|
The accuracy of Landsat 5 multispectral scanner (MSS) data for mapping mangrove in Java is investigated and compared with SPOT 1 very-High-Resolution Scanner (XS) data. Supervised classification on mangrove is performed for SPOT XS data sets collected on August 1988 and Landsat MSS on August 1985. A detailed accuracy assesment is conducted based on ground data collected in 1988 and 1989. These results produce overall mangrove area classification accuracies of 74,2 and 82,8 percent for MSS and XS with 7 and 10 classes, respectively. While the accuracies for predominant categories are similar for both sensors, mangrove discrimination for less commonly occurring and/or spatially heterogeneous categories is improved with the XS data set. The MSS, however, performs similar, or better than the XS in classifying large homogeneous areas. The use of MSS data which is expensive than high resolution data appears promising for mangrove monitoring. |
