基于马尔可夫过程的三江平原土地利用动态变化预测

基于多时相遥感数据，应用GIS技术对三江平原1995和2000年的土地利用数据进行动态分析，计算出1995~2000年的各土地利用类型面积转移矩阵；借助马尔可夫过程建立土地利用转移概率矩阵，预测2005年土地利用结构，并将其与2005年遥感解译的土地利用数据进行比较。结果表明：马尔可夫预测值和遥感解译结果吻合较好，在7种土地利用类型中，除沼泽湿地外（0.26%），其它几种类型的预测值和遥感解译结果的偏差都小于0.20%。同时，利用该模型对三江平原2010年和2015年土地利用结构进行预测，结果表明，在当前稳定的土地利用政策下，未来10 a内，该地区农田和草地面积将显著增加，同时，湿地和林地面积显著减少。农田和草地所占比重将分别从2005年的56.21%和3.71%上升为2015年的63.88%和3.88%；林地、湿地所占比重将从2005年的28.05%、7.46%下降为2015年的25.53%、5.39%。这一预测结果对该区域的生态环境保护和土地资源的合理利用具有一定的参考价值。

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

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.