Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Environmental Microbial & Food Safety Laboratory » Research » Publications at this Location » Publication #254782
Title: Data assimilation in optimizing and integrating soil and water quality water model predictions at different scales

Author
item Pachepsky, Yakov
item JACQUES, DIEDERIK - Belgian Nuclear Research
item Guber, Andrey
item PAN, FENG - University Of Maryland
item HILL, ROBERT - University Of Maryland
item Gish, Timothy
item VAN GENUCHTEN, MARTINUS - Universidade Federal Do Rio De Janeiro
item CADY, RALPH - Us Nuclear Regulatory Commission
item NICHOLSON, THOMAS - Us Nuclear Regulatory Commission

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 6/21/2010
Publication Date: 9/23/2010
Citation: Pachepsky, Y.A., Jacques, D., Guber, A.K., Pan, F., Hill, R., Gish, T.J., Van Genuchten, M., Cady, R., Nicholson, T. 2010. Data assimilation in optimizing and integrating soil and water quality water model predictions at different scales. [abstract]. International Workshop: Optimizing and integrating predictions of agricultural soil and water conservation models at different scales. p. 8.

Interpretive Summary:

Technical Abstract: Relevant data about subsurface water flow and solute transport at relatively large scales that are of interest to the public are inherently laborious and in most cases simply impossible to obtain. Upscaling in which fine-scale models and data are used to predict changes at the coarser scales is the norm in environmental science. Upscaling in general involves two operations – interpolation and simulation. The order in which the two operations are applied determines how the fine-scale models and data are used. This talk focuses on the case when simulation is used first such that the finer-scale simulations are applied across the coarse-scale domain. Since such simulations are affected by uncertainty in the model parameters, they are intrinsically imprecise. Data assimilation (DA) is a method of continually correcting the simulation results based on monitoring. A critical advantage of data assimilation is that model predictions may be corrected at simulation nodes outnumbering the observation points by several orders of magnitude. DA is for this reason very well suited for correcting fine-scale simulations used in the upscaling process. DA is commonly used in atmospheric and ocean sciences, and begins to attract attention also in hydrology. DA-based corrections are based on the magnitudes and correlations of data and model errors. The ensemble Kalman filter use is an efficient way to evaluate model errors, with pedotransfer functions providing a means to build the ensemble. Various features of data assimilation are illustrated using an example involving data obtained with soil moisture sensors.