基于卫星雷达数据的鄱阳湖湿地水面检测与淹露分析

Water surface change detection and analysis of bottomland submersion-emersion of wetlands in Poyang Lake Reserve using ENVISAT ASAR data

以鄱阳湖自然保护区湿地为例,对多时相、多极化的合成孔径雷达(ENVISAT ASAR)数据作主成分分析,得到主成分分量影像,其中次要分量波段包含大部分的局部变异信息,使用主成分分量波段和log算子提取了ASAR影像中湿地淹没变化信息.结果表明,鄱阳湖自然保护区湿地对外湖水位变化有滞后效应,但总体上,其淹没变化仍显著地受到鄱阳湖外湖洪水的影响.与分类后检测法、分类前变化探测法等方法相比,所采用的变化检测方法可得到更为理想的ASAR影像湿地淹没变化信息提取的精度.使用同极化/交叉极化组合模式的ASAR数据对水面变化检测的精度明显优于同极化波段(HH/VV)组合的数据.

The wetlands in Poyang Lake Reserve were selected as the study area. Based on multi-temporal and multi-polarization ENVISAT ASAR data, principal component analysis and generated the principal component images were applied, within which the higher components (i.e., those beyond the first one) defined the majority of localized anomalies. By using the higher component images and log operator, change detection of water surface indicated that although the wetland in Poyang Lake Reserve has a lagging effect on the movements of water level outside the lake basin, evidently the flooded wetland is affected by inundation in Poyang Lake. Compared with the widely used Post-classification Comparison Method and Pre-classification Change Detection Method, this paper presents more precise results and proves its own solution suitable for change detection of flooded areas using ASAR data. Moreover, an image which is made up of cross- and like-polarized signals from multi-temporal ASAR data offers a greater accuracy in water surface change detection than a combination of like-polarized (HH/W) signals.