Research Of Land Use Change Prediction Based On CA-Markov Model In Hadoop Environment

The study of land use / land cover status in different time and space,and the prediction of land use change in the future,can provide a reference for the rational allocation of land resources.There are many kinds of data for land use change prediction,such as remote sensing image Dem image,all kinds of driving factor data,among which remote sensing image data has become a mass of data.The traditional land use change serial prediction algorithm based on CA-Markov model has a long running time under large-scale data,while the Hadoop cloud computing platform has the advantages of high performance,high stability and low cost in big data processing.Therefore,taking Hangzhou as an example,this paper studies the land use change prediction algorithm based on CA-Markov model in Hadoop environment(CLOUD-LUCP),the main tasks are as follows:(1)In order to obtain and parse the data by CLOUD-LUCP algorithm,according to the data structure of land use classification image,driving factor and other data needed by CLOUD-LUCP algorithm and the operation mode of CLOUD-LUCP algorithm,the method of organizing and storing the input data in HDFS is designed.(2)Based on the MapReduce programming model,the distributed parallel design of the traditional CA-Markov model serial prediction algorithm is studied,and the CLOUD-LUCP algorithm is obtained.The algorithm is divided into two parts: The parallel algorithm of Markov model based on cloud environment and the comprehensive evaluation algorithm of land use change based on cloud environment.The CLOUD-LUCP algorithm is used to calculate the land use area transfer matrix,the transition probability matrix,the land use comprehensive evaluation value of the cell,and then determine the change direction of the cell state according to the above data,thus realizing the land use change forecast.(3)The efficiency of the CLOUD-LUCP prediction algorithm and the serial prediction algorithm under different data amounts is compared.The experimental results show that under the cluster environment and test data set,the speed up ratio and the saving time of the parallel algorithm of Markov model based on cloud environment and the comprehensive evaluation algorithm of land use change based on cloud environment increase with the increase of data volume.The speed up ratio of the two-part algorithm reached the highest at 3.43 and 1.86 respectively.(4)The land use changes in 2013 were simulated and compared with the actual interpretation images in the study area.The experimental results show that the Kappa coefficients of the simulation results of nature reserve land,construction land and agricultural land can reach 0.86 0.68 and 0.66 respectively,which can be used to predict the land use change in 2020 in the study area.The results show that the CLOUD-LUCP algorithm is more efficient than the traditional serial prediction algorithm in large scale data.At the same time,the simulation results are reliable and the driving factors and weight parameters used in the algorithm are suitable for the study area,which can be used to predict the land use change in the study area.