Abstract:
Remote Sensing and Geographical Information System are modem tools for
ecosystem management. Remotely sensed data gives convenient and rapid solutions
to problems in a variety of applications.
Land is limited, and vital as it is the main provider of important natural resources. The
fast growing human population has created many problems, due to the increasing
demands for food, water, shelter and fuel. Thus such socio-economic factors often
dictate how land is used regionally.
Land use affects land cover and in turn, changes in land cover affect land use. Thus
land plays a major role in any development process. In tropical countries, due to the
impact of human beings, the rates of change in vegetation cover and land use are high.
Hence frequent updating of land use maps is necessary to provide the information
needed by planners and politicians.
The main objective of this research is to investigate the possibility of using different
remote sensing satellite images for developing a land use/cover monitoring system.
This research is carried out in an area of approximately 400 square kilometres in the
southern part of Sri Lanka. Imageries of SPOT, IRS and Landsat satellites are used.
Different colour combinations are prepared and false colour composite images are
used for image processing.
Maximum likelihood method is used for image classification and the overall accuracy
of the classifications is more than 90%. Using this classification, change detection
matrices are developed to give changes for every land use class considered. A primary
problem encountered in the study area is the mixed pixels. It is difficult to separate
crop land from residential area, as some people reside in houses within the cultivated
area. Filtering techniques can only partially remedy this problem.
In order to monitor the land use/cover, image differencing method is applied and the
extent of the detected changes in terms of pixels or hectares is calculated.
A procedure is proposed as the land use/cover monitoring system using satellite
images. Under this monitoring system, the extent of land use/cover changes can be
computed by using different satellite images with varied spatial and spectral ranges.
