Simulation Of Land Use Change In Charchan Oasis Based On SD Model And CLUE-S Model

As an extreme arid, semi-arid and unique ecological unit, oasis is the basis which maintains human’s existence, activities and development in arid region. In the process of Long-term improper develop and utilize of oasis, human ignored the coordinated development of economy and ecological environment. Land use reflects the most direct, closest, mutual influenced and interactive relationship between human and nature. Land use modelling is an effective method to analyze the causes, process and driving forces of land use. It also helps to understand the ecological effect of land use and offers land use planning and policy-making support.This study takes the typical oasis of southern Tarim Basis of Xinjiang, Charchan Oasis, as an example. Based on the remote sensing interpreted data of 1990, 2001 and 2013, analyzed the change range and dynamic degree of land use types in 1990-2013. We improved CLUE-S model by replacing Logistic regression model with Auto-logistic regression model. Selected several humanistic and natural as spatial driving forces, take the simulated result of SD model as land use demand of CLUE-S model. On that basis, we simulated the land use changing process of Charchan Oasis in the period of 1990-2013 and 2013-2025. Main method and conclusion are as follows:(1) In 1990-2013, land use structure of Charchan Oasis has changed greatly, natural oasis developed into the artificial oasis is the main one, continued expansion of oasis is the general trend. According to the single land use type dynamic degree, cultivated land and construction land are the most active land use types, there are intense transfer interactions between land use types. Comprehensive land use dynamic degree of the three research periods( 1990-2001, 2001-2013, 1990-2013) are round 0.345, combined with single dynamic degree, it indicate that, no matter how rapid and intense the interactions between land use types are, land use change remained in a stable changing state at all times.(2) Take the mutual transfer probability of land use types as the velocity variables and 6 land use types as 6 stocks, built the SD model and simulate the scale of land use of 1990-2013 and 2013-2025. The result of 2013 is ideal, but there is deviation in the result of 2001. Under the assumption of mutual transfer probability in 1990-2013, it reflects that some types of land use were reduced. In the predicted result of land use area 2025, cultivated land will be 36137.32hm2, wood land will be 24477.95hm2, grass land area will be 123270.76hm2, water body will be 7084.09hm2, unused land will be 336760.62hm2, construction land will be 3940.26hm2.(3) Form Auto-logistic regression model by adding spatial autocorrelation factor into the traditional Logistic regression model. Compare fitting degrees of two regression model by ROC and spatial probability distribution map. ROC value of cultivated land, wood land, grass land water body, unused land and construction land are 0.970, 0.827, 0.788, 0.875, 0.850 and 0.955 in Logistic regression model, and it has been improved into 0.977, 0.940, 0.847, 0.894, 0.893 and 0.989. According to the spatial probability distribution map, fitting degree of Auto-logistic regression model is higher than the fitting degree of Logistic regression model.(4) We simulated land use change of Charchan Oasis in 1990-2013 by combining SD model and CLUE-S model, and validated the result of 2001 and 2013 by using Kappa coefficient. Kappa coefficient of 2001 reached 0.7162, conformance level is substantial. Kappa coefficient of 2013 reached 0.8754, conformance level is almost perfect. We predicted land use status of Charchan Oasis in 2025. Result shows the main transfer direction between land use types will be unused land to grass land or wood land, then grass land or wood land to cultivated land, unused land transfer directly to cultivated land, and cultivated land transfer to construction land. From the view of spatial pattern, cultivated land and construction land will be significantly expanded, unused land will be constantly squeezed. Generally, land use scale, spatial patterns and the competitive relation between land use which simulated by SD model and CLUE-S model are conform to the characteristics of expansion of typical desert oasis.