不同尺度遥感数据源的选取将直接影响到作物种植面积测量的精度,研究尺度因子在农作物面积遥感测量中的作用,尺度与面积测量精度的定性和定量关系是非常必要的.为此,本文利用SPOT5卫星数据,以尺度变化对农作物种植面积遥感测量精度影响的分析为主线,运用空间统计分析方法和多种精度评价指标,从不同空间分辨率、不同空间范围、不同农作物百分比等角度系统分析了农作物种植面积遥感测量中的尺度效应问题.为基于多尺度遥感数据复合的农作物种植面积测量业务化运行中的数据选择和精度保证问题提供理论与实验基础.

Themeasurementofcrop planting area is essential to yield estmi ation by remote sensing. The areameasuring accuracy directly affects the reliability and applicability of the results. The application ofmulti-scale remotely sensed data is the inevitable trend. The selection of different scale remotely sensed data sources directly affects the accuracy of measurementof crop planting area. It is necessary and valuable to study the effect of scale on crop planting area measurement, aswell as the relationship ofqualitative and quantitative research. Therefore, this paperused SPOT5 data and analyzed the scale change effect on the accuracy ofmeasurement by remote sensing. Spatial statistic method and manifold accuracy evaluation indiceswere respectively used to analyze them with various spatial resolution, spatial extent and crop percentage. And the study results indicate: (1) with the decrease of the spatial resolution, the average regional accuracywill decrease and the errors of crop planting areameasurementwill increase due to the augmentation ofmixed pixels; (2) with the enlargement of the spatial extent of themeasuring region, the average regional accuracy increases because the proportions error can be counteracted reciprocally ata certain extent; (3) the percentage ofcrop can exerta great influence on the choice ofdatawith differentspatial resolution used in themeasurementofcrop planting area. When\nthe percentage of the crop is above 30%, themean region accuracy atdifferent resolution is basically above 90%. When the percentage reaches 40%—50%, the differences of themean regional accuracies atdifferent resolution reach the least value and is almost above 93%. W ith the percentage unceasingly increasing, the mean region accuracies at lower resolution (30m, 50m, 70m, and 90m) almostkeep unchanged. When the percentage of30m resolution is above 10%,themean region accuracy can stabilize over95% even if themeasuring extent is smal.l The results of this paper provide academic and expermi ental reference to resolve the problem of data selection and accuracy pledge in operational crop planting areameasurementby remote sensing.