Once those maps had been validated, the second task was assessing the accuracy of the AgERA5 dataset compared to actual data from observations of the weather stations. As was mentioned in the introduction section, this task has never been developed in these regions. To do this, the daily value of each meteorological variable (and the PM-ETo derived) in the pixel in which the weather station is located in the reanalysis grid is compared to the actual data registered at the weather station. For this purpose, all the weather stations were employed. Using the ICC, KGE, and the other statistic metrics of Table 2, the accuracy of this dataset was evaluated. Table 3 Statistical metrics. MAE Mean absolute error MAE =1 n∑n j=1Pj−Oj MBE Mean bias error MBE =1 n∑n j=1(Pj−Oj) RMSE Root mean squared error RMSE = 1 n∑n j=1(Pj−Oj)2 √ PBE Percentage bias error PBE =100 ⋅∑n j=1(Pj−Oj) ∑n j=1Oj d Index of Agreement (Willmott, 1981) d=1−∑n j=1(Pj−Oj)2 ∑n j=1(Pj−Oj+Oj−O)2 NSE Nash-Sutcliffe model efficiency(Nash and Sutcliffe, 1970)NSE =1−∑n j=1(Oj−Pj)2 ∑n j=1(Oj−O)2 Where: P j Predicted values O j Observed values OAverage of observed values d has limits of 0, indicating no agreement, and 1, indicating perfect agreement A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 9 Obtaining PM-ETo data requires an agricultural weather network with automated stations that continuously record meteorological variables. Spatial interpolation techniques are then used to estimate these variables at locations without stations. The DRI method is one such spatial interpolation technique. Another approach involves using reanalysis products generated by physically based climate forecast models. These products simulate past atmospheric conditions to provide regular grids of meteorological variables. ERA5-Land and AgERA5 are examples of such products, offering grid resolutions of 9 km. Both are derived from the downscaling of ERA5, a reanalysis product with a coarser 30 km resolution. Numerous studies have evaluated ERA5 and ERA5-Land against actual data from meteorological stations, consistently demonstrating their reliability and identifying specific biases. AgERA5 employs additional algorithms during its production process to mitigate these biases. This study applied the DRI spatial interpolation algorithm to generate high-resolution (100 m) PM-ETo maps for two study areas. The data sources included actual meteorological records from two agro-meteorological networks, as well as data from the ERA5-Land and AgERA5 reanalysis products. The suitability of the bias-corrected AgERA5 reanalysis product for PM-ETo estimation was evaluated by comparing the climate model data with ground-truth observations at the weather station locations. Moreover, the PM-ETo interpolated maps derived from the three data sources were compared. Based on the analysis, the following conclusions were drawn: •Based on the performance and agreement statistics, AgERA5 performed excellently in the JRs and Sicily regions. AgERA5, a biascorrected downscaling of ERA5, effectively removed bias in Sicily when compared to in-situ data, but not in the Jucar system and should be treated with specific techniques before being used in irrigation and hydrological modeling. •AgERA5 showed superior performance in both the JRs and Sicily regions compared to ERA5-Land. •Wind speed remains the most challenging meteorological variable to model accurately, with reanalysis products yet to resolve this issue effectively. Fig. 9. PM-ETo maps from SIAR weather network stations, reanalysis products ERA5Land and AgERA5, and ΔETo among actual and reanalysis products in the JRs region. A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 16 •High-resolution spatially interpolated maps derived from reanalysis products in the JRs region demonstrated better performance compared to Sicily Island. Furthermore, the interpolated maps preserved the characteristics of the ground truth maps’ frequency distribution functions, including the shape of the density function, histogram, box-and-whisker plot, and violin plot. •The interpolated maps