基于气象资料的中国冬小麦收获指数模型

【目的】建立基于气象资料的中国冬小麦收获指数统计模型,为构建作物产量模型提供支持。【方法】获取河南、河北、山东合计30个农业气象观测站近20年冬小麦农业气象观测数据,利用时间序列分析方法提取趋势收获指数,收获指数观测值与趋势收获指数差值即为由气象要素决定的气象收获指数。采用逐步回归分析方法建立基于气象要素的冬小麦收获指数统计模型,模型模拟并反映气象要素变化对冬小麦气象收获指数的影响。【结果】冬小麦关键生育期气象要素与气象收获指数相关关系显著,但单站尺度和区域尺度的显著性水平存在差异。利用209组独立数据,分别在单站和区域尺度对建立的冬小麦收获指数模型进行了验证,单站尺度上模型模拟值与实测值的线性相关系数和斜率分别是0.65和0.4(2n=209,P0.001),均方根误差12.2%,平均偏差-2.4%,拟合指数75.8%,模拟效率42.3%;区域尺度上模型模拟值与实测值的线性相关系数和斜率分别是0.56和0.33(n=209,P0.001),均方根误差13.3%,平均偏差-1.3%,拟合指数为69.0%,模拟效率为31.7%。【结论】基于气象资料构建的冬小麦收获指数模型可以较好地模拟不同气象条件下冬小麦收获指数的动态,该模型可与作物NPP模拟模型相耦合,用于区域尺度上冬小麦产量的模拟研究。

The Harvest Index Model of Winter Wheat in China Based on Meteorological Data

【Objective】 The harvest index model of winter wheat in China based on meteorological data was established for supporting crop yield model. 【Method】 Based on observed data of winter wheat and meteorological data during recent twenty years from thirty agrometeorological stations in Henan, Hebei and Shandong provinces, the trend and meteorological harvest index were extracted by using time-series analysis method. The trend harvest index was separated from the measured harvest index by linear fitting; the meteorological harvest index was then calculated by subtracting the trend harvest index from the measured one, and the relationship between the meteorological harvest index and the meteorological factors was studied by stepwise regression analysis which simulated the influence of the changed meteorological factors on winter harvest index. 【Result】Correlation analysis suggested that the meteorological harvest index of winter wheat significantly correlated with the meteorological factors during the growth period. However, the influencing factors were different in different scales. Both statistical models of single station scale and regional scale were validated, respectively, using other data (n=209). The model validation showed that, the simulated linear regression as against measured harvest index yielded a correlation coefficient (r) of 0.65 with a slope of 0.42, the root mean squared error of 12.2%, the mean bias error of -2.4%, the index of agreement of 75.8% and the model efficiency of 42.3% on the single station scale, and the model almost explained the total variance. The simulated regression against measured harvest index yielded a correlation coefficient (r) of 0.56 with a slope of 0.33, the root mean squared error of 13.3%, the mean bias error of -1.3%, the index of agreement of 69.0% and the model efficiency of 31.7% on regional scale. 【Conclusion】 The established models can well simulate the change of winter wheat harvest index under different meteorological conditions. The winter wheat harvest index model coupled with crop net primary production model can be used to do research on simulating winter wheat yield on the regional scale.