Multi-objective based Cellular Automata-Markov chain modeling for landuse change analysis in Kuala, Langat, Malaysia

Analysis of land use and land cover change is a complex task on account of tensions between land classes where any land category has a series of specific needs for development. This research addresses resolution of a multi-objective land development problem in Kuala Langat district, Malaysia under an integrated model of Cellular Automata-Markov chain (CA-Markov) towards projecting land development for the year 2020. According to the current land use dynamics, four conflicting objectives are identified including urban and urban related development, oil palm development, agriculture development, and forest development. Four groups of evaluation criteria are developed that define the main driving forces of change in each objective. The Analytical Hierarchy Process (AHP) is adopted to assign a weight to each evaluation criteria based on the expert opinions and judgments. Multi-Criteria Evaluation (MCE) technique is used to conduct four disparate suitability analyses. A Multi-Objective Land Allocation (MOLA) analysis is then adopted to analyze four different outcomes of MCE. Simultaneously, Markov chain analysis is conducted to compute the quantitative transitions of each land category between 1997 and 2002 to project land change of the year 2008. The projected 2008 is then validated by real map of the year 2008 based on three validation methods. The overall agreements based on three approaches of quantity disagreement and allocation disagreement, Cramer’s V, and Kappa are 79% (16% allocation disagreement and 5% quantity disagreement), 78%, and 77% (due to location and quantity) respectively. However, the higher accuracy achievementM requires model calibration to eliminate the deviations of projection. To increase the agreement of projection, this research initiates a method for calibration of CAMarkov land change projection. The proposed method is based on integration of cross-tabulation analysis and Markov chain analysis of observed and projected land use data. The method is successfully examined in a specific landscape and the time step. Model validation after calibration process reveals a meaningful increase in the agreement of projected versus observed land use data. The quantity disagreement and allocation disagreement approach measures 15% increase in overall agreement, Cramer’s V measures 13% increase in agreement, and Kappa measures 6% increase in overall agreement due to location and quantity. Finally, the major signals of systematic transition of each land category including net change, swap, gross gain,and gross loss are extracted to compare land transformation process over time. The results demonstrate the high tendency of forest category to systematically lose to ‘other’ land category and the high tendency of ‘other’ category to avoid systematic gain from oil palm category by the year 2020. In the same time, results show the high disinclination of forest category to systematically lose to oil palm category.

Preview PDF FK 2013 127RR.pdf Download (1MB) | Preview

Actions (login required)

View Item