Improving Remote Sensing Identification Accuracy of Mangrove Using Temperature and Moisture Information

The identification accuracy of mangrove is always low by using TM reflective bands due to the similarity of spectra between mangrove and land vegetation, especial water-vegetation mixed pixels. Based on reflective and thermal infrared information in the TM images of the different tide levels, temperature - moisture index (TMI) was proposed. The analysis results show that the thermal infrared band and TMI can obviously improve the separability between mangrove and the other objects based on the tide level information. The thermal infrared band and TMI can also significantly increase the classification accuracy of mangrove by using spectral angle mapping (SAM) supervised classification method comparing with the classification features employed by other researchers. The Kappa coefficient increased 0.14 as well as the commission error of mangrove class decreased 19.9 %, showing that the remote sensing identification accuracy of mangrove can be improved by using the information of tide level, thermal infrared band and TMI.