美国陆地卫星三代光学传感器的辐射转换方程——以鄂尔多斯地区MSS/TM/ETM+影像为例

Radiation Conversion Equations of the Landsat Series of Sensors—A Study Case of MSS,TM and ETM + Images over the ORDOS Region

以鄂尔多斯地区的Landsat MSS/TM/ETM+3种不同影像数据为对象,利用伪不变特征(PIF)法建立辐射归一化方程,对3种传感器红光波段和近红外波段的反射率数据进行辐射归一化处理,并对辐射同化结果从均方根误差、NDVI差值以及动态范围3个方面进行了分析和验证。研究表明,3种传感器对应的波段具有极强的线性相关性(R2>0.92),利用研究得到的转换关系方程实现3种传感器对应波段的高精度转换(RMSE<0.041),校正结果有效地减小了非地物变化引起的辐射差异,实现了多时相、多传感器数据的反射率同化,改善了动态监测效果。

The radiation normalized equations was established by using the method of PIF with the three different images of Landsat MSS/TM/ETM＋ of the ORDOS region. The radiation normalization processing to the three sen-sors′reflectance data of the red and near infrared band was taken, and the root-mean-square error, NDVI difference and the dynamic range were employed in analyzing and verifying the normalized images. The research showed that the corresponding band of three kinds of sensors had a very strong linear correlation （ R2＆gt; 0.92） , and the high ac-curate conversion with the RMSE less than 0.041 for corresponding bands of three sensors was accomplished by u-sing the obtained transformation equations. The calibration results effectively reduced the radiometric difference caused by the non-feature changes. The method of the paper completed the normalization of the surface reflectance for multi-temporal and multi-sensor data, and improved the conditions of the dynamic monitoring.