based on reanalysis datasets retain the same biases previously identified when comparisons were made between actual weather station data and reanalysis data. This indicates that the interpolation method does not alter the meteorological fields between weather stations which means that methods to remove this bias in the original product could be valid to remove it in the high-resolution PM-ETo map produced. Therefore, if the input reanalysis dataset is unbiased, the resulting interpolated map will also remain unbiased. •The DRI interpolation method is suitable for generating high-resolution PM-ETo maps. These PM-ETo spatially interpolated values can be employed for plot-specific irrigation recommendations or as inputs for distributed and semi-distributed hydrological models. Furthermore, the interpolated maps exhibit a quality equivalent to those generated using actual data from weather station networks. This is especially useful for areas with low coverage of weather stations and periods with missing data. •Absolute differences (ΔETo) calculated pixel by pixel between high-resolution spatially interpolated maps using actual data and reanalysis datasets may be useful for identifying areas where the densification of the weather station network with additional stations would be beneficial. CRediT authorship contribution statement Carricondo-Ant´ on Juan: Writing – review & editing, Visualization, Software, Methodology, Investigation, Formal analysis, Data curation. Matteo Ippolito: Writing – review & editing, Validation, Methodology, Data curation. Dario De Caro: Writing – review & editing, Validation, Investigation, Formal analysis, Data curation. Alberto Garcia-Prats: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Methodology, Investigation, Formal analysis, Data curation, Fig. 10. PM-ETo maps from SIAS weather network stations, reanalysis products ERA5Land and AgERA5, and ΔETo among actual and reanalysis products in Sicily region. A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 17 Conceptualization. Juan Manzano-Ju´ arez: Writing – review & editing, Investigation, Formal analysis. Jim´ enez-Bello Miguel: Writing – review & editing, Visualization, Investigation, Data curation. Pulido-Velazquez, Manuel: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This study has received funding from Enhanced Integrated Multiscale Forecasting System For Agriculture, Water And The Environment (CIPROM/2023/5), which is funded by the Conselleria de Innovaci´ on, Universidades, Ciencia y Sociedad Digital de la Comunitat Valenciana, and from the European Union’s Horizon Europe research and innovation programme under the SOS-WATER project (GA no. 101059264). Meteorological data were provided by SIAR: “Sistema de Informaci´ on Agroclim´ atica para el Regadío. Ministerio de Agricultura, Pesca y Alimentaci´ on” and SIAS: “Servizio Informativo Agrometeorologico Siciliano. Dipartimento dell’Agricoltura”. Data availability Data will be made available on request. References Adler, D., Kelly, S.T., Elliott, T., & Adamson, J. (2022). vioplot violin plot. httpsgithub.comTomKellyGeneticsvioplot. Agnew, M.D., Palutikof, J.P., 2000. GIS-based construction of baseline climatologies for the Mediterranean using terrain variables. Clim. Res. 14 (2), 115–127. https://doi.org/10.3354/CR014115. Allen, R.G., Pereira, L.S., Raes, D., & Smith, M. (1998). Crop evapotranspiration: guidelines for computing crop water requirements. ArcGIS Help. https://pro.arcgis.com/en/pro-app/latest/help/main/welcome-to-the-arcgis-pro-app-help.htm. Arnold, J.G., Srinivasan, R., Muttiah, R.S., Williams, J.R., 1998. Large area hydrologic modeling and assessment part I: model development1. JAWRA J. Am. Water Resour. Assoc. 34 (1), 73–89. https://doi.org/10.1111/J.1752-1688.1998.TB05961.X. Ascenso, G., Ficchì, A., Giuliani, M., Scoccimarro, E., Castelletti, A., 2024. Downscaling, bias correction, and spatial adjustment of extreme tropical cyclone rainfall in ERA5 using deep learning. Weather Clim. Extrem. 