Evolution of annual rainfall irregularity in the southwest of the Iberian Peninsula

Evolution of annual rainfall irregularity in the Southwest of the 1 Iberian Peninsula 2 L. García-Barrón1 1 , M. Aguilar2 and A Sousa3 3 1 Department of Applied Physics II. University of Seville, Seville, Spain. [email protected] 4 2 Departament of Physical Geography. University of Seville. Seville. Spain. [email protected] 5 3 Department of Plant Biology and Ecology. University of Seville, Seville, Spain. [email protected] 6 Abstract: 7 The areas of the Iberian Peninsula with Mediterranean climate are characterized by 8 rainfall irregularity. Standard statistical estimation methods provide a limited insight of 9 all the dimensions of such irregularity. Based on different techniques to describe the 10 inter-annual irregularity of rainfall, the authors develop a new method: the disparity 11 indices. These indices are then applied to several historical rainfall series (dating from 12 the end of the 19th century up to the present) from the Southwest of the Iberian 13 Peninsula. Similar rainfall irregularity pattern are found in all weather stations in the 14 studied area confirming their belonging to the same climatic region. The results indicate 15 a relative stability during the first third of the XXth century coinciding with a period of 16 low precipitation and a progressive increase during the last three decades. The use of a 17 new index named Specific Disparity Index has proved be useful in highlighting the 18 irregularity within the rainfall series at each meteorological station. This new index 19 could contribute to monitor future changes in precipitation within the general 20 framework of research on climate change. Although Mediterranean ecosystems are 21 1 Correspondence to: Leoncio García-Barrón. Department of Applied Physics II, University of Seville, Avda. Reina Mercedes s/n, 41012 Seville, Spain. [email protected]. Phone: 0034 954556671; Fax: 0034 954556672 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 2 adapted to important fluctuations in the rainfall regime, this increase in irregularity may 1 affect rivers, wetlands and the hygrophytic vegetation. 2 KEY WORDS Precipitation; rainfall irregularity; variability; disparity index; 3 mediterranean ecosystems; Iberian Peninsula. 4 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 3 1. INTRODUCTION 1 One of the main characteristics of the rainfall regime in extensive areas of the Iberian 2 Peninsula is the great inter-annual irregularity: years with rainfall far below the mean 3 value contrast with others, sometimes following in consecutive years, with high 4 precipitation values. Additionally, intra-annual variability also exists (García-Barrón 5 2007): frequently the rainfall of just a few days represents a significant percentage of 6 the total monthly and annual rainfall values (Peñarrocha et al. 2002). The study of this 7 rainfall variability requires a new approach for very detailed description of variability 8 through a considerable number of alternative spread measurements. 9 Irregularity refers to any changes in the stationarity of climatic phenomena. It has 10 therefore, a broader meaning than variability: whereas variability only measures 11 changes or lack of stability of the variance of a series, irregularity describes any change 12 in the habitual patterns of the meteorological series. Thus, variability is the opposite of 13 homoscedasticity. When analysing meteorological series, irregularity refers to 14 significant changes in the internal organization or evolution pattern (trend, variability or 15 frequency distribution) when considering subsets in a series. A wide range between 16 extreme values and the unusual frequency or persistence of anomalous values in any of 17 the subset series are also indications of irregularity. 18 To a great extent, the analysis of rainfall irregularity has been linked with a search for 19 signs of climate change. The extensive literature addressing the issue of climatic 20 irregularity in the South of Iberian Peninsula has focused on the analysis of trends, on 21 the annual, seasonal or monthly variability patterns (Pita et al. 1997; Santos et al. 2005; 22 Norrant and Douguédroit 2006; Aguilar 2007) or on the occurrence of extreme events 23 (Manrique and Fernández-Cancio 2000; Rodrigo 2002; Haylock and Goodess 2004; 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 4 García et al. 2007; Costa et al. 2008). Other authors have analysed variability related 1 with atmospheric circulation modes (Hurrell 1995; Jones et al. 1997; Rodríguez-Puebla 2 et al. 1998; Rodrigo et al. 2000; Rodrigo and Trigo 2007; Vicente-Serrano and Cuadrat 3 2007) interpreting the atmospheric mechanisms causing such irregularity (Win-Nielse 4 2002; De la Torre 2003; Dünkeloh and Jacobeit 2003; Trigo et al. 2004). 