Landsat5图像的增益、偏置取值及其对行星反射率计算分析

在对比不同时相的数据或建立遥感反演模型时,往往需要将图像的灰度值转换为反射率。增益和偏置是计算反射率时使用的两个基本参数。如果这两个参数发生偏差,反射率的计算结果就有问题,大气校正和其他相关工作的结果也是不可靠的。根据USGS文档和1995年以来的Landsat5遥感图像,对图像头文件中的增益和偏置参数的取值进行了分析,并按照标准单位将增益和偏置进行了转换。为了分析不同增益和偏置取值对反射率计算结果的影响,对比计算了植被和水体的行星反射率值。对2004年7月26日的太湖地区的TM图像。取值分为4种方案: (1)使用遥感软件ENV14．0计算;(2)使用图像头文件中的参数;(3)使用USGS提供的同期参数;(4)使用USGS提供的2003年前的参数。结果表明,不同增益和偏置取值导致计算结果变化较大。如果以方案3的USGS参数计算的结果为标准,那么,其他方案的相对误差值可从0．2％到20％以上。从各个波段的误差分布看,以第3波段的相对误差最小。从工作成果的可比性角度出发,建议统一使用USGS的增益和偏置参数计算Landsat5图像的反射率。

The Gains/Biases Values and Their Effects on the Calculation Results of Planetary Reflectance in Landsat-5 Image

In this paper, the gains and biases (G/B) of landsat-5 image data are summed up according to the USGS document and parameters in the head file of image, and the difference of G/B and its effects on the planetary reflectance is discussed. Planetary reflectance is calculated based on the image with a path of 119 and a row of 38, obtained on 26 July, 2004. Four kinds of G/B are used: 1) ENVI4.0 parameters dated 1986; 2) head file provided by L5 image; 3)the USGS document parameter dated after 2003; and 4) the USGS document parameter dated before 2003. The result shows that it is difficult to compare planetary reflectance value when the source of G/B is different. Taking the results of calculation by USGS parameter after 2003 as a standard, the reflectance of band 3 in terms of error distribution of other bands has the lowest relative error, and that of bands 1,5 and 7 have the highest relative error which may be greater than 20%. In order to make the TM application intercomparable, the planetary reflectance calculated using the G/B parameters provided by the USGS document may be a better choice.