潮滩滩面高程的高光谱遥感反演研究  

Inversion of Tidal Flat Elevation Based on Hyperspectral Remote Sensing

作　　者：陶旭[1] 张东[1] 

机构地区：[1]南京师范大学地理科学学院,江苏南京210046

出　　处：《海洋科学进展》2013年第4期498-507,共10页Advances in Marine Science

基　　金：国家自然科学基金资助项目--潮滩滩面高程的高光谱定量遥感反演方法研究(40606044);浅海潮流振荡边界层动力特性及悬沙垂直分布研究(41076008)

摘　　要：以江苏大丰王港潮滩为研究区,利用从Hyperion高光谱影像和Lidar测量地形资料上提取的地表反射率与高程数据进行统计相关分析,建立单波段模型和多波段线性回归模型,进行了滩面高程遥感反演试验,探讨了利用地表光谱信息直接反演潮滩高程的可行性。研究结果表明:潮滩滩面的沙脊和潮沟等地貌形态对地表反射率有较大影响,需要通过空间分区的方式将潮滩分为平坦区和起伏区,分别进行高程反演;短波红外波段区间的地表反射率对滩面高程变化较为敏感,1 164,1 275,1 336nm波段是高程反演的理想波段,利用理想波段建立的单波段模型和多波段线性回归模型均可有效用于滩面高程反演,但是多波段线性回归模型在高程反演精度上没有明显改善;从精度分析来看,研究区滩面平均高差135.5cm,遥感反演高程与测量高程在北滩MAE、MRE和RMSE分别为7.5cm,6.2%和9.5cm,南滩为36.75cm,141%和45.2cm,表明遥感反演技术能够有效模拟出潮滩地形的平均变化趋势,得到潮滩的概要地形信息,从而为大范围潮滩的冲淤变化分析提供可靠依据。Taking the Wanggang tidal flat of Dafeng City of Jiangsu Province as the study area, two types of elevation inversion models named as single-band linear regression model（SBLRM） and multi-band linear regression model（MBLRM） were built up by using the surface reflectance data extracted from the Hyperi- on hyperspectral remote sensing images and the elevation data measured with Lidar techniques. Using these models, a test of tidal flat elevation inversion based on remote sensing data was carried out and the feasibility that the elevation of the tidal flat can be directly inverted from the surface spectral information was discussed. The results show that： 1） Topography such as sand ridges and tidal creeks on the tidal flat can exert a great impact on the surface reflectance so that the tidal flat should be divided into the flat and the undulant regions spatially and the inversion of the tidal flat elevation should be done separately; 2） The surface reflectance in short-wave infrared ranges is sensitive to the change of the tidal flat elevation. Those bands with central wavelengths at 1 164nm.1 275nm and 1 336nm are ideal for the inversion of tidal flat el- evation. The SBLRM and the MBLRM built by using the ideal bands can be effectively applied to the in- version of the flat elevation, but the MBLRM does not improve the accuracy of the elevation inversion evi- dently; 3） From the view of accuracy analysis, the average elevation difference of the tidal flat is la5.5cm in the study area, and the mean absolute error （MAE）, the mean relative error （MRE） and the root mean square error （RMSE） of the inverted and the measured elevations are respectively 7.5cm, 6.20//oo and 9.5cm for the northern tidal flat and 36.75cm, 141~ and 45.2cm for the southern tidal flat, indicating that the average changing trend of the tidal flat topography can be effectively simulated by using the remote sensing inversion technique. Thus, the outline of the topography of the tidal flat can be obtained and a reliable b

关 键 词：苏北潮滩 潮滩高程 高光谱遥感 

分 类 号：TP7[自动化与计算机技术—检测技术与自动化装置]