5 Numerous studies have addressed climatic variability (Peterson et al. 2001; González 6 Hidalgo et al. 2003; Vicente-Serrano and Cuadrat-Prats 2007) but very few have 7 pursued the analysis of disparity (Martínez et al. 2007), that we define as the degree of 8 contrast between the rainfall values of consecutive years. The study of disparity is an 9 indicator of inter-annual rainfall irregularity, for which there is a rising interest within 10 the field of climatology both its implications for water resources management and from 11 the perspective of climate change and the consequent increase in the occurrence of 12 extreme events. Although these measurements are not strictly predictive, they can 13 conducive towards the improvement of water management practices along with other 14 climatic information. 15 Previous research of the authors of this article in the Southwest of the Iberian Peninsula 16 has focused on the analysis of rainfall annual and seasonal trends, highlighting a clear 17 and statistically significant downward trend in springtime precipitation, which 18 amounted to a 30 % reduction in the course of the 20th century (García-Barrón 2002a), 19 especially during the month of March (Aguilar et al. 2006; Aguilar 2007), and may have 20 been the cause of important environmental impacts (Sousa et al. 2006). 21 Irregularity is also a key factor in decision making and planning, both in agriculture and 22 urban water supply management, introducing risk and uncertainty in forecasting and 23 water assessment. Long term socio-economic planning should be based on the 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 5 knowledge of climatic time series for extended periods. So far, most of the studies 1 concerning rainfall irregularity in the Mediterranean have been devoted to the analysis 2 of extreme events due to the social and economic impacts of floods and droughts (Pita 3 1995; García-Barrón 2002b; Vicente-Serrano 2006; Cuadrat et al. 2007). However, only 4 few studies have related rainfall irregularity with its implications for water management 5 and planning (Giansante et al. 2002). Water management and planning require that 6 consideration is given to the pluri-annual periods in rainfall behaviour as the basis for 7 predicting water availability in reservoirs. The goal of this article is to contribute to a 8 better knowledge of such behaviour. 9 Based on previous research on the identification of trends in precipitation, the overall 10 purpose of this study is to further develop the in-depth analysis of rainfall variability 11 and its temporal evolution for some historical rainfall series from the Southwest of the 12 Iberian Peninsula. For this purpose, the article discusses and applies standard statistical 13 techniques for the description of the temporal evolution of variability and proposes two 14 new indexes that indicate the relative strength of variability in a given lag of time. 15 Although all the sections in the text are obviously linked together, the results obtained 16 in each of them are consistent separetely when analyzing the precipitation series. For 17 this reason, the results obtained are described and discussed for each of the sections. 18 Section 2 describes the study area and the meteorological station from which data have 19 been registered. The environmental characteristics of this area are also described 20 enphasizing the possible impacts caused by an increase in rainfall irregularity. 21 Section 3 provides a statistical description of data and methods used to study the 22 evolution of irregularity, including the definition of two new indices (subsections 3.3 23 and 3.4). Section 4 the cumulative deviation to the mean is used to define 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 6 predominantly dry or wet periods and a moving variation coefficient is calculated. In 1 Section 5 a general disparity index is introduced and calculated to describe the annual 2 rainfall series through a single value. A particular disparity index to measure the degree 3 of change in the variability over time is obtained (Section 6). 4 2. STUDY AREA AND DATA 5 The study area is in the Southwest of the Iberian Peninsula. Only the Iberian 6 Southwestern meteorological stations under Atlantic influence with rainfall data series 7 of over a century were selected for study (Figure 1 and Table1). The records analysed 8 cover the 1882-2005 periods, except for the station in Cordova, whose records started in 9 1894, and Sao Bras de Alportel in 1901. Therefore, the time scale exceeds one century 10 and the surface cell is approximately 400 x 400 km. 11 Fig.1 around here 12 Table 1 around here 13 The region in which the study area is located is particularly interesting from a climatic 14 point of view for a number of reasons: (1) its transitional position between the mid- 15 latitude and subtropical climates; (2) its position between two continents and two water 16 bodies (the Atlantic Ocean and the Mediterranean Sea) and (3) the complex and diverse 17 topography of the Iberian Peninsula relief, which increases the overall irregularity of the 18 climatic variables. Given the high degree of correlation between the precipitation values 19 of the different observatories of the area, the region can be represented by a single 20 precipitation series. Such series will hereafter be referred to as the Southwestern Series. 