CMIP5 future projections in the main water fluxes participating in soil-atmosphere interaction

PENALBA OC; PANTANO VC

Venecia

Congreso; Sixth Annual Conference ?Recent trends in climate sciences, adaptation and mitigation; 2018

The strong and complex soil-atmosphere interaction implies projected changes in precipitation and evapotranspiration impacting on the soil water balance. The several variables involved in this interaction are calculated differently in each Global Climate Model (GCM), according to their physical and thermodynamics equations and specific parameterizations. Therefore, it is necessary to identify those soil water fluxes sensible to changes in the atmosphere for each GCM before analyzing soil response to climate change. In this study,future changes in water fluxes involved in soil-atmosphere interaction are analyzed based on future scenarios of climate change from different GCMs in South America, with special emphasis on south-eastern South America. Firstly, we analyze future changes of precipitation and evapotranspiration, as the main outgoing and incoming fluxes of water. Secondly, focused on a rainfed agriculture region in south-eastern South America, we analyze the climate sensitivity of soil moisture and runoff to precipitation and evapotranspiration. The interest of this region is because the rainfed agricultural production is one of the main economic activities. This region is characterised by a strong feedback between soil moisture and precipitation through the role of evapotranspiration. Therefore, the study of these water fluxes is of special interest for the regional economy because of the impact on the crops´ yields.Monthly precipitation, evapotranspiration, surface and total runoff, surface and total soil moisture content were obtained from seven Global Climate Models selected from the Phase 5 of the Coupled Model Inter-comparison Project (CMIP5): ACCESS 1.0, CanESM2, CESM1 (CAM5), EC-EARTH, IPSL-CM5AMR, MIROC5, MPI-ESM-MR. Two different periods were used: 1970?2005 as climate reference for historical experiment and 2065?2100 for projections under the Representative Concentration Pathways scenarioRCP 8.5.As an example, monthly results are here described for January and July.Initially, precipitation was validated in comparison with reanalysis data from GPCC Full Data Reanalysis Version 6. In January, precipitation is underestimated over the Amazonas and in the south-east of South America by more than 50% in percentage bias whereas it is overestimated over the Andes, south of 40°S and northeast Brazil. In July, precipitation is underestimated in almost all South America except for the Andes and Patagonia. ACCESS 1.0 behaves differently, with overestimation in almost all South America.The agreement between the seven GCMs with respect to projected increases and decreases in precipitation and evapotranspiration was quantified for South America. In particular, future changes were considered when they are higher than 10%. During January, most of the models agree on increase in south-eastern South America, eastern Brazil and Peru, whereas decrease is predicted for extreme north and extreme south of South America. During July, most of the models agree on decrease in the east and northeast of Brazil and center of Chile. Evapotranspiration is also projected to increase in most of South America, but the increase is higher than 10% only in south-eastern South America during January and extreme south of South America during July. In the region of interest, south-eastern South America, both increases in precipitation and evapotranspiration are compensated and the balance between both variables (PET=precipitation-evapotranspiration) presents small change.In order to evaluate how sensitive are soil moisture and runoff to PET values, different conditional probabilities were quantified in south-eastern South America. The coherence was evaluated considering probabilistic intervals of percentiles (P) of the entire frequency distribution: P80. Contingency Table of conditional probabilities was tested through Chi-Square distribution. This analysis reveals which are the more sensible variables for each GCM: surface soil moisture and surface runoff for ACCESS 1.0; surface and total soil moisture and total runoff for CanESM2; surface soil moisture and surface and total runoff for CESM1-CAM5; none for EC-EARTH; surface soil moisture IPSL-CM5A-MR; surface soil moisture for MIROC5; surface and total runoff for MPI-ESM-MR.Finally, future changes in the more sensible variables were analyzed. For example, increase is projected in surface runoffby ACCESS 1.0and CESM1-CAM5in both January and July whereas small change is projected by MPI-ESM-MR. Similarly, increase is projected in total runoff by CanESM2 and CESM1-CAM5 in both January and July whereas small change is projected by MPI-ESM-MR. In the cases that models project increase in runoff, this variable is better responding to changes in precipitation than changes in evapotranspiration. In conclusion, soil response to changes in the interaction with atmosphere (through precipitation and evapotranspiration fluxes) depends on each specific GCM. Even though both precipitation and evapotranspiration are projected to increase, some models showed that the more sensible soil fluxes respond to increases in precipitation. Based on the results of this study, the ensemble of soil moisture and runoff is not recommended but we better suggest to study separately only for those variable sensible to climate change.