Methods for Detecting Land Use Changes Based on the Land Use Transition Matrix

Abstract

Land use change is a key component for global environment change, and also a representation of the impact of human activities on the environment. The cross-tabulation matrix is a fundamental tool in analyses of land use change, but many studies seem to be short of analyzing the matrix in terms of its various components, failing to obtain as much insight as possible concerning the potential processes that determine a general pattern of land use change. This paper aims to use the cross-tabulation matrix to assess the total land use change on the basis of the net change and transfer change. The relative differences between the observed transition and expected transition, which were generated by random transitions, can be used to extract systematic inter-category transitions. An example of changes among five land use categories in Linyi County in Shandong Province illustrated the effectiveness of the suggested method. The amount of total change for cropland was found largest, followed by forest land, unused land, construction land and others. The change in cropland was nearly a pure swap-type of change. In contrast, the unused land indicated no swap change but net change. Changes in others consisted of both swap and net change. The most prominent transition was the conversion from cropland to forest, accounting for 2.63% of the landscape, followed by the conversion from unused land to cropland, and cropland to construction land. If the gains occurred at random categories, it replaced the unused land at a rate eight times of the expected rate. The large conversion from cropland to forest could be attributed to the largest cropland in all categories, since the quantity of the conversion was almost equivalent to what would be expected from a random process. If the processes of loss occurred at random categories and the other land lost, forest tended to replace it at a rate exceeding two times of the expected rate. A relatively large conversion from cropland to forest could be as a result of the smallest area of forest in all land use categories, since the quantity of the conversion was larger than what would be expected from a random process. It was concluded that these methods can facilitate detection of the strongest signals from systematic land use transitions, which would be greatly helpful for analyses and modeling of land use change.