46. https://doi.org/10.1016/J.WACE.2024.100724. Boogaard, H., Schubert, J., De Wit, A., Lazebnik, J., Hutjes, R., & Van der Grijn, G. (2020). Agrometeorological indicators from 1979 to present derived from reanalysis. Copernicus Climate Change Service (C3S) Climate Data Store (CDS). 〈https://cds.climate.copernicus.eu/datasets/sis-agrometeorological-indicators? tab=overview〉. Brown, P.W., 2007. Use of the ASCE standardized reference ET equation by agricultural weather networks in the Western U.S.: current status and future challenges. Examining Conflu. Environ. Water Concerns Proc. World Environ. Water Resour. Congr. 2006 1–10. https://doi.org/10.1061/40856(200)256. Cavalleri, F., Lussana, C., Viterbo, F., Brunetti, M., Bonanno, R., Manara, V., Lacavalla, M., Sperati, S., Raffa, M., Capecchi, V., Cesari, D., Giordani, A., Cerenzia, I.M. L., Maugeri, M., 2024. Multi-scale assessment of high-resolution reanalysis precipitation fields over Italy. Atmos. Res. 312, 107734. https://doi.org/10.1016/J. ATMOSRES.2024.107734. Confederaci´ on Hidrogr´ afica del Jucar (CHJ). (2022). Plan hidrol´ ogico de la demarcaci´ on hidrogr´ afica del Ju  car: ciclo de planificaci´ on hidrol´ ogica 2022 – 2027. Correndo, A.A., Rosso, L.H.M., Hernandez, C.H., Bastos, L.M., Nieto, L., Holzworth, D., Ciampitti, I.A., 2022. metrica: an R package to evaluate prediction performance of regression and classification point-forecast models. J. Open Source Softw. 7 (79), 4655. https://doi.org/10.21105/JOSS.04655). CREA - Consiglio per la ricerca in agricoltura e l’analisi dell’economia agraria - CREA. (n.d.). Retrieved October 15, 2024, from 〈https://www.crea.gov.it/〉. D’Odorico, P., Chiarelli, D.D., Rosa, L., Bini, A., Zilberman, D., Rulli, M.C., 2020. The global value of water in agriculture. Proc. Natl. Acad. Sci. USA 117 (36), 21985–21993. https://doi.org/10.1073/PNAS.2005835117/SUPPL_FILE/PNAS.2005835117.SAPP.PDF. De Caro, D., Ippolito, M., Cannarozzo, M., Provenzano, G., Ciraolo, G., 2023. Assessing the performance of the Gaussian Process Regression algorithm to fill gaps in the time-series of daily actual evapotranspiration of different crops in temperate and continental zones using ground and remotely sensed data. Agric. Water Manag. 290, 108596. https://doi.org/10.1016/J.AGWAT.2023.108596. Dhawan, P., Dalla Torre, D., Niazkar, M., Kaffas, K., Larcher, M., Righetti, M., Menapace, A., 2024. A comprehensive comparison of bias correction methods in climate model simulations: application on ERA5-Land across different temporal resolutions. Heliyon 10 (23), e40352. https://doi.org/10.1016/J.HELIYON.2024.E40352. Di Nunno, F., Granata, F., 2023. Future trends of reference evapotranspiration in Sicily based on CORDEX data and Machine Learning algorithms. Agric. Water Manag. 280, 108232. https://doi.org/10.1016/J.AGWAT.2023.108232. Elliott, R.L., Hubbard, K.G., Brusberg, M.D., Hattendorf, M.J., Howell, T.A., Marek, T.H., & Snyder, R.L. (2000). The role of automated weather networks in providing evapotranspiration estimates. National Irrigation Symposium. Proceedings of the 4th Decennial Symposium, Phoenix, Arizona, USA, November 14-16, 2000., 243–250. Elshinnawy, A.I., Men´ endez, M., Medina, R., 2024. A parameterization for the correction of ERA5 severe winds for extreme ocean wave modelling. Ocean Eng. 312. https://doi.org/10.1016/J.OCEANENG.2024.119048. Erlat, E., Güler, H., 2024. Assessment of changes in absolute extreme temperatures in the Mediterranean region using ERA5-Land reanalysis data. Theor. Appl. Climatol. 155 (9), 9051–9066. https://doi.org/10.1007/S00704-024-05162-8/TABLES/9. Franc´ es, F., V´ elez, J.I., V´ elez, J.J., 2007. Split-parameter structure for the automatic calibration of distributed hydrological models. J. Hydrol. 332 (1–2), 226–240. https://doi.org/10.1016/J.JHYDROL.2006.06.032. Garcia-Prats, A., Carricondo-Anton, J.M., Jim´ enez-Bello, M.A., Manzano Ju´ arez, J., L´ opez-P´ erez, E., Pulido-Velazquez, M., 2023. Dynamic procedure for daily PM56 ETo mapping conducive to site-specific irrigation recommendations in areas covered by agricultural weather networks. Agric. Water Manag. 