Improved SAFRAN forcing and ECOCLIMAP-SG datasets to simulate irrigation over the Ebro basin

Irrigated agriculture plays an important role in the continental water and energy cycles of the basins where it is present. Land-Surface models (LSM) can be used to study and quantify the impact of anthropic processes on the continental water cycle. Therefore, it is necessary to have good quality forcing and physiographic data, including a correct representation of agricultural covers, irrigation methods and actual irrigated areas. 

This work presents four datasets, at a spatial resolution of 1 km, that have been prepared to simulate irrigation-related processes using a LSM over the Ebro basin, the largest Mediterranean Spanish basin, where irrigated agriculture has a large impact on the water cycle. These datasets are: (1) a land cover map, (2) an actual irrigated areas map, (3) a map of irrigation methods per area, (4) and a meteorological forcing dataset.

The most recent version of the ISBA LSM, in SURFEX v9, contains an improved irrigation scheme (Druel et al., 2021), which requires the also recent ECOCLIMAP-SG land cover map (Druel et al., 2021). We validated ECOCLIMAP-SG over the Ebro basin, using SIGPAC data (Agricultural Plot Geographic Information System) provided by the Spanish Ministry of Agriculture, Fisheries and Food. The results showed low F1-score values, indicating a poor representation of agricultural covers. The comparison also showed that ECOCLIMAP-SG overestimated the irrigated surface. Therefore, it was decided to improve the ECOCLIMAP-SG Land Cover Map and create a new map of actual irrigated areas over this basin.

For the improved cover map, SIGPAC information was used. Each agricultural plot was classified, assigning to the informed cultivated species its correspondent ECOCLIMAP-SG cover and replacing it in the original map. The actual irrigated areas map was elaborated combining SIGPAC information of plots prepared for irrigation with data from LAI increments computed for two Summer days (20/08/2017 and 10/08/2019) by LDAS-Monde (Albergel et al. 2017). LDAS-Monde is a tool based on SURFEX able to assimilate satellite-derived LAI from the Copernicus Global Land service in ISBA. The irrigation method map was generated using a simple approach. The irrigation districts were classified between traditional and modernized. In traditional areas, the irrigation method was set to surface irrigation. In modernized areas, plots with herbaceous crops and trees were assigned to sprinkler and drip irrigation respectively. 

In addition, a new version of the SAFRAN meteorological forcing (Vidal et al. 2010, Quintana-Seguí et al, 2017), was developed. For this project, observations from two different sources (Spanish Meteorological State Agency and Catalan Meteorological Survey) were obtained, together with ERA5 data, which is used as first guess (for all variables except Precipitation). SAFRAN has been used to generate forcing data for Precipitation, Temperature, Wind Speed and Relative Humidity.  

The forcing dataset sensitivity has been tested by comparing two SURFEX simulations performed using the new SAFRAN forcing dataset, one with the original ECOCLIMAP-SG maps and another one using the new land cover, actual irrigated areas, and irrigation methods maps.

This work is a contribution to the LIAISE campaign, through the IDEWA project (PCI2020-112043).