21 22 The northern part of the Iberian Peninsula was chosen in order to compare the 23 irregularity pattern of the Southwest zone with another climatic area with different 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 7 rainfall regimes. In the Cantabric area (Bay of Biscay) was four meteorological stations 1 were selected (see Figure 1): Gijon (Asturias), Santander (Cantabria), Bilbao (Vizcaya) 2 and San Sebastian (Guipuzcoa). This group of stations will be called “North-Atlantic” 3 and the Southwest observatories “South-Atlantic”. 4 2.1. Hydrological systems and natural vegetation 5 The main ecosystems in this region are highly adapted to irregular precipitation 6 conditions, as it is the case of the most widespread forest trees in the Iberian Peninsula’s 7 are mainly evergreen oaks (Quercus rotundifolia) and cork oaks (Quercus suber) sensu 8 Galiano (1987). However, due to the existence all along the SW coast of the Iberian 9 Peninsula of continental wetlands, which are highly sensitive to the rainfall regime, the 10 impacts of climate variability upon these valuable natural systems is of particular 11 interest (Sousa et al. 2009, 2010). 12 Assessing the environmental effects that could arise as a result of an increase of rainfall 13 irregularity in the SW of the Iberian Peninsula is extremely complex due to the fact that 14 Mediterranean ecosystems are adapted to high annual and seasonal rainfall variability. 15 Within the hydrological system, river courses —that do only exceptionally have large- 16 volume water flows— show a great drainage capacity with a wide but deep talweg. 17 Many wetlands in the area under study are recharge in a rather epigeal manner 18 (depending directly on the infiltration of rainfall, runoff water or subsurface water). 19 Less frequent are hypogeal water inputs in ponds (depending on water table level) or 20 peat-bogs that infrequently appear in Mediterranean environments. This explains why, 21 in response to past climatic changes (such as the end of the Little Ice Age), each type of 22 wetland in the SW of the Iberian Peninsula has evolved diffently (Sousa and García- 23 Murillo 2003). 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 8 On the other hand, during the last few centuries, the number and the length of short 1 seasonal streams on the Spanish Atlantic coast have decreased radically. This process 2 has intensified from the end of the XIXth century caused by rainfall changes and an 3 increase in the erosion processes that has lead to the clogging of these beds (Sousa and 4 García-Murillo 2001). The intensification of rainfall irregularity increases the risk of 5 this phenomenon (even if other climatic factors are exerting their influence 6 simultaneously). It is especially relevant that a similar process is detected in the creeks 7 of the Portuguese Algarve region (Devereux 1982), indicating that this similar process 8 may be affecting, at least, the whole Southwestern European coast. 9 Vegetation species that predominate in the SW of the Iberian Peninsula, bordering the 10 north of Africa, are those adapted to the conditions of the Mediterranean summer. As 11 pointed out by Rivas-Martínez (1988), in spite of its closeness to the Atlantic Ocean, the 12 Western Andalusian rainfall is markedly Mediterranean, even more so than in the 13 Spanish territories on the Mediterranean Sea coast. To a certain extend, the natural 14 adaptation capacity Mediterranean natural ecosystems prepare them for changes in 15 rainfall irregularity. However, the long-term vegetation show important changes related 16 to water availability indicating that further analysis is required in relation to rainfall and 17 water balance changes. 18 Woods and shrublands in the Southwest of the Iberian Peninsula, such as cork oak or 19 dense juniper trees have a xerophytic character and, therefore, are adapted to the 20 deficient water balances and to the absence of rainfall during the summer season. 21 However, an increase in irregularity may well affect the natural renovation, regeneration 22 and reproduction processes of these plant communities. Thus, an increase in rainfall 23 irregularity together with global warming may intensify the mobilisation of the sandy 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 9 layers and, thus, affecting the regeneration of tree species of the European Southwestern 1 coast, such as the cork oak (Quercus suber), the savin (Juniperus phoenicea subsp. 2 turbinata) or the juniper (Juniperus oxycedrus subsp. macrocarpa). This process has 3 clearly been detected in ancient climatic periods in the same area although at the present 4 time it could be masked by the impact of human activity (Sousa and García-Murillo 5 2003; Sousa et al. 2006). 