Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Hydrology and Remote Sensing Laboratory » Research » Publications at this Location » Publication #361656

Research Project: Integrating Remote Sensing, Measurements and Modeling for Multi-Scale Assessment of Water Availability, Use, and Quality in Agroecosystems

Location: Hydrology and Remote Sensing Laboratory

Title: Uncertainty of reference pixel soil moisture averages sampled at SMAP core validation sites

item CHEN, F. - Science Systems And Applications, Inc
item Crow, Wade
item Cosh, Michael
item COLLIANDER, A. - Jet Propulsion Laboratory
item ASANUMA, J. - University Of Tsukuba
item BERG, A. - University Of Guelph
item Bosch, David - Dave
item CALDWELL, T. - University Of Texas
item Holifield Collins, Chandra
item MARTINEZ-FERNANDEZ, J. - University Of Salamanca
item MCNAIRN, H. - University Of Salamanca
item Starks, Patrick
item SU, Z. - University Of Twente
item WALKER, J. - Monash University

Submitted to: Journal of Hydrometeorology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/6/2019
Publication Date: 7/29/2019
Citation: Chen, F., Crow, W.T., Cosh, M.H., Colliander, A., Asanuma, J., Berg, A., Bosch, D.D., Caldwell, T., Holifield Collins, C.D., Martinez-Fernandez, J., Mcnairn, H., Starks, P.J., Su, Z., Walker, J. 2019. Uncertainty of reference pixel soil moisture averages sampled at SMAP core validation sites. Journal of Hydrometeorology.

Interpretive Summary: Satellite-based soil moisture estimates can be used for a wide range of agricultural applications including: drought forecasting, yield monitoring and flood prediction. Recently, NASA has launched the Soil Moisture Active/Passive (SMAP) mission which represents a significant step forward in our ability to globally monitor water resources within agricultural regions. However, soil moisture products acquired from satellite-based sensors must first be validated via comparisons against ground-based observations. These comparisons can be challenging due to the severe lack of available ground observations in many areas of the world and the sharp contrast between the coarse-scale nature of satellite products (with resolutions >10 km) and the point-scale nature of ground-based soil moisture measurements. This paper derives error estimates for coarse-scale estimates of soil moisture derived from the aggregation of multiple, point-scale soil moisture observations within dedicated SMAP ground validation sites. Such uncertainty assessments are critical for efforts to credibly validate satellite-based SMAP soil moisture estimates using ground-based soil moisture networks. Once achieved, credible ground validation will aid in the promotion of SMAP soil moisture products for important agricultural applications.

Technical Abstract: Despite extensive efforts to maximize ground coverage and improve upscaling functions within core validation sites (CVS) established by the partners of the NASA Soil Moisture Active/Passive (SMAP) mission, spatial averages of point-scale in situ soil moisture observations can never perfectly capture the true spatial average of the reference pixels. Therefore, some level of pixel-scale sampling error from in situ soil moisture observations must be considered in the validation of SMAP soil moisture retrievals utilizing ground data. Using data collected from fifteen SMAP CVS, uncertainties due to spatial sampling errors in the core site average soil moisture (CSASM) are examined and the implications of these errors in CSASM-based SMAP calibration and validation metrics are discussed. The estimated uncertainty of the CSASM temporal average are found to be relatively large, translating into large uncertainty levels for SMAP bias when calculated against CSASM. As a result, the SMAP bias is found to be statistically insignificant at nearly all CVS. Observations from temporary networks suggest that these bias uncertainties may still be underestimated due to under-sampled spatial variability with the existing sensor networks. The unbiased root-mean-square error (ubRMSE) of CSASM – error after removal of temporal averages – are estimated via two distinct error analysis methods: classical sampling theory and triple collocation, both of which suggest that the CSASM ubRMSE is within the range of 0.01 to 0.02 m3/m3. Although limitations in both methods likely lead to underestimation of ubRMSE, the results suggest that the CSASM successfully captures the temporal dynamics of the footprint-scale soil moisture and is a reliable reference for ubRMSE calculations. Spatial sampling errors that widely present and cause large temporal average uncertainties in the SMAP CVS, however, need to be addressed in order to provide reliable CSASM-based bias estimates.