Skip to main content
ARS Home » Research » Publications at this Location » Publication #145439


item Zhang, Xunchang

Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/11/2003
Publication Date: 8/1/2004
Citation: Zhang, X.J. 2004. CLIGEN non-precipitation parameters and their impact on WEPP crop simulation. Applied Engineering in Agriculture. 20(4):447-454.

Interpretive Summary: To make optimal management decisions, knowledge of potential impact of possible future climates on crop and forage production is needed. Computer programs such as climate generators and crop growth models are the most suitable tools for making such impact assessment. The objectives of this work were (i) to evaluate the ability of a climate generator (called CLIGEN) to generate or simulate daily maximum and minimum temperatures, solar radiation, and wind velocity, and (ii) to evaluate the potential impact of generated climates on winter wheat production using a hydrological and crop growth model named WEPP. Four Oklahoma weather stations were used in this study. Measured and CLIGEN-generated climates were compared for each site. The WEPP model, which was calibrated at El Reno, Oklahoma, was run for all sites. Results indicate that CLIGEN generated each of the non-precipitation variables including maximum temperature, minimum temperature, and solar radiation reasonably well. But cross-correlations between precipitation, temperatures, and solar radiation were not properly simulated. As a result, wheat yields predicted using generated climates differed considerably with those predicted with measured climates. However, the yield distribution or average yield trend was well preserved. This suggests average yield trend, rather than yields of individual years should be used in the impact assessment. The overall results show that CLIGEN and WEPP models are appropriate tools to use for assessing climatic impact on crop production. This work provides useful information to researchers and extension professionals for making impact assessment using these models.

Technical Abstract: Physically based response models are the most suitable tools for assessing potential climatic impacts on water resources and crop production. Most response models require daily weather data, which are often synthesized using stochastic daily weather generators. The objectives in this study were to evaluate the ability of the CLImate GENerator (CLIGEN) model to generate non-precipitation parameters, including daily temperatures, solar radiation, and wind velocity at four Oklahoma stations, and to assess potential impact of the generated parameters on simulated productivity of a winter wheat crop (Triticum aestivum L.) using the Water Erosion Prediction Project (WEPP) model. Weather data measured by the National Weather Service and Oklahoma Mesonet at four stations with mean annual precipitation ranging from 420 to 1150 mm were used. Measured, generated, and deliberately synthesized climates were constructed for each site to isolate each variable's impact on predicted crop yields. The WEPP model, which was calibrated using wheat yields and hydrological data collected at El Reno, Oklahoma, was run for all sites. Because of independent assumptions, CLIGEN did not preserve proper serial and cross correlations between daily temperatures, solar radiation, and precipitation. The lack of proper correlations, compounded by model and parameter errors, considerably altered yield prediction for individual years. However, the probability distribution of the predicted yields, including mean and standard deviation, was satisfactorily simulated. This finding suggests yield distribution, rather than individual yields, should be used when conducting impact assessment with these models. The mean relative errors of the predicted yields using measured vs. generated climates with or without an interpolation were -3 and -6%, respectively. This result indicates that interpolated, more time-continuous climate data are superior for natural systems simulations.