A Study on Dynamic Forecast Technique of Regional Rice Productivity

Dynamic forecast of crop productivity is critical for food security and decision-making of agriculture manager. Based on the Rice Cultivational Simulation-Optimization and Decision-making System (RCSODS) and aggregation methods on the model inputs, and combined normal year meteorological data and spatial interpolation technique of weather data, the dynamic forecast technique of regional rice productivity was discussed in this study. The model and the aggregation methods were validated with the experimental data from eight National agro meteorological stations located in Jiangsu province and statistical yields. Rice productivity of Jiangsu was forecasted dynamically with observed meteorological data and normal year weather data generated from the monthly meteorological data. The results showed the RMSEs of development stages and yields between simulated and observed values were 5.0 d and 1344 kg/ha and the NRMSEs were 1.6% and 16.7% respectively, which indicated RCSODS could simulate the rice development process better and yield well. The RMSE and NRMSE between simulated and statistical county yields were 1081 kg/ha and 13.2%, which showed the aggregation method on the meteorological data, soil attributes, variety and cultural method in county level was suitable to the regional application of rice model. The forecast yield would generally be equal to simulated yield with the delay of forecast time.