6 2.2. Water resources and demand 7 Annual renewable (or natural) water resources can be defined as the long-term average 8 freshwater volume supplied naturally by the hydrological cycle, which is the result of 9 total run-off (surface and underground) minus evapotranspiration. To compensate for 10 inter-annual irregularity natural ecosystems accumulate water in the subsoil. However 11 annual renewal rates are also subject to variability. Spatial and temporal variability in 12 rainfall imply that only a portion of these natural resources is actually available at the 13 time and place where it is required. These annual available resources are obviously 14 affected by technological constrains, as well as by socio-economic and institutional 15 considerations. In regions with high climatic variability water management requires the 16 regulation of water flows through a network of reservoirs that store excess resources for 17 their later use and laminate floods. The remarkable variability has historically had 18 important implications for the institutions that must try to manage the irregularity and 19 deal with the risk of water scarcity. 20 The study area is included within two water basins: the Guadalquivir river Basin sensu 21 lato (Cordova, Riotinto San Fernando and Seville) and the Spanish-Portuguese 22 Guadiana river basin (Badajoz). The most important factor in the water resources use 23 and planning of these basins is the prime importance of irrigation agriculture. Although 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 16 Idi =({[ (pi - pi-1)2 + (pi+1 - pi )2] /2}1/2 ) /  i (9) 1 The Specific Disparity Index has the same properties as previously assigned to the 2 General Disparity Index. It should be noted that when disparity is obtained through the 3 squares of the deviations between consecutive elements, so the deviation sign is 4 irrelevant. 5 In each meteorological station the Specific Disparity Index is calculated for each year i 6 (Id) generating a new time series of disparity. 7 4. CUMULATIVE RAINFALL DEVIATIONS 8 Table 2 show a descriptive summary of the annual rainfall series with the average 9 values and variation coefficient for each meteorological station in the study area. 10 Table 2 around here 11 One of the main features of the Mediterranean climate is a marked variability in 12 precipitation patterns. Temporal irregularity has two components: seasonal/intra-annual 13 variability and inter-annual variability. The seasonal variability implies the alternation 14 of a dry season of five to six months, when evaporation exceeds precipitation, and a 15 rainy winter season, which accounts for most of the annual rainfall. The range, 16 persistence and extreme values of inter-annual variability are also extremely high. 17 Torrential precipitation can be extremely intense in most of the area being sometimes 18 over 150-200 mm in twenty-four hours, which represents one-third of the total annual 19 rainfall values (Giansante et al. 2002). 20 The average annual precipitation values are scarce – generally below 600 mm - for most 21 of the meteorological stations. Only the Riotinto and Sao Bras mean value is higher 22 (Table 2) due to the fact that this stations is located on the Southern hillside of Sierra 23 Morena (Sao Bras is closer to the sea). The orientation of these mountains produces a 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 17 barrier effect against the humid winds entering the Southwest of the Iberian Peninsula 1 from the Atlantic Ocean. In order to assess the overall variability for each one of the 2 meteorological stations, the variation coefficient was used as a statistical measure of the 3 deviation of a variable from its mean. 4 In the Iberian Peninsula annual coefficients below 0.2 only occur in the so-called 5 “Green Spain” in the Northern and North-Western part of the Iberian Peninsula under 6 Atlantic influence. Coefficients below 0.3 appear in the Central Plateau and the Douro, 7 Tagus and Ebro river basins whereas values above 0.3 are characteristic of the 8 Mediterranean domain. These geographical differences are consistent with the spatial 9 pattern of daily precipitation described by Martín-Vide (2004) despite the different time 10 scale of analysis. 11 4.1. Analysis of the cumulative rainfall deviations 12 Using Equation 2 the cumulative deviation for the Southwestern series is obtained. 13 Figure 2 represents this series from 1885 to 2005. Downward segments between 14 following years indicate that annual rainfall is below the mean. The increase in the 15 number of these segments within an interval identifies dry periods. Rainfall evolution in 16 the Southwest area of the Iberian Peninsula can be described as follows. After the wet 17 period at the end of the XIXth century the beginning of the XXth is dominated by dry 18 years (1898-1934). The second period (1934-1957) begins with low precipitation values 19 reaching a pick in the early 40s. The last period starts with a steep increase in the 20 precipitation values reaching a maximum value in 1970 followed by a gradual decrease 21 until the end available data. 22 Fig.2 around here 23 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 18 No attempt is being made to identify drought periods or their intensity. Therefore, this is 1 a less stringent method than those used to elaborate the most frequently used drought 2 indexes (Pita, 1995). 