Irrigation timing retrieval at the plot scale using Surface Soil Moisture derived from Sentinel time series in Europe

The computation of the water budget of irrigated fields is generally difficult because of unknown irrigation amounts and timing. Automatic detection of irrigation events could greatly simplify the water balance of irrigated fields. The combination of high spatial resolution and high-frequency SAR (Sentinel-1) and optical satellite observations (Sentinel-2) makes the detection of irrigation events potentially feasible. Indeed, optical observation allows following the crop development while SAR observation can provide an estimation of the Surface Soil Moisture (SSM). However, uncertainties due to acquisition configuration or crop geometry and density might affect the retrieval of SSM. Here, an algorithm for irrigation events detection is assessed considering several aspects that could affect SSM retrieval (incidence angle, crop type, crop development) and specific characteristics of irrigation events (irrigation frequency, frequency of observations). Additionally, an alternative soil water budget model, the force-restore approach, is compared with the original bucket soil water budget algorithm. A European dataset of irrigation events collected during the ESA Irrigation+ project (5 sites in France, Germany, and Italy over three years) is used. The performances are analyzed in terms of the F‑score and the seasonal sum of irrigation. Overall, the analysis corroborated that the scores decrease with SSM observation frequency. The impact of the Sentinel-1 configuration (ascending/descending, close to 39°/far from 39°) on the retrieval results is low. The lower scores obtained with small NDVI compared to large NDVI were almost systematic, which is counter-intuitive, but might have been due to the larger number of irrigation events during high vegetation periods. The scores decreased as irrigation frequency increased, which was substantiated by the fact that the scores were better in France (more sprinkler irrigation) than in Germany (more localized irrigation). The strategy of merging different runs versus the strategy of interpolating all SSM data for one run has produced comparable results. The estimated cumulative sum of irrigation was around -20% lower compared to the reference dataset in the best cases. Finally, the replacement of the original SSM model by the Force-restore provided an improvement of about 6% on the F‑score, and also narrowed the error on cumulative seasonal irrigation. This study opens new perspectives for the advancement of irrigation retrieval at large scale based on SSM data sets through an in-depth analysis of results as a function of satellite configuration, irrigation techniques, and crops.