Comparison of object-based and pixel-based Random Forest algorithm for wetland vegetation mapping using high spatial resolution GF-1 and SAR data

摘要: Abstract Vegetation is an integral component of wetland ecosystems. Mapping distribution, quality and quantity vegetation important for protection, management restoration. This study evaluated the performance object-based pixel-based Random Forest (RF) algorithms mapping using a new Chinese high spatial resolution Gaofen-1 (GF-1) satellite image, L-band PALSAR C-band Radarsat-2 data. research utilized wavelet-principal analysis (PCA) image fusion technique to integrate multispectral GF-1 synthetic aperture radar (SAR) images. Comparison six classification scenarios indicates that use additional multi-source datasets achieved higher accuracy. The specific conclusions this include followings:(1) GF-1, images found statistically significant difference between methods; (2) RF classifications both greater 80% overall accuracy fused with SAR images; (3) improved 3%-10% in all when compared classifications; (4) produced by integration outperformed any lone datasets, 89.64%