Location: Hydrology and Remote Sensing LaboratoryTitle: An improved triple collocation algorithm for decomposing autocorrelated and white soil moisture retrieval errors
|Dong, J. - US Department Of Agriculture (USDA)|
Submitted to: Journal of Geophysical Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/1/2017
Publication Date: 11/17/2017
Citation: Dong, J., Crow, W.T. 2017. An improved triple collocation algorithm for decomposing autocorrelated and white soil moisture retrieval errors. Journal of Geophysical Research. 122(24):13,081-13,094. https://doi.org/10.1002/2017JD027397.
Interpretive Summary: Remotely-sensed surface soil moisture estimates have the potential to aid in a variety of important agricultural and climate applications. However, before this potential can be fully realized, we need to better understand the amount of error these estimates contain. This paper develops and applies a new mathematical technique for deriving a more complete statistical description of errors in existing remotely-sensed surface soil moisture products. The new information this technique provides will eventually be used to develop improved future versions of these products and enhance their usefulness for important water resource and hazard applications including agricultural drought monitoring and flood forecasting. For example, an improved understanding of uncertainty in surface soil moisture estimates will help analysts at the USDA Foreign Agricultural Service (FAS) understand how much consideration should be given to remotely-sensed surface soil moisture products (versus other drought products) when attempting to assess the role of soil water limitations on agricultural productivity in drought-prone regions.
Technical Abstract: If not properly account for, auto-correlated errors in observations can lead to inaccurate results in soil moisture data analysis and reanalysis. Here, we propose a more generalized form of the triple collocation algorithm (GTC) capable of decomposing the total error variance of remotely-sensed surface soil moisture retrievals into their auto-correlated and the white components. Synthetic tests demonstrate the robustness and accuracy of the GTC algorithm even in the presence of significant data gaps. The accuracy of GTC error autoregressive parameter estimates is relatively more sensitive to temporal data availability. In addition, land surface model (LSM) soil moisture predictions collected from phase 2 of the North American Land Data Assimilation System (NLDAS-2) and remotely-sensed surface soil moisture retrievals obtained from the European Space Agency Climate Change Initiative (ESA CCI) soil moisture products are applied for a real data demonstration. ESA CCI-Act (based on scatterometer data) demonstrates largest autoregressive parameters over low LAI areas. Conversely, ESA CCI-Pas (based on radiometer data) has larger error autoregressive parameters over large LAI areas. This suggests that decomposing the errors and considering these colored errors may be particularly necessary under these conditions.