Over the last few decades several vegetation indices were used to map Mangrove forest using satellite images. Difficulty still persists in discrimination of mangroves from non-mangrove vegetation, especially in areas where mangrove species are mixed with other vegetation types. In the present study we have attempted to develop an improved index, which utilizes the information from the Normalized Difference Vegetation Index (NDVI) and the Normalized Difference Water Index (NDWI) of Bhitarkanika mangrove forest of Odisha, India. These indices are negatively correlated (r = -0.988; p < 0.01). Further, the NDWI values were subtracted from the NDVI values at the pixel level. As the outputs are negatively related, subtraction increases the upper and lower range of the overall output, also increasing the distinct values of two classes with near-similar spectral signatures. Same algorithm was applied on mangroves of Sundarbans (r = -0.987) and Andaman (r = -0.989). A comparison between four established indices [NDVI, NDWI, Soil Adjusted Vegetation Index (SAVI), Simple Ratio (SR)] and the newly developed index namely Combined Mangrove Recognition Index (CMRI) were performed. Accuracy assessment using Kappa statistics, revealing that CMRI produces better accuracy (73.43%) compared to other indices, followed by NDVI (56.29%) and SR (48.79%).
Keywords: CMRI; Combined Mangrove Recognition Index (CMRI); Mangrove discrimination; NDVI; NDWI; SAVI; SR.