Assessing the potential of different satellite soil moisture products in landslide hazard assessment

Abstract

Abstract With the development of remote sensing technology, satellite-based soil moisture estimates become more and more available, and the potential of using satellite soil moisture products in landslide hazard assessment has been widely recognized. However, to our knowledge, there is a lack of studies exploring the performance difference of various satellite soil moisture products for such an application. Therefore, this study aims to compare several state-of-the-art satellite soil moisture products on their potentials in landslide applications. The selected products include the ESA CCI soil moisture dataset, the SMAP Level-3 (L3), enhanced Level-3 (L3), Level-4 (L4) surface, and Level-4 (L4) root zone soil moisture datasets. Specifically, the completeness of different datasets is calculated to assess their applicability in practical applications. To investigate the relationship between the soil moisture and the commonly used rainfall information in landslide predictions, the correlation study of the satellite soil moisture with the antecedent cumulated rainfall is also carried out. In addition, to explore whether the satellite soil moisture can provide valuable information for landslide hazard assessment, infiltration events are identified based on the time series of satellite soil moisture, and the significance of event characteristics (such as event duration, soil moisture change, etc.) in landslide occurrence is then investigated with Bayesian analysis. This study is carried out in a landslide-prone area, the Emilia-Romagna region in northern Italy. Results show that the SMAP L4 product does not have any missing values, beneficial to the continuous monitoring of landslides. As for the correlation relationship between soil moisture and antecedent cumulated rainfall, the SMAP L4 product also has more rational spatial distribution of the Pearson correlation coefficients compared with other datasets, which can be better explained by the distribution of slope and TWI (topographic wetness index). Bayesian analysis on the infiltration events shows that our prior knowledge of the probability of landslide occurrence is better improved by using the ‘SMAP L4 root zone soil moisture’-derived infiltration events, indicating its greater potential to be used for landslide hazard assessment in the study region.