Soil Moisture Drydown Detection Is Hindered by Model-Based Rescaling

• Published in: IEEE geoscience and remote sensing letters Vol. 19; pp. 1 - 5
• Main Authors: Raoult, Nina, Ruscica, Romina C., Salvia, M. Mercedes, Sorensson, Anna A.
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE); IEEE - Institute of Electrical and Electronics Engineers
• Subjects: Algorithms; Climate; Continental interfaces, environment; Cumulative density function (CDF) matching; Datasets; Detection; drydown detection; Drying; drying rates; Dynamic range; European Space Agency Climate Change Initiative (ESA-CCI) products; Heavy rainfall; Land surface; Merging; Methods; Moisture; Moisture effects; Ocean, Atmosphere; precipitation; Rainfall; Rescaling; Satellite data; Satellites; Scaling; Sciences of the Universe; Sensors; Soil; Soil moisture; soil moisture (SM); Time
• ISSN: 1545-598X; 1558-0571
• DOI: 10.1109/LGRS.2022.3178685

The rate at which soils dry out after heavy rain has huge impact on the climate. It is important that these drying rates, known as soil moisture (SM) drydowns, are well represented in climate simulations. Satellite data allow us to study these events globally. Although there are many individual satellite sensors we can use, their data have gaps, both spatially and temporally. Merged products, such as the ones from the European Space Agency, collate data from these sensors to create datasets that are as complete as possible. However, we find that the merging algorithms used to create such products can hinder the detection of drydowns events calling for caution when using such datasets. The smaller SM dynamic range imposed on this combined dataset during its creation hinders drydown detection when using methods based solely on SM dynamics. Although fewer drying events are detected, the drydown time scales are mostly unchanged. Detection methods using external precipitation products are less affected by this rescaling, and however, we detect far fewer events and drydown time scales tend to be longer than when using SM-based methods.