Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: USING REMOTE SENSING & MODELING FOR EVALUATING HYDROLOGIC FLUXES, STATES, & CONSTITUENT TRANSPORT PROCESSES WITHIN AGRICULTURAL LANDSCAPES Title: The optimality of potential rescaling approaches in land data assimilation

Authors
item Yilmaz, Mustafa
item Crow, Wade

Submitted to: Journal of Hydrometeorology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: September 28, 2012
Publication Date: April 1, 2013
Citation: Yilmaz, M.T., Crow, W.T. 2013. The optimality of potential rescaling approaches in land data assimilation. Journal of Hydrometeorology. 14:650-660.

Interpretive Summary: Estimating soil moisture using satellite-based observations and hydrological models is critical for agricultural drought monitoring. Modeled and remotely-sensed estimates are commonly merged together to maximize the accuracy of soil moisture assessments. Such merging, however, requires that systematic differences between soil moisture time series estimates (e.g., different means and contrasting dynamic ranges) be corrected before the estimates are combined. In this study, we evaluate the optimality of commonly-used observation-model matching techniques and offer a solution that yields the most accurate merged soil moisture estimate possible. The results of this study can be used to maximize the value extracted from current soil moisture modeling and observations resources and - therefore - the accuracy of existing agricultural drought monitoring systems.

Technical Abstract: It is well-known that systematic differences exist between modeled and observed realizations of hydrological variables like soil moisture. Prior to data assimilation, these differences must be removed in order to obtain an optimal analysis. A number of rescaling approaches have been proposed for removing systematic dfferences between models and observations. These methods include rescaling techniques based on: matching sampled temporal statistics (i.e. variance), minimizing the least-squares distance between observations and models, and the application of triple collocation. Here we evaluate the optimality and relative performances of these rescaling methods both analytically and numerically and find that a triple collocation-based rescaling method results in an optimal solution whereas variance matching- and least squares-regression approaches result in only approximations to this optimal solution.

Last Modified: 9/21/2014
Footer Content Back to Top of Page