Hydrol. Earth Syst. Sci., 24, 4503–4521, 2020 https://doi.org/10.5194/hess-24-4503-2020 © Author(s) 2020. This work is distributed under the Creative Commons Attribution 4.0 License. Specific climate classification for Mediterranean hydrology and future evolution under Med-CORDEX regional climate model scenarios Antoine Allam1,2, Roger Moussa2, Wajdi Najem1, and Claude Bocquillon1 1CREEN, Saint-Joseph University, Beirut, 1107 2050, Lebanon 2LISAH, Univ. Montpellier, INRAE, IRD, SupAgro, Montpellier, France Correspondence: Antoine Allam ([email protected]) Received: 18 February 2020 – Discussion started: 25 March 2020 Accepted: 27 July 2020 – Published: 16 September 2020 Abstract. The Mediterranean region is one of the most sen- the 2070–2100 period served to assess the climate change sitive regions to anthropogenic and climatic changes, mostly impact on this classification by superimposing the projected affecting its water resources and related practices. With mul- changes on the baseline grid-based classification. RCP sce- tiple studies raising serious concerns about climate shifts and narios increase the seasonality index Is by C80 % and the aridity expansion in the region, this one aims to establish aridity index IArid by C60 % in the north and IArid by C10 % a new high-resolution classification for hydrology purposes without Is change in the south, hence causing the wet sea- based on Mediterranean-specific climate indices. This clas- son shortening and river regime modification with the migra- sification is useful in following up on hydrological (water tion north of moderate and extreme winter regimes instead resource management, floods, droughts, etc.) and ecohydro- of early spring regimes. The ALADIN and CCLM regional logical applications such as Mediterranean agriculture. Olive climate models (RCMs) have demonstrated an evolution of cultivation is the characteristic agricultural practice of the the Mediterranean region towards arid climate. The classes Mediterranean region. The proposed approach includes the located to the north are slowly evolving towards moderate use of classic climatic indices and the definition of new cli- coastal classes, which might affect hydrologic regimes due matic indices, mainly precipitation seasonality index Is or to shorter humid seasons and earlier snowmelts. These sce- evapotranspiration threshold SPET, both in line with river narios might look favourable for Mediterranean cultivation; flow regimes, a principal component analysis to reduce the however, the expected impact on water resources and flow number of indices, K-means classification to distribute them regimes will surely expand and directly hit ecosystems, food, into classes, and finally the construction of a decision tree health, and tourism, as risk is interconnected between do- based on the distances to class kernels to reproduce the clas- mains. This kind of classification might be reproduced at the sification without having to repeat the whole process. The global scale, using the same or other climatic indices specific classification was set and validated by WorldClim-2 at 1 km to each region, highlighting their physiographic characteris- high-resolution gridded data for the 1970–2000 baseline pe- tics and hydrological responses. riod and 144 stations’ data over 30 to 120 years, both at monthly time steps. Climatic classes coincided with a geo- graphical distribution in the Mediterranean ranging from the most seasonal and driest class 1 in the south to the least sea- 1 Introduction sonal and most humid class 5 in the north, showing the cli- matic continuity from one place to another and enhancing the Mediterranean climate is a result of a complicated global cy- visibility of change trends. The MED-CORDEX ALADIN clonic system swiping a large evaporative basin. The distri- and CCLM historical and projected data at 12 and 50 km res- bution of marine and continental air masses creates an alter- olution simulated under the RCP4.5 and 8.5 scenarios for nation of low-pressure zones coming over from Iceland and Published by Copernicus Publications on behalf of the European Geosciences Union. 