A thesis submitted in partial fulfillment of the requirements for the degree of Masters of Engineering

Thesis (M.Eng.) - Asian Institute of Technology, 2002

This study focussed on the comparison and assessment of three crop growth models under tropical climatic conditions. CERES-Maize, CropSyst and SWACROP models were selected for the study. Secondary data from eight experiments, conducted at four different sites in Thailand to run CERES-Maize model, were used as inputs in the other two models. The theoretical structure of the models was compared and all the three models were run to simulate moisture content in different soil layers, evapotranspiration, maximum leaf area index, biomass and grain yield. It was found that CropSyst requires many cultivar and site specific data. SW ACROP does not require much site specific and cultivar specific data and also does not take into account the initial nitrogen and organic matter present in the soil hence it produced poor simulation and predicted biomass, yield, max maximum leaf area index poorly. Biomass, grain yield, moisture content in different soil layers and evapotranspiration were well simulated by CERES-Maize and CropSyst. Moisture content in different soil layers showed similar variation with time in CERES-Maize and SWACROP. However CropSyst did not give very good simulation of soil moisture in all the layers except the top layer. CERES-Maize and CropSyst produced almost the same potential evapotranspiration while SW ACROP predicted higher potential evapotranspiration. All the three models simulated different actual evapotranspiration but CERES-Maize gave higher amount of actual evapotranspiration. Statistical parameters showed that CERES-Maize predicted biomass, yield and maximum leaf area index well (%RMSE = 7.43, 9.53 and 7.03; CD= 0.94, 0.83 and 0.52 for biomass, yield and max. LAI respectively). CropSyst predicted biomass and yield very well however it did not give good simulation of maximum leaf area index (%RMSE = 4.27, 8.66 and 30.58; CD = 0.96, 0.83 and 0.16 for biomass, yield and max. LAI respectively). SW ACROP could not simulate any of the three parameters very well and gave poor simulations (%RMSE = 15.45, 18.51 and 13.18; CD= 0.72, 0.73 and 0.26 for biomass, yield and max. LAI respectively). Biomass and yield were significantly underestimated by SW A CROP. This could mainly be attributed to the lack of cultivar specific input requirements. Sensitivity analysis showed that models are sensitive to certain input parameters e.g., potential kernel number in CERES-Maize, specific leaf area in CropSyst and maximum water use efficiency in SW A CROP, which need to be determined by experiments.