Monitoring Mangrove Forest Dynamics of the Sundarbans in Bangladesh and India using Multi-Temporal Satellite Data from 1973 to 2000

Background

In the Sundurbans of India and Bangladesh, the authors measured the extent and condition of the mangrove forest at three intervals using GeoCover datasets (Landsat MSS, TM, ETM+) with the goal of assessing the current extent of the remaining forest, measuring change in the extent of the forest in different time periods, and identifying areas of intensive deforestation or degradation and changes in patterns of canopy density.

Research Goals & Methods

The data was classified into mangrove, non-mangrove, flooded, barren lands, and water bodies using a supervised Maximum Likelihood Classification based on training regions. Change detection was then carried out using post-classification techniques designed to minimize the common sources of error. The authors carried out a post-classification change matrix function to examine net gain and loss of the forest between 1970-2000s, 1970-1990s, and 1990-2000s. NDVI range was also calculated and used to reveal relative patterns of canopy closure. Finally, a confusion matrix was prepared using training points collected from QuickBird images, aerial photographs, and mangrove forest classification maps.

Conclusions & Takeaways

The area of mangrove forest decreased by 1.2% in the 25 years of the study. However, the rate of change was not uniform throughout; from the 1970s to 1990s, mangrove forest area increased by 1.4%, and from 1990s to 2000s, the area decreased by 2.5%, changes which are well within the error margin for satellite-based classification. The authors calculated that while total mangrove area remained the same the transition between land categories was much greater than net change. They attribute the changes, concentrated in the outer periphery and near the shoreline, to possible tidal variations, encroachment, erosion, aggradation, and mangrove rehabilitation programs.