U015-08
Error Characteristic Assessments of Soil Moisture Estimates from Satellites and Land Surface Models : Focusing on Forested and Irrigated Regions
Friday, 11 December 2020: 17:56
Hyunglok Kim1,2, Jean-Pierre Wigneron3, Sujay V Kumar4, Jianzhi Dong5, Wolfgang Wagner6, Michael H Cosh7, David D Bosch8, Chandra Holifield Collins9, Patrick j Starks10, Mark S Seyfried11 and Venkataraman (Venkat) Lakshmi1, (1)University of Virginia, Engineering Systems and Environment, Charlottesville, VA, United States, (2)University of Virginia, School of Data Science, Charlottesville, VA, United States, (3)ISPA, INRA, Bordeaux, France, (4)NASA Goddard Space Flight Center, Greenbelt, MD, United States, (5)USDA ARS, Hydrology and Remote Sensing Laboratory, Beltsville, MD, United States, (6)Vienna University of Technolog, Vienna, Austria, (7)U. S. Dept. of Agriculture, Beltsville, MD, United States, (8)USDA, ARS, Tifton, GA, United States, (9)Agricultural Research Service Tucson, Tucson, AZ, United States, (10)USDA-ARS, El Reno, OK, United States, (11)USDA-ARS, Northwest Watershed Research Center, Boise, ID, United States
Abstract:
The moisture in the topsoil layer (~10 cm) is one of the most important hydrological variables to understanding the water cycle and climate systems. This is because the partitioning of outgoing energy flux into latent and sensible heat fluxes can be described by the variability of surface soil moisture (SSM). Over the past four decades, global-scale SSM data have been estimated by several satellite systems and land surface models. As these data have different levels of accuracy, it is important for researchers to understand the error characteristics of various SSM products before harnessing them for any scientific or practical applications.
In this study, we focused mainly on investigating the error characteristics of SSM taken from four satellites and land surface model (LSM) data sets over forested and irrigated areas as these areas have been regarded as challenging to estimate SSM. We used triple collocation analysis (TCA) and ground-based SSM measurements to calculate the error statistics of each product from the most widely used satellites and LSM data: The Advanced Scatterometer (ASCAT), the Soil Moisture and Ocean Salinity (SMOS), the Advanced Microwave Scanning Radiometer 2 (AMSR2), the Soil Moisture Active Passive (SMAP), the European Centre for Medium-Range Weather Forecasts Reanalysis 5 (ERA5), and the Global Land Data Assimilation System (GLDAS). In the present study, we proposed to consider the standard deviations of all possible TCA-based numbers that could be derived from combinations of 6 different products to produce robust results.
Thorough this research, we showed the advantages and disadvantages of current satellite and model-based SSM products focused on forest and irrigated areas, and we found that the choice of triplets for TCA can have a dramatic impact on the final results.
