Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: Spatially Relevant Stochastic Weather Simulation Model Development for Biological and Hydrological Applications

Authors
item Johnson, Gregory
item Hanson, Clayton
item Lu, Yunyun - UNIVERSITY OF IDAHO
item Richardson, Clarence

Submitted to: Agricultural and Forest Meteorology Conference Proceedings
Publication Type: Proceedings
Publication Acceptance Date: September 26, 1995
Publication Date: N/A

Interpretive Summary: For many applications in the biological and hydrological sciences, particularly for computer model simulations of processes such as crop growth or erosion, it is necessary to use statistical climate models. Such models take advantage of basic climate information and the random nature of phenomena to simulate natural, day-to-day variability in weather. In many locations of interest, however, "basic" climate information is simply not available. In such cases it is often the practice to estimate these base-line values from nearby locations where quality weather records are available. However, in locations where climate is quite variable because of the presence of mountains, water bodies, and other natural features nearby locations may not be representative of the location in question. To circumvent this problem, and to enable the accurate simulation of long weather records for virtually any location, a procedure has been developed to map these basic parameters in any terrain. The procedure will result in the capability to simulate climate in any terrain. It is now being tested over a two- state region, and the results are presented in this paper.

Technical Abstract: For many applications in the biological and hydrological sciences, particularly for computer model simulations of processes such as crop growth or erosion, it is necessary to use stochastic weather models. Such models take advantage of basic climate information (such as means and variances) and the random nature of phenomena to simulate natural, day- to-day variability in weather. In many locations of interest, however, "basic" climate information is simply not available. In such cases it is often the practice to estimate these base-line values from nearby locations where quality weather records are available. However, as spatial complexity of climate increases simple interpolation procedures are unacceptable. To circumvent this problem, and to enable the accurate simulation of long time series of weather for virtually any location, regardless of topographic complexity, a method of spatially distributing the parameters of a first-order, Markov-chain-based stochastic weather model (delivering model output on a daily time step) is introduced. The USDA-ARS USCLIMATE weather generator is chosen, and the PRISM climatic interpolation model is used for stochastic parameter distribution. The methodology is described in terms of an application to a climatically-diverse region.

Last Modified: 12/26/2014
Footer Content Back to Top of Page