Software Tools for Weed Seed Germination Modeling

Abstract

The next generation of weed seed germination models will need to account for variable soil microclimate conditions. To predict this microclimate environment we have developed a suite of individual tools (models) that can be used in conjunction with the next generation of weed seed germination models. The three tools that will be outlined here are GlobalTempSIM, GlobalRainSIM, and the soil temperature and moisture model (STM2). Each model was compared with several sets of observed data from worldwide locations. Overall, the climate predictors compared favorably. GlobalTempSIM had a bias between −2.7 and +0.9 C, mean absolute errors between 1.9 and 5.0 C, and an overall Willmott d-index of 0.79 to 0.95 (where d = 1 represents total agreement between observed and modeled data) for 12 global validation sites in 2007. GlobalRainSIM had a bias for cumulative precipitation ranging from −210 to +305 mm, a mean absolute error between 29 and 311 mm, and a corresponding d-index of 0.78 to 0.99 for the sites and years compared. The high d-indices indicate that the models adequately captured the annual patterns for the validation sites. STM2 also performed well in comparisons with actual soil temperatures with a range of −2 to +4.6 C biases and mean absolute errors between 0.7 and 6.8 C, with the d-index ranging from 0.83 to 0.99 for the soil temperature comparisons. The soil moisture prediction annual bias was between −0.09 and +0.12 cm3 cm−3, mean absolute errors ranging from 0.02 to 0.16 cm3 cm−3, and possessed a d-index between 0.32 and 0.91 for the validation sites. These models were developed in JAVA, are simple to use, operate on multiple platforms (e.g., Mac, personal computer, Sun), and are freely available for download from the U.S. Department of Agriculture Agricultural Research Service website (http://www.ars.usda.gov/Services/docs.htm?docid=11787).