287, 108415. https://doi.org/10.1016/J.AGWAT.2023.108415. Gomis-Cebolla, J., Rattayova, V., Salazar-Gal´ an, S., Franc´ es, F., 2023. Evaluation of ERA5 and ERA5-Land reanalysis precipitation datasets over Spain (1951–2020). Atmos. Res. 284, 106606. https://doi.org/10.1016/J.ATMOSRES.2023.106606. Gourgouletis, N., Gkavrou, M., Baltas, E., 2023. Comparison of empirical ETo relationships with ERA5-land and in situ data in Greece. Geographies 3 (3), 499–521. https://doi.org/10.3390/geographies3030026. Gupta, H.V., Kling, H., Yilmaz, K.K., Martinez, G.F., 2009. Decomposition of the mean squared error and NSE performance criteria: implications for improving hydrological modelling. J. Hydrol. 377 (1–2), 80–91. https://doi.org/10.1016/J.JHYDROL.2009.08.003. A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 18 Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Hor´ anyi, A., Mu˜ noz-Sabater, J., Nicolas, J., Peubey, C., Radu, R., Schepers, D., Simmons, A., Soci, C., Abdalla, S., Abellan, X., Balsamo, G., Bechtold, P., Biavati, G., Bidlot, J., Bonavita, M., Th´ epaut, J.N., 2020. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc. 146 (730), 1999–2049. https://doi.org/10.1002/QJ.3803. Ippolito, M., De Caro, D., Cannarozzo, M., Provenzano, G., Ciraolo, G., 2024. Evaluation of daily crop reference evapotranspiration and sensitivity analysis of FAO Penman-Monteith equation using ERA5-Land reanalysis database in Sicily, Italy. Agric. Water Manag. 295, 108732. https://doi.org/10.1016/J. AGWAT.2024.108732. James, G., Witten, D., Hastie, T., Tibshirani, R., & Taylor, J. (2023). An Introduction to Statistical Learning. https://doi.org/10.1007/978-3-031-38747-0. Kodysh, J.B., Omitaomu, O.A., Bhaduri, B.L., Neish, B.S., 2013. Methodology for estimating solar potential on multiple building rooftops for photovoltaic systems. Sustain. Cities Soc. 8, 31–41. https://doi.org/10.1016/J.SCS.2013.01.002. Koo, T.K., Li, M.Y., 2016. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J. Chiropr. Med. 15 (2), 155. https://doi.org/ 10.1016/J.JCM.2016.02.012. Kottek, M., Grieser, J., Beck, C., Rudolf, B., Rubel, F., 2006. World Map of the K¨ oppen-Geiger climate classification updated. Meteorol. Z. 15 (3), 259–263. https://doi. org/10.1127/0941-2948/2006/0130. Kuhn, Max, 2008. Building predictive models in R using the caret package. J. Stat. Softw. 28 (5), 1–26. https://doi.org/10.18637/jss.v028.i05. Legates, D.R., McCabe, G.J., 1999. Evaluating the use of “goodness-of-fit” Measures in hydrologic and hydroclimatic model validation. Water Resour. Res. 35 (1), 233–241. https://doi.org/10.1029/1998WR900018. Lemon, J., & others. (2006). Plotrix: a package in the red light district of R. R-News, 6(4), 8–12.. Martínez-Lüscher, J., Teitelbaum, T., Mele, A., Ma, O., Frewin, A.J., Hazell, J., 2022. High-resolution weather network reveals a high spatial variability in air temperature in the Central valley of California with implications for crop and pest management. PLOS ONE 17 (5), e0267607. https://doi.org/10.1371/ JOURNAL.PONE.0267607. McGraw, K.O., Wong, S.P., 1996. Forming inferences about some intraclass correlation coefficients. Psychol. Methods 1 (1), 30–46. https://doi.org/10.1037/1082989X.1.1.30. Mit´ aˇ sov´ a, H., Mit´ aˇ s, L., 1993. Interpolation by regularized spline with tension: I. Theory and implementation. Math. Geol. 25 (6), 641–655. https://doi.org/10.1007/ BF00893171/METRICS. Mu˜ noz-Sabater, J., Dutra, E., Agustí-Panareda, A., Albergel, C., Arduini, G., Balsamo, G., Boussetta, S., Choulga, M., Harrigan, S., Hersbach, H., Martens, B., Miralles, D.G., Piles, M., Rodríguez-Fern´ andez, N.J., Zsoter, E., Buontempo, C., Th´ epaut, J.N., 2021. ERA5-Land: A state-of-the-art global reanalysis dataset for land applications. Earth Syst. Sci. Data 13 (9), 4349–4383. https://doi.org/10.5194/ESSD-13-4349-2021. Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part I - A discussion of principles. J. Hydrol. 10 (3), 282–290. https://doi.org/ 10.1016/0022-1694(70)90255-6. Negm, A., Jabro, J., Provenzano, G., 2017. Assessing the suitability of American National Aeronautics and Space Administration (NASA) agro-climatology archive to predict daily meteorological variables and reference evapotranspiration in Sicily, Italy. Agric. For. Meteorol. 