摘要
为获取沉陷区积水深度，尝试采用LANDSAT5（TM）影像建立水深遥感定量关系．首先对TM影像进行几何校正、大气校正、沉陷区提取等步骤，然后输出像元反射率值，并与水深实测控制点坐标匹配，使水深值与反射率值对应．计算TM影像各波段及波段组合反射率与水深值之间的相关性，其中TM4波段的相关性最大（0．561），以TM4波段为例建立水深与像元反射率间的回归模型．模型以水深值2rn为阈值，包括水深实测值小于2rn和在2～6rn的2个回归方程．将模型应用到淮南潘一矿的沉陷区，计算反演水深值与实测值误差．结果表明：水深值小于2m的区域平均绝对误差0．1591m，平均相对误差13．367％；水深值在2～6m范围内的区域平均绝对误差0．57181m，平均相对误差15．02％．
To measure the water depth in ponding subsidence, a model based on TM image was proposed. After geometric calibration, atmospheric correction and ponding subsidence extraction, the pixels reflectivity was exported. In order to found the relation between actual water depth and pixels reflectivity, the pixels reflectivity were matched to control points. The corre- lation coefficients between reflectance of bands（combined bands） and water depth were calculated. The correlation coefficient between reflectance of TM4 and water depth is the maximal （0. 561）. The regression model was established between the actual measured water depth and reflectance of TM4. The threshold of the model is 2 m, including two regression equations which correspond to actual measured water depth less than 2 m and water depth from 2 m to 6 m. The model was applied to measure the difference between the actual water depth retrieved water depth in ponding subsidence of Huainan. The results show that the mean absolute error is 0. 159 1 m and the mean relative error is 13. 367%, when the actual measured water depth is less than 2 m. The mean absolute error is 0. 571 8 m and the mean relative error is 15.02%,when the actual measured water depth is from 2 m to 6 m.
出处
《中国矿业大学学报》
EI
CAS
CSCD
北大核心
2007年第4期537-541,553,共6页
Journal of China University of Mining & Technology
基金
国家重点基础研究发展规划(973)项目(2002CB211707
2006CB202209)
国家自然科学基金重大项目(50490271)
"十五"国家科技攻关重点项目(2001BA803B0403)
国家创新研究群体基金(50221402)
教育部"长江学者和创新团队发展计划"(IRT0408)
国家杰出青年基金项目(50025413)
教育部科学技术研究重大项目(306002)
关键词
沉陷区
水深
遥感
回归模型
ponding subsidence
water depth
remote sensing
regression model
