NOAA/AVHRR数据的几何精纠正方法研究

利用NOAA／AVHRR资料进行省级水稻遥感长势监测与估产是以高精度配准为前提的，因此首先论述了在利用TBUS报进行系统纠正的基础上，还必须进行几何精纠正的必要性及其方法的选择；提出综合利用ARC/INFO和ENVI的功能建立几何精纠正的标准空间，通过各种变换进行图像增强处理，仔细选取地面控制点；利用59次浙江省范围的NOAA/AVHRR数据计算出不同方法中的误差，首次通过方差分析和多重比较进行统计检验。结果表明：RST，二元一次多项式拟合法和二元二次多项式拟合法之间的RMS差异达到极显著水平，选定二元二次多项式拟合法作为几何精纠正方法，灰度值重采样采用三次卷积插值法，可作为采用NOAA/AVHRR资料开展省级水稻面积遥感估算、长势监测与估产的基础．

Accurate geometric correction for NOAA/AVHRR data.

It was pointed out in this paper that accurate geometric correction is the presupposition of rice growth monitoring and yield estimation using NOAA/AVHRR data at province level,therefore it is necessary to select an accurate geometric correction method which should be easy to use for non professional users and to assure the accuracy.After the standard space was established using ARC/INFO and ENVI function and a lot of images transform and enhancement was performed, the ground control points were selected.The combined RMSs were calculated for 59 NOAA/AVHRR data of Zhejiang province, The result of variance ahalysis and multiple comparisons showed that the RMSs differences of RST,1st degree polynomial and 2ed polynomial warping methods were significant.At last,the 2ed polynomial warping methods was selected as accurate geometric correction method and cubic convolution re sampling method was available for sub image of NOAA/AVHRR data for rice growth monitoring and yield estimation.