基于PCA和SVM的遥感影像水体提取方法及验证

针对内陆湖泊水质及光谱特性空间差异性大、支流水系结构复杂而导致的遥感影像水体提取精度低的问题,提出了结合光谱主成分分析(PCA)及支持向量机(SVM)的PCA-SVM水体提取算法。基于GF-1卫星遥感影像,对原始影像光谱波段特征进行PCA降维,从中优选熵、方差、差异性纹理特征向量,结合原始波段及归一化差分水体指数(NDWI),构建了8维特征向量,并基于SVM算法提取湖泊水体。以巢湖洪水期与非洪水期影像为研究实例,分别采用NDWI法、传统SVM算法及PCA-SVM算法对水体进行提取,并进一步基于PCA-SVM算法对2020年汛期巢湖洪水期淹没演变过程进行反演和跟踪,定量解析特征向量组合及SVM惩罚系数C对水体提取性能的影响。结果表明：PCA-SVM算法提取的湖泊完整、支流连续,显著改善了含蓝藻水体漏提、建筑物误提等问题;洪水期和非洪水期提取结果的F1分数分别为95.08%和97.95%,虚警率分别为5.43%和1.13%,提取精度显著高于NDWI法和SVM算法。     

The glaciers of the Hindu Kush Himalayas: current status and observed changes from the 1980s to 2010

The fate of the Hindu Kush Himalayan glaciers has been a topic of heated debate due to their rapid melting and retreat. The underlying reason for the debate is the lack of systematic large-scale observations of the extent of glaciers in the region owing to the high altitude, remoteness of the terrain, and extreme climatic conditions. Here we present a remote sensing–based comprehensive assessment of the current status and observed changes in the glacier extent of the Hindu Kush Himalayas. It reveals highly heterogeneous, yet undeniable impacts of climate change.