3 If individual rainfall series were analysed a similar pattern would be identify despite the 4 lack of a complete coincidence among the meteorological stations under analysis. The 5 definition of these periods coincides with those described by Aguilar and Pita (1996) for 6 the Spanish Southwest. 7 4.2. Analysis of the long-term variability 8 Each year is assigned with a variability number obtained with the Coefficient of 9 Variation calculated with the ten previous values. Figure 3A and B shows the evolution 10 of 11-year running coefficient of variation for (A) San Fernando and Seville and (B) 11 Sao Bras and Riotinto, and their corresponding long-term variability obtained by a 12 polynomial fitting line. Long-term variability values were high at the end of the 19th 13 century and decreased later between 1900 and 1930. Since then the curve has been 14 increasing sharply up to 1950 and more smoothly since the mid 70s. The general pattern 15 described above applies to all the rainfall series and throughout the period, although 16 some difference can be detected. Overall the most interesting feature is the clear 17 increase in variability in the last three decades of the 20th century in all meteorological 18 stations. 19 Fig.3 around here 20 Figure 4 shows the results obtained from the six meteorological stations studied 21 together (Badajoz, Cordova, San Fernando, Sao Bras de Alportel, Seville and Riotinto). 22 Thus, the overall variability of the SW area of the Iberian Peninsula can be appreciated 23 by smoothing the extreme anomalies of each station when constructing a single serie. 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 19 Fig.4 around here 1 5. THE DISPARITY IN RAINFALL. THE GENERAL DISPARITY INDEX 2 The General Disparity Index provides a single value (Equation 3) for each annual 3 rainfall series in each meteorological station. Values of this index are shown in Table 3. 4 A positive association takes place between the values of the variation coefficients in the 5 complete series, as shown in Section 4, and the corresponding general disparity indexes. 6 It should be noted that the variation coefficient is not dependent on the temporal order. 7 It only depends on the deviations of elements from the average. On the other hand, two 8 sets of data with the same elements show different disparities as a result of their relative 9 positions, i.e. as a function of the chronological order of the series. The General 10 Disparity Index is larger than the variation coefficient to the mean minimizes the 11 procedure when the distribution approaches randomness. 12 Table 3 around here 13 The annual rainfall irregularity in the Iberian Peninsula has been characterized by 14 several authors. Lana and Burgueño (2000) have elaborated non-dimensional indexes 15 based directly on the addition of the absolute differences between consecutive years of 16 the series, divided by the mean rainfall average and the number of years. A drawback of 17 this method is that different series with the same mean value and the same addition of 18 absolute differences will generate the same value. The general disparity index 19 overcomes this problem by amplifying the effect of irregularity due to the “2nd-level 20 differences”. For the same sum of absolute values of the difference between the central 21 element and the lateral ones, dab and dbc, the disparity index is higher when the 22 difference │dab -dbc│ is higher. 23 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 20 Instead of using the differences of the absolute values of consecutive elements divided 1 by the number of years, Martín-Vide et al. (2001) used the logarithm of the absolute 2 value of the ratios. Nevertheless, this method does not take into account that it assigns 3 equal values to non-comparable transitions like decreasing from 600 mm to 300 mm 4 (half), and increasing from 600 mm to 1200 mm (double). Another indirect method is 5 proposed to asses irregularity based on fractal dimension, the rescaled range analysis, 6 associated to the internal dependence of the series. Warning about its generalised 7 application is given as the result is affected by the sample size used in the procedure 8 (Martín-Vide et al. 2001). 9 There should not be differences between disparity indexes measuring the same 10 phenomenon. A high degree of correspondence (with a correlation coefficient greater 11 than 0,95) is observed between the three indexes, when considering syncronous 12 precipitation series of more than 30 years and from observatories belonging to the same 13 geographical unit. Any of these indexes can be used to compare the irregularity between 14 the different observatories. Although the absolute values of these indexes may not be 15 the same, the spatial distribution of disparity within the same geographical unit is 16 similar no matter which method is used. 17 As mentioned before, these methods provide a single value for each meteorological 18 station and therefore they are focused on the spatial distribution of rainfall irregularity. 