A comprehensive assessment of SM2RAIN-NWF using ASCAT and a combination of ASCAT and SMAP soil moisture products for rainfall estimation

Abstract

Rainfall estimation using remote sensing products is an alternative to in situ measurement rainfall due to their high temporal and spatial resolution. Using satellite soil moisture (SM) observations in the SM to Rain (SM2RAIN) algorithm have a great potential to estimate rainfall. SMA2RAIN-NWF algorithm is a reinforced version of a SMA2RAIN algorithm which was developed to estimate rainfall through the integration of the SM2RAIN algorithm and the net water flux (NWF) model. A new release of SMA2RAIN-NWF algorithm uses the Advanced Microwave Scanning Radiometer 2 (AMSR2) SM dataset as input datasets. The aim here is to assess the SMA2RAIN-NWF by using multiple SM products including ASCAT, and their integration in four aggregations (AGGR) periods (1, 7, 14, and 30 days) by comparing with rainfall observation of 15 stations over the Lake Urmia basin, Iran for the period January 2015 to December 2019. The Discrete Cosine Transform (DCT) method is applied to fill the gap in the satellite SM time series. Moreover, the effect of land cover classes (grasslands, croplands, and urban) on rainfall estimation is investigated. Considering the Kling-Gupta efficiency (KGE) and correlation coefficient (R) values in comparisons of calibration and validation revealed that urban areas experienced a minimum decrement rate (2–5 %). A comparison of three SM products (ASCAT, ASCAT+SMAP, and ASCAT+DCT) show that all products had a high performance on a daily time scale in term of the KGE and R. The results showed that algorithm performance gradually rose via an increase in AGGR levels, reaching KGE and R values of 0.8 and above. Furthermore, the comparison of SM2RAIN-NWF and SM2RAIN show an improvement of SM2RAIN-NWF performance across various AGGRs. © 2022 Elsevier B.V.