RADARSATSNBSAR数据在大面积水稻估产中的应用研究

由于雷达遥感的全天候、全天时的优势,使之成为南方大范围农业信息动态监测的最佳遥感手段。为了适宜于运行系统,本研究采用加拿大雷达卫星（RADARSAT）窄波扫描模式（SNB）数据,以广东省为例进行了大范围水稻估产。通过建立稻作图谱,解决了在地形复杂、农业集约化程度低、水稻田分布不规则,且地块间田间管理水平差异很大等诸多因素影响下的水稻信息提取问题;在野外观测站采集的水稻生长期生理生态数据的基础上,建立了基于RADARSAT SNB SAR的雷达遥感时序信息水稻估产模型。通过2000年早、晚两季水稻估产的实践证明,此模型估产的精度在平原区达到95％,而在复杂的丘岭谷地则需进一步提高图像预处理的精度,改进特殊地段估产模型的精度。从实用性而言,这是一套高效、经济的技术方法,易于投入实际运行阶段。

RICE YIELD ESTIMATION IN REGIONAL SCALE BY USING RADARSAT SNB SAR IMAGES

Radar remote sensing is the most appropriate for agricultural monitoring and crops yield estimating as the inherent prepotency and its capability of all-weather and day/night imaging. Especially, cultivated areas are most often cloudy and rainy in South China. For this reason, RADARSAT SNB SAR is the dominant data source in tropic and sub-tropical regions and also provided re-visit schedules suitable for monitoring in a regional scale. In the year 2000, it has been successfully done the rice yield estimation for early rice and late rice in whole province of Guangdong, China. Through setting up the rice cropping calendar, a solution of rice information extracting difficulty was found for reducing the influence of undulation landform, low level of intensive agriculture, irregular distribution of paddy and different rice cropping system. A rice yield estimating model with time series RADARSAT SNB SAR data was established for the region. And a yield mapping was produced with the classified result of the model. The final result shown, a high accuracy result was given in the plain but the result had to be improved at pre-processing and estimating models for the valleys of the hills and the low mountains. Nevertheless, the whole procedure was in the high efficient and quite economic and ease for putting in use as an operational system.