19 Specific Disparity Index is presented in the next section to achieve the main goal of this 20 study: rainfall irregularity evolution. 21 6. THE DISPARITY IN RAINFALL. SPECIFIC DISPARITY INDEX 22 The Specific Disparity Index is calculated for each year of the rainfall series in each 23 meteorological station using Equation 9 presented in subsection 3.4. Figure 5 show the 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 21 evolution of the Specific Disparity Index in Seville-Sao Bras and in Badajoz-Cordova 1 areas. When compared with other variables, the most outstanding characteristic in the 2 graphs is jigsaw evolution of the index and the wide range of variation during short 3 periods of time. However, in the course of a century, an outstanding stability of the 4 mean can be observed. A similar behaviour of the index is also observed within each 5 region. If the XXth century is subdivided into four quarters, the first quarter can be 6 visualised as one with plenty of low-disparity years; the second quarter has plenty of 7 high disparity years; the third has low values again and finally, in the last quarter of the 8 century, both the maximum and minimum increase. 9 Fig.5 around here 10 Figure 6A show the disparity index values during the last decades of the XXth century 11 calculated for the series of Seville highlighting a linear growth trend. Unlike the 12 previous decades, a marked growing trend and persistence is observed (continuous line). 13 Figure 6B represents the Southwestern series of annual average values of the index, 14 which is similar to the one from Seville station. It is noteworthy that there is a upward 15 trend in the set of maximum values (dotted line); this justifies 65 % of the internal 16 variance. From this data we can inferred that this permanent growth in irregularity is a 17 characteristic of the rainfall evolution in the South Atlantic area of the Iberian 18 Peninsula, and has not previously been observed with such intensity and duration. 19 Fig.6 around here 20 Table 4 shows a summary of the average values and standard deviation of the Idi series. 21 For each of the meteorological stations, the average value of the Specific Disparity 22 Index is lower than the corresponding value of the General Disparity Index, although a 23 good association level is maintained. The variation coefficient values in these disparity 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 22 series are far higher than those in the original rainfall series indicating that the method 1 highlights rainfall irregularity and is suitable for its analysis. 2 Table 4 around here 3 The Table 5 showing the correlation coefficients among the series of disparity indexes, 4 that range from 0.5 to 0.7 (relatively low thus indicating that the temporal evolution of 5 the meteorological stations of Southwestern are not fully uniform). 6 Table 5 around here 7 In order to compare rainfall irregularity evolution in the Southwest of the Iberian with 8 another climatic zone a group of meteorological stations have been selected from the 9 North of the Iberian Peninsula. A regional “North-Atlantic” series was calculated for the 10 period 1975 to 2005 being each year value the mean from the stations rainfall records 11 (Gijon, Santander, Bilbao, San Sebastian). Table 6 and Figure 7 show the differences 12 between the two climatic groups of stations in the Iberian Peninsula (South-Atlantic and 13 North-Atlantic) in terms of their mean values and Disparity Indices calculated with 14 different procedures from 1950 to 2005. 15 Table 6 around here 16 The Specific Disparity Index evolution (Figure 7A) and the 11-year Moving Coefficient 17 of Variation (Figure 7b) are presented for both groups of meteorological stations, 18 “South-Atlantic” and “North-Atlantic”. This Figure 7 clearly shows the higher 19 variability of the Southwest observatories (“South-Atlantic”) from the North stations of 20 the Iberian Peninsula (“North-Atlantic”). Both areas also differ in their temporal 21 evolution, something that had never been concluded in other publications. Figures 7A 22 and 7B show the different values of the Specific Disparity Index and the Coefficient of 23 Variation in the Cantabric area (Bay of Biscay). Figure 7A does not show an increase in 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 23 irregularity in the last decades, something that is outstanding for the meteorological 1 stations of the Southwest area (“South-Atlantic”). 