A thesis submitted in partial fulfilment of the requirement for the Degree of Master of Science, School of Environment, Resources and Development

Thesis (M.Sc.) - Asian Institute of Technology, 1994

Forest cover loss is becoming a major concern in the conservation and rehabilitation of the environment. This exploratory study aims to find out the forest cover loss of Gazipur district in central Bangladesh. This research presents a descriptive profile of the loss of forest cover in the past 18 years (1974-1991) in Gazipur. A model is developed for the study of deforestation based on the correlation ship between population density and forest cover (%) in Gazipur. An attempt has also been made to identify the cause for forest cover loss. A set of guidelines has been proposed for sustainable development and future management of forest cover in the study area . Remotely sensed data (Landsat TM ) satellite imagery integrated with geographical information system formed the basis for mapping forest cover of the study area. The GIS layers include river, road, railway network, urban settlement, forest cover type and forest boundaries. The mapping was performed using field survey, aerial photograph, topomap, forest cover map and Landsat Thematic mapper satellite data. Landsat TM 1991 image was used for updating the forest cover map . A questionnaire survey was conducted for collecting preliminary data. A total of 126 respondents were interviewed using the survey questions as an instrument in data gathering. The data gathered were analyzed by employing a descriptive and comparative statistical method using the application software {u2022}statistical Package for Social science' (SPSS) . The study revealed that the forest cover loss during the past 18 years was serious. The average rate of forest cover loss was as high as 8.79% per year. The rate of forest cover loss varied in different years. From 1989 to 1991 the rate of forest cover loss was lower compared to that during 1974 to 1989. It was found that the two most important changes were transformation of forest area to non- forest and the non forest area to forest area . Reasons for changes f rom forest to non-forest was mainly due to population increase, unclear tenural condition or ownership, timber harvesting, new settlement, forest road construction . Changes from the non-forest to forested area, occurred mainly due to the implementation of a social forestry project and home garden in the area . In addition, this study evaluates the potential of measuring/mapping forest cover using Landsat Thematic mapper. Landsat TM bands 2, 3, 4 was found to be the best for the discrimination of forest and non forest area in the study area. Higher spectral reflectance was observed for forest as compared to paddy, scrubs and mixed area. Unsupervised classification of the Landsat TM data did not clearly identify all cover types. Supervised classification was tested by stand agreement. Interpreting Landsat TM image errors interpretation were categorized as commission, omission and were further classified into errors of identification. over 55 percent of all errors were less than 2.0 hector in size. Satellite remote sensing techniques with GIS technology have provided an efficient and cost effective method of streamlining and updating maps. Finally, although the growth of population is crucial to loss of forest cover, much of this decrease can be ascribed to poor policies and management related to the forest resource, and change in the economic and social structure. Furthermore, improvement of living condition of local people was an important factor. Recommendations include improvement of forest management systems, institution, and forest market.