244–245, 111–121. https://doi.org/10.1016/J. AGRFORMET.2017.05.022. Niazkar, M., Piraei, R., Menapace, A., Dhawan, P., Torre, D.D., Larcher, M., Righetti, M., 2024. Bias correction of ERA5-Land temperature data using standalone and ensemble machine learning models: a case of northern Italy. J. Water Clim. Change 15 (1), 271–283. https://doi.org/10.2166/WCC.2023.669. Ninyerola, M., Pons, X., Roure, J.M., 2000. A methodological approach of climatological modelling of air temperature and precipitation through GIS techniques. Int. J. Clim. 20, 1823–1841. https://doi.org/10.1002/1097-0088. Pelosi, A., Chirico, G.B., 2021. Regional assessment of daily reference evapotranspiration: Can ground observations be replaced by blending ERA5-Land meteorological reanalysis and CM-SAF satellite-based radiation data? Agric. Water Manag. 258. https://doi.org/10.1016/J.AGWAT.2021.107169. Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2009. Aquacrop-The FAO crop model to simulate yield response to water: II. main algorithms and software description. Agron. J. 101 (3), 438–447. https://doi.org/10.2134/AGRONJ2008.0140S. Tarek, M., Brissette, F.P., Arsenault, R., 2020. Evaluation of the ERA5 reanalysis as a potential reference dataset for hydrological modelling over North America. Hydrol. Earth Syst. Sci. 24 (5), 2527–2544. https://doi.org/10.5194/HESS-24-2527-2020. Taylor, K.E., 2001. Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res. Atmospheres 106 (D7), 7183–7192. https://doi.org/ 10.1029/2000JD900719. Tomas-Burguera, M., Beguería, S., Vicente-Serrano, S., Maneta, M., 2018. Optimal Interpolation scheme to generate reference crop evapotranspiration. J. Hydrol. 560, 202–219. https://doi.org/10.1016/J.JHYDROL.2018.03.025. Tovar-Pescador, J., Pozo-V´ azquez, D., Ruiz-Arias, J.A., Batlles, J., L´ opez, G., Bosch, J.L., 2006. On the use of the digital elevation model to estimate the solar radiation in areas of complex topography. Meteorol. Appl. 13 (3), 279–287. https://doi.org/10.1017/S1350482706002258. Treppiedi, D., Cipolla, G., Francipane, A., Noto, L.V., 2021. Detecting precipitation trend using a multiscale approach based on quantile regression over a Mediterranean area. Int. J. Climatol. 41 (13), 5938–5955. https://doi.org/10.1002/JOC.7161. Vanella, D., Longo-Minnolo, G., Belfiore, O.R., Ramírez-Cuesta, J.M., Pappalardo, S., Consoli, S., D’Urso, G., Chirico, G.B., Coppola, A., Comegna, A., Toscano, A., Quarta, R., Provenzano, G., Ippolito, M., Castagna, A., Gandolfi, C., 2022. Comparing the use of ERA5 reanalysis dataset and ground-based agrometeorological data under different climates and topography in Italy. J. Hydrol. Reg. Stud. 42, 101182. https://doi.org/10.1016/J.EJRH.2022.101182. Vicente-Serrano, S.M., Lanjeri, S., L´ opez-Moreno, J.I., 2007. Comparison of different procedures to map reference evapotranspiration using geographical information systems and regression-based techniques. Int. J. Climatol. 27 (8), 1103–1118. https://doi.org/10.1002/JOC.1460. William Revelle. (2024). psych Procedures for Psychological, Psychometric, and Personality Research. httpsCRAN.R-project.orgpackage=psych. Willmott, C.J., 1981. On the validation of models. Phys. Geogr. 2 (2), 184–194. https://doi.org/10.1080/02723646.1981.10642213. Xie, Z., Yao, Y., Tang, Q., Liu, M., Fisher, J.B., Chen, J., Zhang, X., Jia, K., Li, Y., Shang, K., Jiang, B., Yang, J., Yu, R., Zhang, X., Guo, X., Liu, L., Ning, J., Fan, J., Zhang, L., 2024. Evaluation of seven satellite-based and two reanalysis global terrestrial evapotranspiration products. J. Hydrol. 630, 130649. https://doi.org/ 10.1016/J.JHYDROL.2024.130649. Yu, X., Qian, L., Wang, W., Huo, X., Hu, X., Wang, Y., 2023. Assessing and comparing reference evapotranspiration across different climatic regions of China using reanalysis products. Vol. 15, 2027, Water 2023 15 (11), 2027. https://doi.org/10.3390/W15112027. A. Garcia-Prats et al. Journal of Hydrology: Regional Studies 60 (2025) 102531 19