2 Fig.7 around here 3 7. CONCLUSIONS 4 This article characterises the rainfall evolution throughout the XXth century in the South 5 Atlantic area of the Iberian Peninsula. The annual rainfall series has been submitted to 6 three methods during the analysis: deviations to the mean, variability and disparity. For 7 each case, new temporal series were generated. The proposed indexes are mutually 8 complementary and should be viewed as consecutive steps in the study of precipitation 9 irregularity. The precipitation cumulative deviation graph is a first easy approach to the 10 analysis of pluviometric disparity, though the main purpose of this technique is to show 11 the temporal distribution of dry and wet spells in the series. The cumulative deviations 12 discriminates changes between dry and wet periods of the annual rainfall series. 13 The evolution of irregularity was studied by means of the running variation coefficient, 14 calculated for an eleven-year period. The results indicate a relative stability during the 15 first third of the XXth century and a progressive increase during the last three decades. 16 During the first third of the century, the low variability coincides with a period of low 17 precipitation. The most outstanding feature is the increase of the inter-annual rainfall 18 variability in the Southwest of the Iberian Peninsula during the last third of the XX 19 century up to the present. 20 In order to characterise the irregularity of a series, better results are obtained with the 21 general disparity index calculated with the consecutive terms of the series rather than 22 with the variation coefficient based on the deviations from the mean. Other disparity 23 index proposed by other authors shows high correlation when applied to long series. 24 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 24 These indexes can be used to analyse the spatial distribution of disparity. To this 1 respect, the general disparity index gives insights into the aspects of rainfall irregularity, 2 but is not suited for a diachronic analysis. 3 The specific disparity index gives more accurate information on the temporal evolution 4 of irregularity in precipitation series and allows making comparisons between different 5 periods of the series. Therefore, the specific disparity index is more suited to dynamic 6 studies. As a result, the general and specific disparity indexes complement each other in 7 temporal and spatial distribution analysis of pluviometric irregularity. 8 The use of the Specific Disparity Index has proved useful in highlighting the irregularity 9 within the rainfall series at each meteorological station. Disparity is higher between 10 1925 and 1950, with a continued growth during the last quarter of the XXth century. It 11 therefore seems that the Southwest of the Iberian Peninsula has experienced a 12 progressive increase of rainfall irregularity in terms of its variability and disparity. 13 Although Mediterranean ecosystems are adapted to important fluctuations in the rainfall 14 regime of these areas, this increase in irregularity may affect rivers, wetlands and the 15 vegetation (in particular hygrophytic vegetation, as well as the regeneration of some 16 xerophytic woody communities). In fact, an important reduction has been detected in 17 the number and length of the seasonal streams of the European Southwest area (Sousa 18 and García-Murillo 2001). 19 Both the numerical analysis and the observation of the graphs from each one of the 20 meteorological stations confirm this positive trend in rainfall irregularity. If this trend 21 goes on in the future, it is likely that the frequency and length of both wet and dry spells 22 will increase. In conclusion, the new index improves the existing procedures for the 23 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 25 analysis of precipitation irregularity and could contribute to monitor future changes in 1 precipitation within the general framework of research on climate change. 2 ACKNOWLEDGEMENTS 3 Special thanks to Consuelo Giansante for her enormous work revising the English 4 manuscript and for her valuables comments and suggestions. We are grateful to the 5 Spanish National Agency of Meteorology (AEMet) for providing the precipitation data 6 to carried out this study. This study has been financed by the Spanish Ministry of 7 Education and Science (Project CGL2006-07194/BOS and Project CGL2009-10683). 8 REFERENCES 9 Aguilar M (2007) Cambios y tendencias recientes en las precipitaciones de Andalucía. In: 10 Sousa A, García-Barrón L, Jurado V (eds) Climate Change in Andalusia: trends and 11 environmental consequences. Consejería de Medio Ambiente de la Junta de Andalucía, 12 Seville, pp 99-116 [Online]. Available: 13 http://www.juntadeandalucia.es/medioambiente/web/Bloques_Tematicos/Educacion_Y_Participacion_Amb 14 iental/Educacion_Ambiental/el_cambio_climatico_en_andalucia/capitulo5.pdf 15 Aguilar M, Pita MF (1996) Evolución de la variabilidad pluviométrica en Andalucía: su 16 repercusión en la gestión de recursos hídricos. In: Marzol MV, Valladares P (eds) Clima 17 y agua: la gestión de un recurso climático. Tabapress, Madrid, pp. 299-310 18 Aguilar M, Sánchez E, Pita MF (2006) Tendencia de las precipitaciones en marzo en el 19 sur de la Península Ibérica. In:. Cuadrat