Mapping of mangrove forests is a very difficult task due to its complexity in physical characteristics. However, its protection and maintenance area necessity, as they are an ecological hotspot. Synthetic Aperture Radar(SAR) data integrated with optical data has been widely used for wetland application for better classification results. This study assess the applicability of dual polarimetric C-band data (Sentinel-1) in mangroveforest mapping together with Sentinel-2 optical data. Along with that the significance of commonly used spectral indices and SAR parametersare also examined. It was performed by running a separability analysis using Jeffries-Matusita (JM) distance and a multicollinearity analysis using Spearman’s rank correlation. Google Earth Engine platform was used for preprocessing and Random Forest (RF) classification of data.The result generated from multi-season data collected over a year indicated that selection of significant features for classification improved theclassification accuracy of both the optical and the SAR data. The overall accuracy of the optical image and the SAR image increased by 7.0% and 5.0%, respectively. Moreover, integration of feature subsetted fromoptical and SAR data again resulted in a rise in accuracy compared to classification results of SAR and optical data alone. JM distance and correlation analysis identified 23 - 33 features as significant for classification among initial 63 features of optical and SAR data, by reducing data dimensionality. This paper comprehensively discusses the importance of selecting suitable features for the classification of mangrove forests.