Parameter Estimates 1 MLR Model using All Significant Variables 1

73 use transition WW 16, and OO 26. The decreasing distance from Other lands will increase the likelihood of land use transition GO 15, all land use transitions from Settlement, and all land use transition from Other lands. The human environment contributes 15 parameters on the land use change model of Siak District. Most of parameters in human environment have positive coefficients with value which are close to 0.000 for most land use transitions in Siak District, and only distance from road and population density have different coefficients characteristics. The distance from road is the parameter that contributes in decreasing the likelihood of most land use transitions, and only land use transition FS 4, CG 8, CS 9, CO 10, and GC 12 have positive coefficients which are close to 0.000. These mean that the increasing distance from road will decrease the likelihood of most land uses to transform into other land uses, and reversely the closer distance from road will increase its likelihood. The population density has positive coefficients which will contribute the increasing of the likelihood of all land use transitions in Siak District. The increasing population density will increase the likelihood of land use transitions which happen in Siak District. The policy contributes 3 parameters on the final model that are Forestry Spatial Plan at national level, Spatial Plan at province level, and Forestry and Crop Spatial Plan at district level. The Forestry Spatial Plan at national level contributes the effect to the likelihood of most land use transitions negatively, and only likelihood of land use transition WW 16, OG 24, OS 25, and OO 26 which will be increased because of the implementation of its spatial plan. In other word, the Forestry Spatial Plan at national level may drive the land use transitions happen in Siak District which prevent most of land uses to be transformed into other land uses, keep the Wetlands and Other lands in stable condition, and in the same time increase the utilization of bareland as the major land use in Other lands by replanting vegetation and developing the settlement area. The Spatial Plan at province level has little different effect to the likelihood of land use transitions rather than the Forestry Spatial at national level. The Spatial Plan at province level prevent most of land uses to be transformed into other land uses, keep the Grassland, Wetlands and Other lands in stable condition, and in the same 74 time increase the utilization of Other lands by replanting vegetation. However, the Spatial Plan at province level also increases the likelihood of land use transition FG 3, FS 4, and FO 5. The last parameter in policy theme which has significant contribution on the final model is Forestry and Crop Spatial Plan at district level. In general Forestry and Crop Spatial Plan at district level contributes contrastopposite effect rather than two spatial plans that have been discussed. Its spatial plan increase the likelihood of most land use transitions in Siak District, whereas decrease only the likelihood of land use transition FC 2, WW 16, and OO 26. The discussion about parameter estimates has been done in order to reveal the effect of significant parameters to the land use transitions, so that it can support the understanding of the land use change process in Siak District generally. Furthermore, there are some facts which can be learned from the parameter estimates produced. First fact, the MLR model is generalized model that can be applied to model land use change in Siak District which forces all significant parameters determined to every land use transitions that have been detected. Every land use transitions have various parameters coefficients which can be used to describe the direct effects of parameters to the land use transitions that happen in Siak District. Finally, the parameters on natural environment, human environment, and policy drive the land use change in Siak District in different manners.

1.4.2.4 Model Validation 1

st Scenario The model validation has been conducted in order to validate the final model of land use change in Siak District, which has been developed by using MLR model analysis, to the actual condition of land use change 2005 – 2008. The aim of this model validation is to examine the performance of the final model whether the final model could fit the actual spatial data and could also be used for projecting the near future condition of land use change in Siak District or not. This model validation has been conducted by comparing the projection of land use transitions 2005 – 2008 with the actual land use transitions 2005 – 2008. The projection of land use transitions 2005 – 2008 were created by simulating the 75 parameter estimates produced in MLR model analysis and the actual spatial data layers in 2005 into MLR model equation Equation 1 and 2. The parameter estimates in logistic regression contain the coefficients of the parametersvariables β included in the final model, and it summarizes the effect of each parameter. The spatial data layers in 2005 see Appendix 4 which would be functioned as parametersvariables x in the MLR model equation have been prepared by the same procedures when preparing the spatial data layers in 2002 for MLR model analysis. Table 18. MLR Model Equation: Logit Functions and Conditional Probability of Each Land Use Transition Equation 1. Logit Functions Equation 2. Conditional Probability of Land Use Transition The coefficient of the parameters β i in Parameter Estimates and spatial data layersparameters 2005 x i have been simulated on MLR model equation by using raster calculator, so it would produce the conditional probability maps of land use transitions during 2005 – 2008. Two steps of computations have been done in order to simplify the conditional probability maps simulations that were Logit Functions Equation 1 and Conditional Probability of Land Use Transition Equation 2. The computation of coefficient of the parameters β i and spatial data layersparameters 2005 x i on MLR model equation produced 26 conditional probability maps of outcome categories in accordance with number of land use transitions in Siak District. These conditional probability maps show the probability of each land use transition may occur in the research area, with