MODEL PARAMETERIZATION AT DIFFERENT SCALES: HOW DO WE ESTIMATE AND INCORPORATE SPATIAL INFORMATION?

Authors: Green, Timothy; Ascough Ii, James; Ahuja, Lajpat; Ma, Liwang; Erskine, Robert - Rob

Submitted to: Workshop Proceedings
Publication Type: Abstract Only
Publication Acceptance Date: 4/2/2004
Publication Date: 4/20/2004
Citation: Green, T.R., Ascough Ii, J.C., Ahuja, L.R., Ma, L., Erskine, R.H. 2004. Model parameterization at different scales: how do we estimate and incorporate spatial information?. Intl. Workshop on Applications, Enhancements and Collaboration of ARS RZWQM and GPFARM Models, April 20-22, 2004, Fort Collins, CO.

Interpretive Summary: Models and model spatial units are applied at different scales without explicitly adjusting parameter values. Thus, the appropriate scale of application for a given model and its default parameters are often vague at best. Scale-dependence in model parameters usually results from heterogeneity in the underlying properties below the scale of application. It is important to recognize process and dimensionality changes with scale. Different processes may become dominant at different scales. If the processes and spatial/temporal distributions of parameter values are known, one can derive the scaling behavior of these parameters, i.e., effective or apparent parameter values for different scales. For unsaturated media, the effective parameters have been shown to be both scale-dependent and state-dependent. Such phenomena add complexity to both parameterization and model formulation. Thus, parameter estimation should not be viewed independently of the model structure, and detailed knowledge of the active processes and sub-scale patterns of heterogeneity may be required to rigorously address the issues of parameterization and scale effects. Methods of estimating spatial data and incorporating such data into models will be discussed. In agricultural systems modeling, the estimation problem is compounded by interactions in space and time between various system components. Opportunities for future research will be presented.

Technical Abstract: Models and model spatial units are applied at different scales without explicitly adjusting parameter values. Thus, the appropriate scale of application for a given model and its default parameters are often vague at best. Scale-dependence in model parameters usually results from heterogeneity in the underlying properties below the scale of application. It is important to recognize process and dimensionality changes with scale. Different processes may become dominant at different scales. If the processes and spatial/temporal distributions of parameter values are known, one can derive the scaling behavior of these parameters, i.e., effective or apparent parameter values for different scales. For unsaturated media, the effective parameters have been shown to be both scale-dependent and state-dependent. Such phenomena add complexity to both parameterization and model formulation. Thus, parameter estimation should not be viewed independently of the model structure, and detailed knowledge of the active processes and sub-scale patterns of heterogeneity may be required to rigorously address the issues of parameterization and scale effects. Methods of estimating spatial data and incorporating such data into models will be discussed. In agricultural systems modeling, the estimation problem is compounded by interactions in space and time between various system components. Opportunities for future research will be presented.