Evaluation of agro-hydrological subseasonal-to-seasonal forecasts for tactical decision making in crop irrigation

The timing of many agricultural activities depends on weather. Farmers typically rely on short term weather forecasts for operational decision making. Climate forecasts with longer lead times may offer additional guidance, particularly in dry seasons, when the irrigation water allowance might not be sufficient. Farmers may want to make tactical decisions to anticipate a possible shortage of water before end of season. The main objective of this paper is to investigate the use of sub-seasonal to seasonal (S2S) forecasts with lead times of up to six weeks for estimating future irrigation. The case study site is located in Hamerstorf, Northern Germany. Soil-Water-Atmosphere-Plant (SWAP) models of winter wheat and sugar beet were set up to derive auto-irrigation rules from observed scheduling. These models were applied to schedule future irrigation driven by ECMWF S2S ensemble climate forecasts. The performance of irrigation estimation up to six weeks in advance was tested, with forecasts updated weekly. Results show a decrease in irrigation forecast skill from the first- to the sixth-week lead time. The mean absolute error (MAE) of cumulative irrigation within one crop season using first-week forecasts is about 40 mm (≈2 doses) for winter wheat and 30 mm (≈1 dose) for sugar beet, increasing to 70 mm (≈3 doses) and 80 mm (≈3 doses), respectively, with a six-week S2S lead time. Although the results indicate a loss in forecast accuracy with longer lead times, on average, applying median of S2S forecasts has potential to increase crop water productivity (WPc) in irrigated years.