In this work, surface soil moisture (SSM) datasets at different spatial resolutions (1 km and plot-scale), derived from Sentinel-1 observations, are used as input into the soil moisture (SM)-based inversion algorithm to retrieve information on irrigation volumes. The method is applied over an agricultural area of about 7000 ha falling within the Upper Tiber River valley (central Italy). For this area, information about irrigation water consumption in the period 2017–2020 is used as a benchmark. A district-scale analysis is carried out by comparing the performances of three different SSM datasets: two 1 km resolution products (Copernicus and RT1), and the S2MP, a plot-scale product developed by merging Sentinel-1 and Sentinel-2 observations. At the district level, the best performances are obtained through the Copernicus SSM, providing a median yearly relative error of 17.5%. RT1 SSM shows an overestimation lower than 30% compared to the actual irrigation volumes for two of the four considered irrigation seasons. The lowest performances are found for the S2MP dataset, with irrigation estimates much larger than the actual irrigation amounts. At the plot scale, overestimates (BIAS = 19.75 mm/14-day) and underestimates (BIAS = 14.88 mm/14-day) are obtained in the irrigation seasons of 2017 and 2018, respectively.
Quantifying Irrigation Volumes Using Sentinel-1 Soil Moisture Data in Central Italy
Vergni, L.
;Dari, J.;Todisco, F.;Vizzari, M.;Saltalippi, C.;Venturi, S.;Casadei, S.;
2023
Abstract
In this work, surface soil moisture (SSM) datasets at different spatial resolutions (1 km and plot-scale), derived from Sentinel-1 observations, are used as input into the soil moisture (SM)-based inversion algorithm to retrieve information on irrigation volumes. The method is applied over an agricultural area of about 7000 ha falling within the Upper Tiber River valley (central Italy). For this area, information about irrigation water consumption in the period 2017–2020 is used as a benchmark. A district-scale analysis is carried out by comparing the performances of three different SSM datasets: two 1 km resolution products (Copernicus and RT1), and the S2MP, a plot-scale product developed by merging Sentinel-1 and Sentinel-2 observations. At the district level, the best performances are obtained through the Copernicus SSM, providing a median yearly relative error of 17.5%. RT1 SSM shows an overestimation lower than 30% compared to the actual irrigation volumes for two of the four considered irrigation seasons. The lowest performances are found for the S2MP dataset, with irrigation estimates much larger than the actual irrigation amounts. At the plot scale, overestimates (BIAS = 19.75 mm/14-day) and underestimates (BIAS = 14.88 mm/14-day) are obtained in the irrigation seasons of 2017 and 2018, respectively.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
