Statistical assessment of irrigation water over wide areas by the use of ground and SENTINEL-2 data

Water is an essential resource for every-day life and its availability is a fundamental pre-requisite for the sustainable development of earth population. This is particularly the case for semi-arid regions, which, as a consequence of climate change, are also affected by a reduction of water availability and an increase of air temperature. In these areas, a large amount of water is required also for agricultural purposes and, in order to promote a sustainable use of this natural resource, it is necessary to quantify this amount and define its impact on the water cycle of a certain area. Therefore, our investigation focuses on the need to quantify the total amount of irrigation water (IW) required over a growing season, in a specific area where annual crops are cultivated both in rainfed and irrigated conditions. Statistical approaches might be applied to infer IW measured at some fields to the whole investigated area. However, this approach implies the availability of a large number of field measurements which however might not be sufficient to fully characterize the area. There are in fact problems in defining the size of the sample, in identifying IW for some crops, etc.. One of the possible ways to overcome these issues and increase the precision of the ground sampling is by applying the regression (or ratio) estimator to independent datasets informative on IW. Specifically, this method can be applied to the IW maps which have been recently produced using meteorological data and Sentinel-2 MSI NDVI imagery. An example of this approach is provided in the Grosseto area (10x10 km2) for the year 2018.