Climate Services for hidrology and agriculture: co-design and operation of services based on seasonal forecasting

Current context of climate change is impacting many sector’s activities: rising temperatures or the increased frequency of extreme events is affecting water or energy demands, or the way and schedule for crop growing. Climate services based on seasonal forecasts can provide a useful tool helping to adapt current practiques to a new scenario.

However, and with the exception of certain “windows of opportunity”, skill of seasonal forecasts is in general limited within mid-latitudes, so the development of a service needs a careful selection of the sources of information, postprocessing and calibration. Here, we present two examples of services for water and agriculture sector making use of seasonal forecasts, both co-developed with users:

The first one has been developed jointly with water authorities in Spain, providing probabilistic forecasts of both high resolution precipitation and inflow for the majority of Spanish reservoirs during the 5 month period spanning from November to March.

The service makes use of different sources of information, including different runs from ECMWF-SEAS5 and empirical forecasts based on snow accumulation. The close relationship between accumulated winter precipitation and North Atlantic Oscillation over Iberia is used to calibrate and weight the ensemble based on a “Best NAO” hybrid forecast, improving the skill. Additionally, a downscaling methodology combining analogs and regression is applied, providing improved skill over a wider area.

The information is then feeded in a web visor available through AEMET web, developed jointly with users. Every year, information for the upcoming winter and new version of the web application is presented, and discussion on results, usability and possible improvements take place, keeping the service updated to user needs.

The second one provides crop yield forecasts for raind fed winter cereals (barley and wheat) over Castille and Leon. A vegetal model is run for every grid point every day of the season using observations. A particular day of the season, the model is coupled using medium range forecasts and then using both an ensemble based on climatology and ECMWF-SEAS5 downscaled forecasts, and results are compared. A set of synthetic observations, coupling vegetal model to ERA5, is produced and validated, in order to evaluate the performance. Information is then provided to users through the stakeholders in the region.