Evaluating irrigation demand forecasts from S2S/agro-hydrological modelling with field experiments in Northern Germany in the context of farmer decision support

Optimizing water use efficiency and crop yield are important objectives of irrigated agriculture. For planning near future irrigation, farmers can rely on weather forecasts, which cover a time horizon of up to two weeks. This information is then used to make decisions about agricultural activities, including irrigation. However, a gap exists between weather forecasting and climate prediction, which poses challenges for decision-making in the medium-term crop season. The sub-seasonal to seasonal (S2S) range, spanning from two weeks to one season, bridges this gap. In this study we investigate if S2S forecasts combined with an agro-hydrological model can extend the time horizon of farmers’ decision decision-making compared to a traditional week-to-week schedule.

A case study was conducted for the Northern German Hamerstorf experimental field, which is operated by the Chamber of Agriculture of Lower Saxony to provide weekly consulting and decision support services for regional farmers in the fields of fertilisation and irrigation. Irrigation is triggered at 35% and 50% of available water capacity and the annual crop yield for these irrigation scenarios is evaluated. In this research a SWAP (soil-water-atmosphere-plant) model was calibrated and validated using observed field data from the experiments. The calibrated model was then coupled with the reforecast S2S ensemble dataset. To evaluate the performance of the S2S/agro-hydrological model, we used the ECMWF (European Centre for Medium-Range Weather Forecasts) S2S ensemble and simulated the future irrigation water demand for the next two, four and six weeks. Simulated crop yield, irrigation water demand and the results of auto-scheduling irrigation over the recent five irrigation seasons (2018-2022) were evaluated and compared with a reanalysis using observed climate and with the experimental field practise.

First results confirm that uncertainty increases with the lead time of the forecast, but a major aspect for irrigation planning is the start and end of dry periods. There, uncertainty is less compared to the uncertainty of future rain, which recommends further exploration of the value of S2S forecasts in agricultural decision support.