Scenario 3 Scenario-based land use modeling by using CLUE-S model

41 There are also several limitations related to the method and findings of this research, including justification in parameter settings and land use conversion elasticity that only based on local knowledge and interview with local community. In the previous sub-chapter, it can be seen that in order to determine the temporal dynamic of the simulation by using reversibility of land use changes, three different decision rules are used. To produce equal conversion behavior, the coefficient of elasticity is set in dimensionless range 0-1 to determine level of reversibility. The elasticity of grassland area is 0.4, estate is 0.6, and forest is 0.4. Since there is no exact method in determining the elasticity level, empirical assessment of land use behavior and multi-temporal observation then will be a major contribution in the improvement of the model. Another issue is related to the reliability of driving factors involved in this model. The implicit assumption in such an approach is that the driving factors are stable during the modeling time period and influence the dynamic of land use in the area. This assumption ignores the possibility of driving factors to change and their effects to the land use change. For example the development of local road and additional facilities that may respond to changes in land use. The CLUE-S framework is capable in involving dynamic driving factors as additional variables to assess future land use pattern, therefore, if data about new dynamic driving factors are available, it could therefore be useful to improve the utility of scenarios.

2.5. Conclusion

It was found that land use changes in this area were mainly dominated by expansion of settlement area with the annual rate of change 13 during 1991-2009, followed by forest 5.05 and grassland 4.72 in the same period. Estate is decrease in period 1991-2009 with annual rate of change -6.21 . The trend of land use change in 1991-2009 showed this area faces the expansion of settlement, forest and grassland area and the decrease of estate. Based on the findings, it can be concluded that settlement area were significantly increase during the period 1991- 2009 and causing the pressures in the study area. 42 The selection of driving factors that has significant effect to land use change was conducted by using logistic regression analysis. In case of forest, the quantitative analysis of driving factors of land use change by using logistic regression showed that the driving factors that significantly influence forest area were population density, elevation, distance to main road, distance to public facility and distance to education facility. Based on the calculation results, it can be concluded that the most significant driving factor of forest area was distance to education facility and followed by distance to public facility, elevation, distance to main road, and population density. The relationship between driving factors and land use change showed that distance to primary road, distance to public facility and education facility have positive effect to forest area to change which means that the higher the value of these factors, the higher the probability of land use to change. In contrast, population density, and elevation have negative effect that implies that the higher the value of these factors, the decrease the probability of land uses to change. The goodness of the statistical measurement revealed that ROC values for urban water area, grassland area, estate area, settlement area and forest area were 0.903, 0.701, 0.780, 0.813 and 0.994, which indicated that the probability of land uses built from these models were capable to represent land use changes and empirical analysis by using logistic regression method was satisfactory to examine the relationship between driving factors and land use change in study area. In order to improve the utility of scenario regarding to the deficiency of data and methods, this research has conducted several approaches, includes reducing the uncertainty of data classification and combining qualitative and quantitative approach in driving factors selection.