Soil moisture estimates over sporadically flooded farmlands: synergies and biases of remote sensing and in situ sources

CAPPELLETTI, LUCÍA M.; SÖRENSSON, ANNA A.; SALVIA, MERCEDES; RUSCICA, ROMINA C.; SPENNEMMAN, PABLO; FERNANDEZ-LONG, MARIA E.; JOBBÁGY, ESTEBAN

INTERNATIONAL JOURNAL OF REMOTE SENSING

TAYLOR & FRANCIS LTD

Año: 2022 vol. 43 p. 6979 - 7001

0143-1161

Soil moisture plays a key role in hydrological processes in ecosystems and regulates water and energy exchanges between the surface and the atmosphere. Global coverage of surface soil moisture (SSM) satellite estimates makes them a fundamental source of information, while the validation of these estimates is usually based on in situ measurements, ideally from networks that cover areas similar to the satellite data resolution. However, as we expose in this study, both SSM data sources face challenges over extremely flat regions with large SSM variability. A homogenous farming region in the subhumid Pampas of Argentina, characterized by a large interannual rainfall variability as well as a marked annual cycle of rainfall and cropping, was taken as a case study. The region is almost devoid of irrigation and drainage infrastructure, is subject to large episodic flood and waterlogging events and holds an in situ network belonging to the Argentinean National Commission for Space Activities. This in situ network was set to evaluate the soil moisture estimated by satellite missions, such as SMAP and SAOCOM. However, several of these sites have been placed close to homesteads in a more uniform perennial vegetation than the prevailing seasonal crop. In this work, we examine how this placement bias influences SSM dynamics and its interpretation. We find that in situ data fails to capture the large seasonal and daily SSM variability caused by the cropping dynamics as well as the situation of waterlogging. As for the satellite SSM estimates, provided by the SMOS and SMAP missions in this study, while they capture the impact of cropping on SSM, data gaps can hinder robust statistical analysis. During periods of waterlogging, SSM values can lie outside the dynamic range considered valid by satellite missions, and thus are usually removed by users, creating ‘blind spots’ for high soil water content stages in flood-prone lands. Our study underlines the importance of using multiple sources of information to interpret the hydrological status, including data from in situ measurements and remote sensing estimations of SSM as well as, when available, locally collected information such as reports from national, sub-national and private agro-industrial agencies.