RS Application for conducting change detection within the Sundarban Mangrove Forest, Bangladesh to meet REDD+ initiatives

The U.S. Forest Service (USFS) provided technical support to the Resource Information Management System (RIMS) unit of the Forest Department (FD) of Bangladesh in developing a method to monitor changes within the Sundarbans Reserve Forest using remote sensing and GIS technology to meet the Reducing Emissions from Deforestation and Degradation (REDD+) initiatives within Bangladesh. It included comparing the simple image differencing method with the Z-score outlier change detection method to examine changes within the mangroves of Bangladesh. Landsat data from three time periods (1989, 1999, 2009) were used to quantify change within four canopy cover classes (High, Medium, Low, and Very Low) within Sundarbans. The Z-score change analysis and image differencing was done for all the 6 reflective bands obtained from Landsat and two spectral indices NDVI and NDMI, derived from these bands for each year. Our results indicated very subtle changes in the mangrove forest within the past twenty years and the Z-score analysis was found to be more useful in capturing these subtle changes than the simple image difference method. Percent change in Z-score of NDVI provided the most meaningful index of vegetation change. It was used to summarize change for the entire study area by pixel, by canopy cover classes and the management compartment during this analysis. Our analysis showed less than 5% overall change in area within the mangroves for the entire study period. Percent change in forest canopy cover reduced from 4% in 1989-99 to 2% by 1999-2009 indicating an increase in forest canopy cover. Percent change in NDVI Z-score of each pixel was used to compute the overall percent change in z-score within the entire study area, mean percent change within each canopy cover class and management compartments from 1989 to 1999 and from 1999 to 2009. The above analysis provided insight to the spatial distribution of percent change in NDVI between the study periods and helped in identifying potential area for management intervention. The mean distribution of change from both study periods was observed within ± 20% SD.Our results were in agreement with the independent field study conducted by the US Forest Service earlier the same year for biomass and carbon stock estimation. The 10m field plots that showed a decline in carbon stock between 1995 and 2010 overall coincided with the compartments or region that showed a decline in forest canopy cover between 1999 and 2009 from the present analysis. These results led us to believe that the Z-score analysis can be a potential quantitatively rigorous tool to quantify change in ecosystems that are mostly stable and do not undergo drastic land use or land cover change. The field and remote sensing study together provided important scientific information and direction for future management of the forest resources, baseline information for long term monitoring of the forest, and identifying potential REDD+ Carbon financing projects in Sundarbans, as well as other potential REDD+ sites within forested area of Bangladesh. Given the rising concern and interest in REDD+ initiative we consider the Z-score analysis to be a potential tool in monitoring and providing a quick spatial assessment of change using remote sensing technology.