Submitted to: Meeting Abstract
Publication Type: Proceedings
Publication Acceptance Date: 1/7/2007
Publication Date: 3/26/2007
Citation: Jackson, T.J., Cosh, M.H. 2007. Soil moisture algorithm validation with ground based networks. In: Proceedings of the Electromagnetic Remote Sensing, March 26-30, 2007, Beijing, China. p. 507-510.
Technical Abstract: Validation satellite-based soil moisture algorithms and products is particularly challenging due to the disparity of scales of the two observation methods, conventional measurements of soil moisture are made at a point whereas satellite sensors provide an integrated area/volume value over a large area. It is further complicated by the multiple scales of variability of the soil moisture field related to geographic units defined by land cover, soils, and topography, and as the result of rainfall events and climate. Soil moisture products from satellite sensors have to be validated because the retrieval algorithms utilize formulations, parameters and ancillary data that have not been thoroughly developed and verified. In addition, due to the complexity in scaling points to large footprints it will always be impossible to translate first principle physics/electromagnetics to these levels. Historically, validation of a soil moisture product was not a part of passive microwave remote sensing satellite missions until the AMSR instruments were launched. Validation programs for the AMSR instruments on the NASA EOS Aqua and the JAXA ADEOS-II (AMSR-E) platforms are currently ongoing (Njoku et al. 2003), and include many of the elements that are necessary. In order to address the validation issues noted, as part of the AMSR-E validation activity, several networks of insitu soil moisture sensors were established using the infrastructure already in place as part of research watersheds. These networks, typically consisting of 15 or more points, provide estimates of the average soil moisture over watersheds and surrounding areas that approximate the size of the AMSR passive microwave footprint. This is performed on a continuous basis (reporting at least every hour minutes), partially in real time. The same soil moisture/temperature instrument was used at all sites and watersheds, which provides an estimate of the 0-5 cm soil moisture. Four watersheds in different vegetation/climate regions of the U.S. were selected. All instrumentation was installed prior to the launch of AMSR-E in 2002. There are now close to five years of observations available. Quality control of the data has included short-term field experiments at some of the watersheds to verify calibration and scaling. To date the data sets from Walnut Gulch, Arizona and Little Washita, Oklahoma have been analyzed. The soil moisture algorithms used by NASA and JAXA for their respective standard products were compared to the network observations, along with an alternative algorithm. The results indicated that each algorithm has different performance statistics that depend upon the site, which will be presented in detail. A positive outcome of the analysis is that it appears that the algorithms can perform within acceptable error bounds. The issues addressed here are common to both current and future satellite missions. Insitu observations are critical to validation, however, an integrated approach using in situ networks, field campaigns and comparison to other satellite products is desirable.