Land Use Change Schemes

57 Figure 21. Land Use Change Scheme 2002 – 2005 in Siak District Figure 22. Land Use Change Scheme 2005 – 2008 in Siak District The land use change scheme 2002 – 2005 shows that all land use categories tended to not transform into other land uses stable condition with high probabilities. Wetlands, which are assumed in stable condition, and Settlements, 58 are the land use categories which have the highest probability for being in stable condition. The probability of Forest land, Cropland, and Grassland tended to be in stable condition were 70, 62, and 47 respectively. The land use change scheme 2002 – 2005 also show that the three dominant land use categories in Siak District, which are Forest land, Cropland, and Grassland, transformed each other which constructed the triangle of major land use transitions with reciprocal transitions . Forest land transformed into Cropland and Grassland, and it contributed deforestation in Siak District with total probability 33 of Forest land is deforested. Reversely, only 9 of Cropland and 8 of Grassland transformed into Forest land reforestation during 2002 – 2005. In the same period, Other lands also contributed reforestation with probability 12 of Other lands were reforested. During 2002 – 2005, Other lands tended to contribute the area expansion of the three dominant land use categories with significant probabilities, which majority of Other lands transformed into Cropland and Grassland with probability 50 and 28 respectively. The land use change scheme 2005 – 2008 shows the similar scheme with the scheme 2002 – 2005, that all land use categories tended to be in stable condition with high probabilities, which Wetlands and Settlements, are the land use categories which have the highest probability for being in stable condition. However, the probability of Forest land and Grassland to be in stable condition slightly decreased which became 69 and 44 respectively, whereas the probability of Cropland to be in stable condition increased dramatically which became 72. The triangle of major land use transitions with reciprocal transitions among Forest land, Cropland, and Grassland did not happen during 2005 – 2008. The transformations from Forest land to Cropland and Grassland deforestation happen in one-way transitions. The deforestation probability increased which became 37 of Forest land was deforested. Reversely, the transformations from Cropland and Grassland to Forest land reforestation did not count as major land use transitions during 2005 – 2008. Small portion of Cropland and Grassland instead transformed into Other lands during 2005 – 2008, and reversely the majority of Other lands transformed into Cropland and Grassland with probability 52 and 37 respectively. 59 The illustrations and the discussion about land use change scheme show that the land use change schemes in Siak District were different during two time periods, which means the major land use transitions also changed during 2002 – 2005 and 2005 – 2008. The situation that should be highlighted and considered by Siak District is the increasing of deforestation probability as one of major land use transitions during 2002 – 2008. This situation was getting bad, since the reforestation was not done significantly and was not also visible as major land use transitions according to the land use change scheme 2005 – 2008. If this situation continues, it is possible that the Forest land in Siak District, which the majority is Peatland Forest, will continues to decline and probably in the future will be exhausted and will be replaced by Cropland. This situation is also motivated by the increasing probability of Forest land and other land uses to transform into Cropland and probability of Cropland to be in stable condition. Figure 23. Crop Plantation Cropland Category in Siak District

4.4 Land Use Change Model and Significant Variables

The land use change modeling by using Multinomial Logistic Regression MLR has been conducted in this research in order to determine the significant variables which might affect the land use transitions in the research site, and also to find the adequate land use change model by considering the significant variables which have been determined. The MLR model has been chosen for modeling land use change in this research because its model can accommodate the categorical data of land use transitions as a dependent variable, and the relevant driving factors which could be a mixture of binary, continuous and categorical variables as independent variables. Furthermore, the logistic regression has no assumptions about the distributions of the independent variables Xie et al. 2005; 60 Tabachnick and Fidell 2006. The independent variables do not have to be normally distributed and linearly related which usually follow the land use change cases, so that the driving factors which involve in the model do not require data normalization. The land use change modeling activity has been started with data preparation for MLR modeling, and then continued with the MLR model analysis, and the last was model validation in order to examine the performance of MLR model that has been developed.

4.4.1 Data Preparation

Data layer preparation is the most fundamental process in order to condition the spatial and non-spatial data into statistical MLR model, and at the same time considering the MLR model to be applied into spatial manner, since the dependent and independent variables for land use change modeling may have different types of data binary, continuous, or categorical and spatial resolution. In order to accommodate the nature of spatial analysismodeling which is applied in this research, all variables involving in the model should be conditioned to have the same spatial characteristics, such as the type of spatial data raster or vector, spatial extent, and spatial resolution. According to the methods which have been developed and described in Chapter 3, the land use change modeling in this research would be processed on ERDAS Imagine raster data with 100 x 100 meters spatial resolution, and cover the Siak District’s spatial extent. Thus, either spatial or non-spatial data should be processed to meet its spatial characteristics. Here, data categorization, conversion, and spatial analysis techniques should be applied, so that it would produce spatial data layers which have the same spatial characteristics and are ready to be used on MLR modeling. The dependent variable which would be involved in MLR model analysis was the categories of land use transitions 2002 – 2005, and the independent variables were the relevant driving factors which have been collected in Field Data Collection activity. The spatial data layers of the independent variables were divided into three themes that were natural environment, human environment, and policy in order to get better understanding of the process of land use change and also reveal which variables having the significant role in driving the land use