Abstract
It is the well-known method to analyze the land use transition using a Markov chain model. This approach would assume that the land use transition had specific rules obeying larkov property, and estimate its transition matrix from two historical data of the sample area. The general versatillity and high efficiency in terms of computional efforts is the strengh of that method . However ,that has problems that it is difficult to modelize and forecast micro and random transitions that have spatially hetrogeneous rules. In this paper I propose a new approach to modelize the random land use transition applying Hidden Markov Models(HMM), that contain the symbol sequences that can be observed and the unobservable states obeying Markov process. In this investigation, I verify the effectiveness of HMM for estimating land use transition , by comparison of the extrapolation efficiency from numerical sample data, assuming the land use data consists of various areas obeying spatially different transition rule, with the model based on a Markov chain model.