JM et al (eds) Publicaciones de la Asociación 20 Española de Climatología (AEC) Serie A 5, Zaragoza, pp 41-51. [Online] Available: 21 http://www.aeclim.org/images/stories/articulos_pdf/Vcongreso/ponencia_cambio_climatico/aguilar%20alba 22 .pdf 23 Alexandersson H (1986) A homogeneity test applied to precipitation data. J Climatol 24 6:661-675 25 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 32 Legend of the figures 1 Fig. 1 Map with the location of the meteorological stations used in the study area 2 Fig. 2 Cumulative deviations of annual rainfall of the Southwestern series 3 Fig. 3 Evolution of 11-year running coefficient of variation for (A) San Fernando and 4 Seville and (B) Sao Bras and Riotinto, and their corresponding long-term variability 5 obtained by a polynomial fitting line 6 Fig. 4 Evolution of 11-year running coefficient of variation for Southwestern series 7 obtained by polynomial fitting line method. 8 Fig. 5 Specific Disparity Index in (a) Seville and Sao Bras and (b) in Badajoz and 9 Cordova areas 10 Fig. 6 Specific Disparity Index values during the last decades of the XXth century 11 calculated for the series of Seville (a) and the Southwestern series (b). Linear fitting 12 model to the series (continous line), relative maximum values and explained variance. 13 Fig 7. Rainfall irregularity evolution comparison between the SW area (South-Atlantic) 14 and the North area (North-Atlantic) of the Iberian Peninsula (A) Moving Variation 15 Coefficient and polynomial fitting (B) Specific Disparity Index and polynomial fitting. 16 17 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 33 Figure 1 1 2 3 Figure 2 4 5 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 34 Figure 3 1 2 Figure 4 3 4 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 35 Figure 5 1 2 Figure 6 3 4 Figure 7 5 6 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 36 Legend of the tables 1 Table 1 Geographical coordinates, province, time period and institution responsible of 2 data 3 Table 2 Mean values (mm) and Variation Coefficient of the annual rainfall series of the 4 Southwest observatories of the Iberian Peninsula 5 Table 3 General Disparity Index (ID), index from Lana and Burgueño (2000) and 6 Martín-Vide et al. (2001) of the annual rainfall series of the Southwest observatories of 7 the Iberian Peninsula 8 Table 4 Statistical characteristics of the annual Specific Disparity Index of the 9 Southwest observatories of the Iberian Peninsula 10 Table 5 Correlation coefficients among the series of Specific Disparity Indexes of the 11 Southwest observatories of the Iberian Peninsula 12 Table 6 Rainfall series and General disparity Index comparison between the SW area 13 (South-Atlantic) and the North area (North-Atlantic) of the Iberian Peninsula (1950- 14 2005) 15 16 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 37 Table 1 1 Station Time period Province Longitude Latitude Institution Porcentage of missing values Southwest of the Iberian Peninsula (“South-Atlantic”) Badajoz 1882-2005 Badajoz 6º 49'45'' W 38º 43' 02'' N Spanish National Agency of Meteorology 1.2 Cordova 1894-2005 Cordova 4º 51' 02'' W 37º 50' 40'' N Spanish National Agency of Meteorology 0.2 Riotinto 1882-2005 Huelva 6º 36' 17'' W 37º 42' 00'' N Riotinto Mining Company 0.0 San Fernando 1882-2005 Cadiz 6º 12' 20'' W 36º 27' 56'' N Spanish Navy 2.9 Sao Bras de Alportel 1901-2005 Algarbe 7º 52' 12'' W 37º 08' 59''N Portuguese Institute of Meteorology 4.5 Seville 1882-2005 Seville 6º 00' 30'' W 37º 21' 55'' N Spanish National Agency of Meteorology 0.7 North of the Iberian Peninsula (“North-Atlantic”) Bilbao 1950-2005 Vizcaya 2º 54' 21'' W 43º 17' 52'' N Spanish National Agency of Meteorology <1.0 Gijon 1950-2005 Asturias 5º 18' 31'' W 43º 32' 18'' N Spanish National Agency of Meteorology <1.0 San Sebastian 1950-2005 Guipuzcoa 2º 02' 22'' W 43º 18' 27'' N Spanish National Agency of Meteorology <1.0 Santander 1950-2005 Cantabria 3º 49' 10'' W 43º 25' 48'' N Spanish National Agency of Meteorology <1.0 2 3 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 38 Table 2 1 2 3 Badajoz Cordova Riotinto San Fernando Sao Bras Alportel Seville Mean 471.7 640.8 750.6 565.6 820.3 574.4 Variation Coefficient 0.27 0.34 0.32 0.28 0.31 0.32 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 39 Table 3 1 2 3 Badajoz Cordova Riotinto San Fernando Sao Bras Alportel Seville ID 0.36 0.42 0.45 0.38 0.42 0.44 Lana and Burgueño (2000) index 0.28 0.33 0.35 0.30 0.34 0.35 Martín-Vide et al. (2001) index 0.28 0.35 0.37 0.32 0.35 0.36 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 40 Table 4 1 2 3 Badajoz Cordova Riotinto San Fernando Sao Bras Alportel Seville Mean 0.31 0.37 0.39 0.33 0.37 0.38 Standard Deviation 0.17 0.19 0.21 0.18 0.17 0.21 Evolution of rainfall irregularity in the SW of the Iberian Peninsula 41 Table 5 1 Riotinto San Fernando Seville Cordova Sao Bras Alportel Badajoz 0.65 0.56 0.58 0.65 0.58 Riotinto 1 0.56 0.61 0.55 0.66 San Fernando - 1 0.64 0.51 0.59 Seville - - 1 0.67 0.50 Cordova - - - 1 0.50 2 3