A Distributed Cotton Growth Model Developed from GOSSYM and Its Parameter Determination
- Xin-Zhong Liang *ab,
- Min Xub,
- Wei Gaoc,
- K. Raja Reddyd,
- Kenneth Kunkele,
- Daniel L. Schmoldtf and
- Arthur N. Samel
- a Dep. of Atmospheric and Oceanic Science, Univ. of Maryland, College Park, MD 20740, and Dep. of Atmospheric Sciences, Univ. of Illinois, Urbana, IL 61801
b Earth System Science Interdisciplinary Center, Univ. of Maryland, College Park, MD 20740, and Division of Illinois State Water Survey, Institute of Natural Resource Sustainability, Univ. of Illinois, Champaign, IL 61820
c USDA UV-B Monitoring and Research Program, Natural Resource Ecology Lab., and Dep. of Ecosystem Science and Sustainability, Colorado State Univ., Fort Collins, CO 80523
d Dep. of Plant and Soil Sciences, Mississippi State Univ., Mississippi State, MS 39762
e Cooperative Institute for Climate and Satellites, North Carolina State Univ., Asheville, NC 28801, and National Oceanic and Atmospheric Administration, National Climatic Data Center, Asheville, NC 28801
f USDA National Institute of Food and Agriculture, Washington, DC 20024; and
Prediction of cotton (Gossypium hirsutum L.) production under a changing climate requires a coupled modeling system that represents climate–cotton interactions. The existing cotton growth model GOSSYM has drawbacks that prohibit its effective coupling with climate models. We developed a geographically distributed cotton growth model from the original GOSSYM and optimized it for coupling with the regional Climate–Weather Research Forecasting model (CWRF). This included software redesign, physics improvement, and parameter specification for consistent coupling of CWRF and GOSSYM. Through incorporation of the best available physical representations and observational estimates, the long list of inputs in the original GOSSYM was reduced to two parameters, the initial NO3 amount in the top 2 m of soil and the ratio of irrigated water amount to potential evapotranspiration. The geographic distributions of these two parameters are determined by optimization that minimizes model errors in simulating cotton yields. The result shows that the redeveloped GOSSYM realistically reproduces the geographic distribution of mean cotton yields in 30-km grids, within ±10% of observations across most of the U.S. Cotton Belt, whereas the original GOSSYM overestimated yields by 27 to 135% at the state level and 92% overall. Both models produced interannual yield variability with comparable magnitude; however, the temporal correspondence between modeled and observed interannual anomalies was much more realistic in the redeveloped than the original GOSSYM because significant (P = 0.05) correlations were identified in 87 and 40% of harvest grids, respectively. The redeveloped GOSSYM provides a starting point for additional improvements and applications of the coupled CWRF–GOSSYM system to study climate–cotton interactions.Please view the pdf by using the Full Text (PDF) link under 'View' to the left.
Copyright © 2012. . Copyright © 2012 by the American Society of Agronomy, Inc.