利用MODIS植被指数时间序列这一特性，以北京市通州及周边为实验区，冬小麦种植面积为研究对象，提出 了农作物种植面积指数模型(Pan-CPI模型)的概念，并构造了冬小麦特征物候期植被指数与种植面积的定量函数关系， 通过样区TM影像求解关键参数，对研究区冬小麦种植面积测量方法进行了试验研究。研究结果表明：(1)Pan-CPI模 型能够很好地反映特定目标农作物种植面积状况，为基于植被指数时间序列影像识别农作物种植面积提供了新方法； (2)精度分析结果表明：Pan-CPI模型具有很高的稳定性，且不受样本变化的影响，只要达到满足模型计算的样本量(如： 5%)，多次测量结果间具有很好的一致性。选取MODIS 6×6像元大小的窗口时，TM样本的复相关系数(R2)稳定在0.85 左右，与TM结果比较，窗口相对精度稳定在95%左右，区域精度稳定在92%以上，经调整的区域精度高达96%以上； (3)对于种植结构复杂、目标作物种植破碎的地区，Pan-CPI模型可以充分利用MODIS植被指数时间序列的优势，有效改 善TM单时相和多时相提取信息因时相缺失无法表征作物变化的不足。

Other than the refl ectance comparing to natural vegetation, various types of crop have their own representative phenological calendar features. This dramatic change of crops along seasons makes a great difference to the regular order changes of natural vegetation. MODIS-VI time series become the best indicator for these phonological features. In this paper, a new crop area index, called Pan-CPI, is proposed to refl ect the quantitative functional relationship between the MODIS-EVI time series and crop planted area. The research region is located at Tongzhou, Beijing and its surrounding areas. The winter wheat planted area was determined by the key parameters of the Pan-CPI model from the samples collected by TM images and MODIS-EVI time series. The results demonstrated that: (1) The Pan-CPI model can well monitor the goal crop area and provide a new method for crop area estimation based on MODIS-EVI time series. (2) Accuracy analysis shows that: As long as the population of samples meet the requirement of model calculation (for example: 5%), the Pan-CPI model has a high stability to get a high consistency between multiple measurements and will not be infl uenced by different samples. While the size of stats window is 6×6 MODIS pixels, the multiple correlation coeffi cient (R2) reach above 0.85.Compared with the result of TM, the Overall Windows Accuracy stabilizes at around 95%, Total Quantity Accuracy stabilizes above 92% and Post-Adjusted Total Quantity Accuracy reaches up to 96%. (3) For the area with complex, fragmental cultivation structure, the Pan-CPI model can provide a more reasonable crop area estimation than those results from TM, which may easily be infl uenced by the loss of images in key phenology phases.