Jalilvand, Ehsan; Tajrishy, Masoud; Ghazi Zadeh Hashemi, Sedigheh Alsadat; Brocca, Luca, 2019, Quantification of irrigation water using remote sensing of soil moisture in a semi-arid region, Remote sensing of environment 231 (2019): 111226. doi_10.1016/j.rse.2019.111226,
URL: http://www.cnr.it/prodotto/i/405572

Irrigated agriculture is the principal consumer of fresh water resources. Most countries do not have a precise measurement of water consumption for irrigation. In this study, an innovative approach is proposed that allows for estimation of irrigation water use at the catchment scale based on satellite soil moisture data. To this end, the SM2RAIN algorithm, which had been originally developed for estimation of rainfall from the soil moisture observations, is adopted. The satellite soil moisture observations obtained from Advanced Microwave Scanning Radiometer 2 (AMSR2) along with different rainfall and evapotranspiration (ET) products in the period 2012-2015 are used as the input to the model. The methodology is tested in the agricultural plains of southern Urmia Lake, which is one of the main agricultural plains in Iran for which actual irrigation data is available. The results reveal that the proposed approach can capture the overall irrigation pattern, although; it is systematically overestimating irrigation volume compared to observed irrigation data. Thus the bias is calculated over largely non-irrigated pixels and used to modify the model estimates. The bias-corrected results show good agreement with the in situ irrigation data. In particular, the average model performance in the irrigated pixels in terms of R and RMSE (mm/month) are (0.86 and 12.895) respectively. Accuracy varied depending on the inputs, with improvement in order of 11% and 42% in R and RMSE depending on the inputs chosen. The method is also applied to less irrigated areas that result in obtaining significantly lower irrigation rates. The low spatial resolution of soil moisture products (i.e. ~50 km) makes it difficult to capture the irrigation water of small irrigated croplands. Unreliable rainfall and ET data can also lead to the over/underestimation of irrigation. In spite of the above limitations (particularly lack of reliable ET dataset), the proposed model can still capture the irrigation pattern, given that strong soil moisture signal from irrigation is detected by the satellite.

Data from https://intranet.cnr.it/people/