GCOM-W AMSR2 soil moisture product validation using core validation sites

Rajat Bindlish* (Corresponding Author), Michael H. Cosh, Thomas J. Jackson, Toshio Koike, Hideyuki Fujii, Steven K. Chan, Jun Asanuma, Aaron A. Berg, David D. Bosch, Todd G. Caldwell, Chandra Holifield Collins, Heather McNairn, Jose Martinez-Fernandez, John H. Prueger, Tracy Rowlandson, Mark Seyfried, Patrick J. Starks, Marc Thibeault, R. van der Velde, Jeffrey P. WalkerEvan J. Coopersmith

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

46 Citations (Scopus)
13 Downloads (Pure)


The Advanced microwave scanning radiometer 2 (AMSR2) is part of the global change observationmission-water (GCOM-W). AMSR2 has filled the gap in passive microwave observations left by the loss of theAMSR-earth observing system (AMSR-E) after almost ten years of observations. Both missions provide brightness temperature observations that are used to retrieve soil moisture estimates at the near surface. A merged AMSR-E and AMSR2 data product will help build a consistent long-term dataset; however, before this can be done, it is necessary to conduct a thorough validation and assessment of the AMSR2 soil moisture products. This study focuses on the validation of the AMSR2 soil moisture products by comparison with in situ reference data from a set of core validation sites around the world. A total of three soil moisture products that rely on different algorithms were evaluated; the Japan Aerospace Exploration Agency (JAXA) soil moisture algorithm, the land parameter retrieval model (LPRM), and the single channel algorithm (SCA). JAXA, SCA, and LPRM soil moisture estimates capture the overall climatological features. The spatial features of the three products have similar overall spatial structure. The JAXA soil moisture product shows a lower dynamic range in the retrieved soil moisture with a satisfactory performance matrix when compared to in situ observations [unbiased root mean square error (ubRMSE) = 0.059 m3/m3, Bias = -0.083 m3/m3, R = 0.465]. The SCA performs well over low and moderately vegetated areas (ubRMSE = 0.053 m3/m3, Bias = -0.039 m3/m3, R = 0.549). The LPRM product has a large dynamic range compared to in situ observations with a wet bias (ubRMSE = 0.094 m3/m3, Bias = 0.091 m3/m3, R = 0.577). Some of the error is due to the difference in observation depth between the in situ sensors (5 cm) and satellite estimates (1 cm). Results indicate that overall the JAXA and SCA have the best performance based upon the metrics considered.

Original languageEnglish
Article number8233417
Pages (from-to)209-219
Number of pages11
JournalIEEE Journal of selected topics in applied earth observations and remote sensing
Issue number1
Early online date20 Dec 2017
Publication statusPublished - 1 Jan 2018


  • In situ networks
  • Passive microwave
  • Soil moisture
  • Validation


Dive into the research topics of 'GCOM-W AMSR2 soil moisture product validation using core validation sites'. Together they form a unique fingerprint.

Cite this