基于WOFOST作物模型的玉米区域干旱影响评估技术

Evaluation of regional drought disaster to maize using WOFOST crop growth model

为探讨作物区域干旱影响评估技术,拓展农业气象灾害区域影响评估方法,基于WOFOST作物模型,以西南玉米干旱为研究对象,对发育期、光合生产等模块进行了改进与提高,利用西南地区8个代表性站点的玉米田间观测数据和同期逐日气象数据对模型进行了适宜性检验,在此基础上,选择西南地区历年典型干旱年份,模拟分析了玉米的产量变化趋势,并与实际减产率作了对比分析。结果表明:改进后,玉米生育期模拟值与实测值归一化均方根误差(NRMSE)由原来的3.25%~6.95%降低到1.48%~3.07%,R^2由原来的0.57~0.79提高到0.63~0.99,平均模拟精度由原来的74.12%提高到78.9%。玉米产量模拟值与实测值NRMSE范围由原来的7.88%~11.99%降低到3.07%~6.79%,R2由原来的0.52~0.93提高到0.77~0.98,平均模拟精度由原来的75.7%提高到80.95%。对1987年、1992年及2006年西南地区关键生育期典型干旱年份产量模拟平均精度分别是69.8%、78.1%与75.9%。结合上述分析可得出模型对发育期与产量的模拟精度都有不同程度地提高,模型对西南玉米主产区干旱影响评估有很好的反应,对干旱分布范围与分布规律的模拟值与实际情况基本接近,表明该方法可为区域干旱影响评估提供一种更为科学的评估技术。

Timely and accurate assessment of the impacts of drought on maize in the maize producing areas in Southwest China for regional agricultural production planning has become increasingly important with more frequent and severe drought events. In order to effectively study the impact of drought on crop production, the assessment method of agricultural meteorological disaster should be expanded. This study evaluated the application of WOFOST model using field experimental data of maize and the parallel daily meteorological data collected at eight typical stations in Southwest China. The WOFOST model was improved for the phenology and photosynthesis modules before application. Based on the improved crop model, the historical impacts of drought on maize production in terms of change trend in maize yield for typical drought years were analyzed. The results showed that the normalized root mean squared errors（NRMSE） between the simulated and observed maize phenology reduced from the range of 3.25%–6.95% to 1.48%–3.07%, R2 increased from 0.57–0.79 to 0.63–0.99 and the average simulation precision increased from 74.12% to 78.9%. NRMSE between simulated and observed maize yields reduced from the ranges of 7.88%–11.99% to 3.07%–6.79%, R^2 increased from 0.52–0.93 to 0.77–0.98 and the average simulation precision increased from 75.7% to 80.95%. The average simulation precisions of maize yield were 69.8%, 75.9% and 78.1% in typical drought years of 1987, 1992 and 2006, respectively. The above results showed that the simulation precision of the phenology and yield of maize improved significantly and the model was therefore fully applicable in simulating the impacts of drought on maize production. The study provided a scientific technique for the drought assessment on regional maize production in Southwest China.