SPATIAL MODELING TO UNDERSTAND LAND COVER DYNAMICS USING MARKOV CHAIN IN THE UPPER CITARUM WATERSHED

• Main Author: Wijayasari, Winda
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/73363
• Institution: Institut Teknologi Bandung

Modeling regarding land cover will continue to be needed, especially in developing countries, because it is related to the increasing population and the need for agricultural production and settlements. The watershed is an area that has essential functions. Land cover changes must be monitored because they are closely related to water availability and the land's ability to prevent floods naturally. The emergence of NASA's product, MODIS MCDQ12, which is specifically made with an algorithm for land cover, can potentially improve the land cover model because it offers more accurate data in the land classification process. Markov Chain is very efficient in seeing changes in model land cover because it can simultaneously include analysis of changes in space and time. So that land changes can be wellstudied pixel to pixel. Using the Markov chain, the error value generated from the Upstream Citarum watershed projection for the 2001-2020 period is 15.9%. Surface water data is also involved to see the effect of water availability in driving land change. Both data are combined by resampling the MODIS and surface water data into the exact resolution of 500 m. Land cover change in the Upper Citarum area was massive in 2015-2020. This was due to water availability in the Upper Citarum, which became the main attraction for urbanization and accelerated development in the peri-urban areas of the city of Bandung. Surface water data is added to test the relationship between water availability and land change trends. As a result, the transition matrix shows that locations with ample water availability tend to experience changes in land use, starting from changes in the forest to agriculture, then agriculture to builtup areas. This research was completed using code in the R language, which was then collected into a new package called Mallacha which offers functions for conducting land cover analysis using the Markov chain