利用历史产量序列重构二维时变矩阵进行自然正交分解 ,得到了 3种影响产量变化的主要成份。通过分析发现 ,这些主要成份不仅包括能反映生产力变化的趋势分量和能表达气候影响效果的短期振荡分量 ,还发现了一种由气候决定的趋势作用分量 ,对产量有着不可忽视的影响。这些分量相互独立 ,无预先设定的函数模型 ,为产量分析和预报提供了较为客观和完备的历史变动序列 ,对提高产量预报精度有着重要意义。 Using the historical production sequence to reconstruct the two-dimensional time-varying matrix for natural orthogonal decomposition, we obtained three main components that affect the yield change. Through the analysis, we found that these main components include not only the trend component that can reflect the change of productivity and the short-term oscillation component that can express the effect of climate, but also a climate-dependent trend component that has an unnegligible impact on output. These components, independent of each other and without pre-defined function models, provide a more objective and complete sequence of historical changes in production analysis and prediction, which is of great significance for improving the yield forecast accuracy.