Analysis of the past and future water resources of the Pyrenees by means of a land-surface simulation

The Pyrenees are the "Water Towers" of several key river basins in France, Andorra and Spain, being the Adour-Garonne and the Ebro the largest ones. The water of these basins is used by agricultural and industrial economic sectors which have a significant socioeconomic impact. Furthermore, the water of these rivers also sustains ecosystems which have an intrinsic value and provide ecosystem services to society. For this reason, an assessment of the past and future evolution of the water resources of the Pyrenees is necessary. Until now, these assessments have often been done at the basin or at the national level, but never the water resources of the Pyrenees were assessed as a whole. This is the main aim of the PIRAGUA project, within which we develop our research.

In order to simulate the continental water cycle of the Pyrenees we have used the SASER (SAFRAN-SURFEX-Eaudyssée-RAPID) modeling chain. SAFRAN is a meteorological analysis system, that allows us to create a gridded dataset of all the variables needed by the SURFEX land-surface model. SURFEX’s outflows (runoff and drainage) are used by Eaudyssée and RAPID to calculate streamflow.

Until now there were two separate implementation of SAFRAN in France (8 km resolution) and Spain (5 km resolution). For this project we have taken the climatic zone level SAFRAN data of both countries and interpolated it to a new common grid at a resolution of 2.5 km. The dataset covers a domain that includes the Adour-Garonne, the Ebro and all other Pyrenean river basins, its time period is 1979/80-2014/15 (which will be extended to 2016/17). The RAPID river routing scheme has been implemented in the simulation domain using HydroSheds to describe the river network.

In order to simulate the future evolution of the continental water cycle we use the Pyrenean climate scenarios developed within the CLIMPY project. These include precipitation and maximum and minimum temperature. SURFEX needs other variables too, such as wind speed, relative humidity and radiation. We solve this problem using an analog based approach similar to Clemins et al (2019).

The simulated streamflow is compared to observed streamflow of natural basins. The results show that 18 (out of 38) non influenced stations present a KGE of daily streamflow larger than 0.5. For monthly streamflow, KGE is larger than 0.5 on 22 stations (out of 38).

The next steps of our research are to quantify the improvement due to the increased resolution (comparing to a lower resolution simulation), calculate trends of relevant variables at the sub-bassin scale and compared them to the observed ones in the past, and analyze future trends of these variables. Finally, we will assess the impacts of these changes on water resources.

This research is funded by the EFA210/16-PIRAGUA project, within the INTERREG V-A España-Francia-Andorra POCTEFA2014-2020 program.