Regional Application for Wheat Phasic Development Model and Comparative Analysis.

Abstract£ºThe key technique is to extended Mechanism-based wheat phenological model to the large region. Then, we compare the results of WheatGrow and CERES-Wheat phenological in the large region. TPS method was used to generate daily meteorological data of large region. Average regional sowing date of wheat was interpolate using TPS method through the using of historical sowing date of many years. Through the combination of the MCMC method and the two wheat growth models and using data of 98-03 wheat measured phenology to estimate the probability distribution of the region-specific parameters. Through the calibration in single point and region from 2003-2005, we extented the wheat phenological model to large region and solved the problem of scaling-up of crop models. The observed values agreed well with the estimated values in the calibration of single point. By analyzing the result in regional scale, WheatGrow model have higher prediction accuracy than CERES-Wheat. But the uncertainty brought by crop parameters are also higher than CERES-Wheat model. Moreover, the prediction accuracy and uncertainty in the large region is affected greatly not only by parameter estimation in single point. Therefore, in order to improve the accuracy of the simulation results in large region, through the growth period model system calibration, making a single simulation and experimental results ecological point more consistent, thereby increasing the region's forecast growth period results. Through combination the TPS interpolation method with MCMC-based parameter estimation techniques can effective extended mechanism-based model to regional scales, and quantify the uncertainty coming from the parameter estimation. WheatGrow and CERES-Wheat have their advantages in the application of large region and could be corrected systematic to improve the regional model prediction accuracy. Key words: Wheat; Phasic development; MCMC; Regional application