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 #306427

Title: Stand-alone error characterisation of microwave satellite soil moisture using a Fourier method

item SU, CHUN-HSU - University Of Melbourne
item RYU, D. - University Of Melbourne
item Crow, Wade
item WESTERN, A. - University Of Melbourne

Submitted to: Remote Sensing of Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/1/2014
Publication Date: 11/14/2014
Publication URL:
Citation: Su, C., Ryu, D., Crow, W.T., Western, A. 2014. Stand-alone error characterisation of microwave satellite soil moisture using a Fourier method. Remote Sensing of Environment. 154:115-126. DOI: 10.1016/j.rse.2014.08.014.

Interpretive Summary: Within the past decade, substantial progress has been made in the remote sensing of surface soil moisture content using satellite-based sensors. Accurate monitoring of soil moisture conditions is important for a range of agricultural applications including: drought monitoring, fertilizer scheduling, and irrigation management. Satellite-based soil moisture estimates are increasingly being used in these important application. However, proper use of remotely-sensed estimates requires information concerning their accuracy and reliability. This paper describes the application of a new time series method to estimate the statistical magnitude of error in surface soil moisture estimates obtained from satellite remote sensing. This accuracy information is very valuable for any water resource or agricultural decision support system that is attempting to utilize uncertain satellite-derived soil moisture estimates to arrive at a particular management decision.

Technical Abstract: Error characterisation of satellite-retrieved soil moisture (SM) is crucial for maximizing their utility in research and applications in hydro-meteorology and climatology. Error characteristics can provide insights for retrieval development and validation, and inform suitable strategies for data fusion and assimilation. Error estimation is typically implemented in the time domain through direct comparison against ground observations or triple collocation (TC) with multiple coincident data sets. The frequency (Fourier) domain can provide an alternative framework for examining erroneous SM records. In particular the analysis of the difference between the empirical power spectra of these data and a water balance may be a useful means to quantify the errors and circumvent the need for other data sets. This work tests the utility of a Fourier method proposed by Su et al. (2013a) by applying to two passive and active microwave satellite SM products from AMSR-E (Advanced Microwave Scanning Radiometer – Earth Observing System) and ASCAT (Advanced Scatterometer of MetOp-A). Their errors are estimated over Australia and compared against the TC estimator. The Fourier estimator shows very good agreement with TC, with strong linear correlations of 0.80–0.92 in terms of error variance and (normalised) fractional error, but negative bias of order 0.01 m^3m^-3 is also noted. Their differences are related to the power spectral estimation and model inversion methods of SF that do not fully account for the variances of all frequency components, amongst other reasons. Spatial analyses of the derived (SF and TC) error maps for the two satellite products are performed using principal component analysis to discern influence of higher vegetation, rainfall and spatial heterogeneity in topography and soil type on increased retrieval errors. Seasonal analysis of the errors discovers systematic temporal variability in errors due to variability in rainfall amount, and less so with changing vegetation density.