The Markov Process for the Forecast of Land Use Dynamic Changes in Sanjiang Plain

Abstract

Regarded Sanjiang Plain as the research object,based on multi-temporal remote sensing data,dynamic process of the land use was analyzed using 1995 and 2000 year's land use maps.The transition probability matrix about the land use type between 1995 and 2000 was calculated.Under the help of the Markov Process,the land use transition probability matrix was built to predict the land use situation in year 2005.At first,the remote sensing interpretation data of 2005 was used to validate the Markov Process,showing that the Markov Process is efficient and suitable for prediction in the area.Except for wetland(0.26%),other land-use types difference between prediction and interpretation value is less than 0.20%.Then,the land use pattern in 2010 and 2015 was predicted using the Markov Process.The results find that if the current land use policy is kept in the next 10 years,the areas of farmland and grassland will markedly increase,and the areas of wetland and forest land will obviously decrease.The share of farmland and grassland increase from 56.21% and 3.71% in 2005 to 63.88% and 3.88% in 2015;the share of forest land and wetland decrease from 28.05% and 7.46% in 2005 to 25.53% and 5.39% in 2015.According to the results of the prediction,the land use pattern can be adjusted,which may serve as a scientific basis for regional ecological environment protection and land resource rational utilization.