Statistical Analysis Land Use Scenario Development by using CLUE-S Model

25 Based on the findings about land use changes in year 1991 and 2009 above, it can be noted that urban settlement area experienced the most significant development followed by forest and grassland. However estates tend to decrease and river is stable during the 1991-2009 period.

2.4.1.3. The Trends of Land Use Change during 1991-2009

According to the change detection for the land use classification year 1991 and 2009, it can be seen that settlement, forest and grassland area are increasing over time, whereas estate tend to decrease and river remains constant. The trends of land use changes of the area during 1991-2009 are presented in following figure. Figure 6. Trends of land use change year 1991-2009

2.4.2. Analysis of driving factors of land use changes

2.4.2.1. Logistic regression results

The results of logistic regression between land use and independent variables are presented in this section. Each land use has independent variables or driving factors that influence to its pattern. In logistic regression analysis, five classes of land use are included in the regression calculation. The selection of the significant and non-significant independent variables is based on a enter procedure see 2.3.3.3.. The variables, which have coefficient values below 0.01 significant thresholds, are categorized as significant and the 26 variables above 0.02 will be classified as non-significant and automatically removed in the calculation process. All of significant variables are automatically selected in the results of enter procedure and will be used in the calculation of land use probability. Practically, enter procedure in logistic regression is done by putting all independent variable in the process at the same time in the iteration process. The result can be seen in the Table 10. Table 10. Results of logistic regression between land use pattern and driving factors Variable Water Grassland Estate Settlement Forest Total number of observed grid cell 88918 88918 88918 88918 88918 POPDEN 2.404 -5.964 -1.374 1.081 -2.913 ELEVS -27.566 .968 1.056 -1.060 -8.369 ROADS -22.757 3.714 -.355 -2.106 .949 PUBFAC -6.285 -1.100 -8.848 8.679 EDUFAC -.570 -6.851 -5.745 20.718 HEAFAC -11.171 4.251 -2.487 .748 COECO 9.398 -.821 1.596 -1.022 CONSTANT 1.892 -2.353 1.983 .303 -8.541 EXP � 6.633 0.095 7.264 1.354 5120.462 ROC 0.903 0.701 0.780 0.813 0.994 Enter procedure; Significant at: 0.01 entry level and 0.02 removal Has significant value above 0.02 See Appendix 5.; Not statistically significant value above 0.02

2.4.2.2. Logistic regression interpretation

The number of driving factors used in this regression analysis is 7 variables. Each driving factor has a different effect on every type of land use. The effect of each driving factor is indicated by the coefficient β in logistic regression result, which presents how much variance from the use of land that can be explained by the driving factors. A large positive β value indicates a strong positive relationship between the independent and dependent variable land use change, while a large negative β value indicates a strong negative correlation with land use change. Besides the effect of driving factors, the results of logistic regression can be used to indicate which driving factor that has the biggest influence to the land use change. How big the influence of each driving factor is indicated in exponential β value that shows the effect of increasing one unit of driving factor to the land use change Borzacchiello, et al 2010, Gyawali et al, 2004. Based on the findings of