Climate Change and Rice Yields in Diverse Agro-Environments of India. I. Evaluation of Impact Assessment Models

Abstract

This paper reports results of a comparison of two popular rice growth models- Ceres-Rice and ORYZA1N for the same input conditions. Both models use different approaches for simulating growth and yield, are sensitive to climate change parameters, and represent two major schools of crop modelling. A dataset of 32 experiments consisting of 98 treatments was assembled from an extensive literature search. These experiments were conducted over the period of 1980–1993 in diverse Indian locations from 11° N–33° N. The treatments varied in N management, sowing dates, varieties and seasons. The flowering duration in the dataset varied between 37 and 86 days and grain yields between 2587 kg ha⁻¹ and 8877 kg ha⁻¹. Seven treatments from this dataset, one for each variety, were selected for calibration. The genetic coefficients of different varieties used in the analysis for both models were estimated from this short-listed dataset by repeated iterations until a close match between simulated and observed phenology and yield was obtained in these treatments. Similarly 11 treatments with zero or low N applications were used for calibration of initial soil N for different locations. The remaining 80 treatments were used for validation of the models. Both models predicted satisfactorily the trends of leaf area and dry matter growth, grain number, days to flowering and grain yields. Simulated yields were within +15% of the measurements. Considering that the field measurements also generally have 10–15% error and that the treatments widely varied in weather conditions, particularly in temperature, it was concluded that both models are adequate to simulate the effects of climate change on rice yields in diverse agro-environments of India that are free from all pests.