Evaluation of Copernicus and SMAP soil moisture products in an almond orchard in southeastern Spain

Soil moisture (SM) plays an important role in the interactions between the atmosphere and the land surface, and has been widely recognized as a key variable of the climate system. Over the last decades, several global satellite products have been generated to monitor SM at different spatial and temporal resolutions. To use these products it is important to validate them with in-situ observations. In this study, the performance of the Soil Water Index (SWI) and Surface Soil Moisture (SSM) Copernicus’s products and the Soil Moisture Active Passive (SMAP) SMAP L3_SM_P_E product was evaluated over an irrigated almond orchard located in the semiarid area of Tarazona de la Mancha, Spain (39.2660N, -1.9397W). The almond trees were planted in 2017 in a homogeneous field of about 10-ha. The Copernicus SSM and SWI products at 1-km spatial resolution provide daily SM images covering Europe since 2015. The SSM is retrieved from the Sentinel-1 radar backscattering and SWI combines Sentinel-1 and Metop ASCAT data. The Level-3 SMAP product provides SM data every 2-3 days retrieved by the SMAP radiometer. SMAP L3_SM_P_E has a spatial resolution of 9 km. Here, the moisture content in the topsoil (5 cm) estimated by the satellite products was evaluated against observed SM measurements for the 2019-2023 period. Even though the field sensor registers SM data at different depths (10-120 cm), the SM observations of the first 10 cm were used to analyze the remote sensing products. The accuracy of the products was defined using the following statistics; the determination coefficient (R²), the root mean square difference (RMSD), bias, and the unbiased root mean square difference (ubRMSD). The results obtained show that in general, the evaluated products capture the temporal variability of the SM measurements. For SSM and SMAP differences against in-situ data resulted in RMSD of about 4.34 vol% and 4.85 vol%, respectively. Also, SMAP overestimates the observed data with a considerable bias of 3.60 vol%. These deviations could be due to the coarse spatial resolution, however, it achieves the highest correlation (R²=0.64). SSM shows a good agreement with in-situ measurements, yielding the lowest bias (bias=0.10 vol%), but poorer R² than the other evaluated datasets (R²=0.22). The SWI product (RMSD=3.79 vol%, ubRMSD=3.73 vol%, R²=0.36, and bias=0.66 vol%) performs the best compared to SMAP and SSM. These results obtained in almonds are comparable to validation results published for other regions and land covers. Therefore, it is possible to indicate that satellite SM data and, specifically the SWI product could benefit the local water resources management.