by Sukmo Pinuji (Indonesia)
Understanding the characteristics of agricultural expansion, particularly oil palm, is important to study its impact on the world’s land. Land use modelling is a tool that can be used to help to understand the key process of oil palm expansion, to assess the current state, the drivers, the processes and the impact of oil palm expansion. Using spatial datasets from different sources, this research models the process of land use change using IDRISI Land Change Modeller to understand the follow up oil palm land use after deforestation events in Indonesia, as well as to predict where the deforestation will likely to occur due to the process of oil palm expansions. Artificial Neural Network method was used to build sub-models during the observation period for the year of 2000 – 2006, while Markov Chain Method was used to predict future land use in 2009.
The results shows that the actual trend of deforestation and oil palm expansion during observation and prediction periods was change significantly, thus affecting the model accuracy in predicting changes, especially for the class that dominated the change after deforestation, such as other class in Riau and oil palm class in West Kalimantan and East Kalimantan. In Riau, other class were overestimated 47.63% more than the actual. In East Kalimantan and West Kalimantan was predicted by the model 32.86% and 54.57% less than the actual increases. Apart from those two classes, the following land use after deforestation events were predicted better by the model, and only gives small differences with the actual changes. We also found that distance from the existing oil palm plantations was the most significant variable in the process of deforestation and oil palm expansion. We conclude that Artificial Neural Network and Markov Chain Method are useful to model and to predict land use change following deforestation and agriculture expansion, but failed to account external factors of deforestation and agriculture expansion drivers that were not included during the modelling process.
Keywords: deforestation; oil palm; land use change modelling; Artificial Neural Network; Land Change Modeller; agriculture expansion.