Assessment and Analysis of Crop Irrigation by the Combination of Modelling and Remote Sensing Techniques

Crop irrigation should be properly monitored to plan the use of land water resources and their repartition among competing activities. Appropriate statistical approaches can be applied to infer the irrigation water (IW) supplied over a cropped area relying on ground observations and the outputs of calibrated crop development models. The collection of such reference samples over relatively large areas and multiyear periods, however, is often hampered by practical problems that limit the possibility of obtaining precise estimates of the IW actually supplied. One of the possible ways to overcome these issues and increase the precision of the IW observations is through a regression correction versus wall-to-wall IW covariates obtained from remotely sensed images. Specifically, the correction of the reference IW observations can be performed using mapped IW estimates yielded by the combination of meteorological data and Sentinel-2 NDVI images. This strategy was tested in a 10 × 10 km² agricultural area in Southern Tuscany (Central Italy) during 2018–2022. The high correlations found between the reference and remotely sensed IW values allowed us to obtain satisfactory results for all years. The regression corrections applied had very high relative efficiencies (> 30) and notably enhanced the IW precisions obtained from the reference samples. The dynamics of the corrected IW observations were finally analysed versus the possible drivers, yielding useful indications for the management of local water resources.