A STUDY ON MANGROVE COVER DYNAMICS AND CLIMATIC VARIABILITY IN SUNDARBANS USING REMOTE SENSING AND MACHINE LEARNING

Abstract

The dynamics of forest density of Sundarbans is crucial for Bangladesh. Climatic variability has a gigantic impact on this mangrove forest cover dynamicity. Changes in temperature, precipitation, and evapotranspiration patterns have been proposed as factors influencing changes in the composition and spatial distribution of mangrove species. To monitor this changing pattern of mangrove forest in Sundarbans and its climatic variability advanced techniques such as remote sensing and machine learning need to be applied. The main purpose of this research was to analyze the relationship between meteorological variables and dynamics of mangrove forest density mainly normalized difference vegetation index (NDVI) and to predict it based on this meteorological variable over the Sundarbans, using the last 30 years’ time period (1992 to 2021) data. To investigate the changing pattern, the overall 30 years were split into ten years as follows (i) 1992 to 2001, (ii) 2002 to 2011, and (iii) 2012 to 2021. Firstly, the 30-year Landsat satellite data was analyzed using the Google Earth engine to determine the changing nature of NDVI at Sundarbans; then, the meteorological variables changing nature were analyzed along with the NDVI to see the relationships among them. Finally, machine learning algorithms (Random Forest, Support Vector Machine with Radial Basis Function, and Support Vector Machine with Polynomial Function were used to predict the NDVI over Sundarbans using the relationships between NDVI and maximum & and minimum temperature, precipitation, and potential evapotranspiration. It has been found that the Random Forest model performs well for predicting the NDVI over Sundarbans as it has a lower mean squared and mean absolute error.