4504 A. Allam et al.: Specific climate classification for Mediterranean hydrology under Med-CORDEX RCM the Persian Gulf or high-pressure zones from Siberia and the method depends on the objective and its specificity. There Azores (Clerget, 1937). The seasonal shifts of these zones are several modes of climatic classification: (a) genetic clas- are magnified by the North Atlantic Oscillation (NAO) that sifications related to meteorological causes and the origin plays an important role in shaping Mediterranean climate of air masses (Bergeron, 1928; Barry and Chorley, 2009); and influencing the evolution of farming and social activities (b) bioclimatic classifications based on the interrelation be- in the long term (Rodwell and Hoskins, 1996). During the tween vegetation type and climate (Holdridge, 1947; Mather positive phases of the NAO, oceanic disturbances bring the and Yoshioka, 1968; Harrison et al., 2010); (c) an agro- most humid to northern Europe and the less humid to north- climatic method based on the assessment of the rainfall– ern Africa and the Middle East (Douguédroit and Lionello, evapotranspiration balance for the estimation of agricultural 2015). This continuous alternation of high- and low-pressure, productivity (Thornthwaite, 1948); and (d) climatic methods cold and humid winters followed by hot and dry summers based on precipitation and temperature indices similarly to marks the Mediterranean seasonality, which makes the re- the classification of Köppen (1936) updated by Peel in 2007 gion attractive for social activities, hence its sensitivity to cli- (Peel et al., 2007) and which remains the most used. mate change and anthropogenic pressures (PlanBleu, 2012). There are several climate classification studies of the A north–south general precipitation and evapotranspiration Mediterranean region; among these we cite Köppen–Geiger gradient has been identified in Tunisia through the analysis classification at the global scale in which the Mediterranean of directional variograms that results from partial gradients climate is very distinctive (Köppen, 1936; Peel et al., 2007; evolving through seasons (Slimani et al., 2007; Baccour et Eveno et al., 2016). Köppen’s classification divides the globe al., 2012; Feki et al., 2012). These spatial gradients mainly into 30 climate zones and relies on a partition hierarchy. depend on topographic structures through the interception It is based on precipitation and temperature indices where of rainfall-generating air masses. The climatic classification Mediterranean climate corresponds to dry hot or dry warm will try to identify the general spatial gradients across the summer where the precipitation in the driest month in sum- Mediterranean. Hydrologically, the precipitation seasonal- mer is either below 40 mm or below the third of the pre- ity characterizing the Mediterranean climate is reflected in cipitation in the wettest month in winter (Cs) and the air the flow regimes of Mediterranean rivers, as pointed out by temperature of the warmest month is above 22 ◦C (Csa) or Haines (1988), who classified the Mediterranean rivers under the number of months with air temperature above 10 ◦C ex- Group 12 Winter Moderate hydrologic regimes, Group 13 ceeds 4 (Csb). The (Cs) climate does not reign all over Extreme Winter, and Group 14 Early Spring and found a the Mediterranean region, as Köppen (B) classes are also clear relation to the Köppen Csa and Csb climates and a observable. (B) classes correspond to arid climate in gen- close equivalent of the “Mediterranean Seasonal” categories eral, with (BWh) the desertic and hot climate that domi- of Gentilli (Haines et al., 1988). Seasonality is the main fac- nates Egypt and Libya characterized by very low precipi- tor in the Mediterranean, but to our knowledge its use is still tation (MAP < 5 × Pthreshold with Pthreshold D 2 × MAT) and limited as a characterizing index for climatic and hydrologi- high temperature (MAT ≥ 18 ◦C), (BSk) the arid steppe cold cal classification. Climate change is expected to have severe climate that dominates south-eastern Spain characterized by consequences for Mediterranean runoff, with a serious risk low precipitation (5×Pthreshold > MAP > 10×Pthreshold) and of freshwater availability decrease of 2 % to 15 % for 2 ◦C low temperature (MAT < 18 ◦C), (Cf) the temperate climate of warming (Cramer et al., 2018) and a significant increase without any dry season in the regions of Thessaloniki and in drought periods, particularly in the south and east (Hre- Veneto, and finally (D) cold climate present further north. iche et al., 2007; Cudennec et al., 2007; Garcia-Ruiz et al., On the other hand, and at a global scale, some regions 2011; Verdier and Viollet, 2015). The CMIP5 simulations share a similar Mediterranean (Cs) climate, such as Cali- (Coupled Model Intercomparison Project, Phase 5) expected fornia, Chile, South Africa, and Australia (Fig. 1). Rivoire a mean precipitation decrease of −4 % ◦C−1 and a temper- et al. (2019) classified 160 Mediterranean rain gauges ac- ature increase of 20 %, more than the global average, with cording to monthly net precipitation (P − ET0). The clas- a maximum precipitation reduction reaching −7 % ◦C−1 in sification showed a marked distinction between two clus- winter in the southern Mediterranean region and −9 % ◦C−1 ters, with northern stations having a precipitation deficit from in the summer in the northern region (Lionello