Dynamics prediction of land use changes using cellular automata and artificial neural network modeling

Abstract

BACKGROUND AND OBJECTIVES: Land use change has become one of the main issues in environmental studies and sustainable development at the global level. Although several studies have explored land use change at the regional level, an in-depth understanding of the patterns and drivers of land use change in Pakpak Bharat Regency, North Sumatra, Indonesia, is still very limited. As of now, no existing predictive model can deliver long-term perspectives on land use change trends within the region. A lack of comprehensive data and analysis can obstruct the creation of effective and sustainable spatial planning policies. Consequently, a focused study on this matter is necessary to create a pertinent scientific framework that can assist in guiding decision-making processes. METHODS: The current study utilized spatial analysis alongside cellular automata and artificial neural network modeling techniques to project and predict land use transformations in Pakpak Bharat Regency in 2030. Geographic information system with the Molusce plugin were utilized in this study. The analysis consisted of two stages, namely land use interpretation and land use projection modeling. To ascertain the primary regional commodity, data was gathered from field studies related to soil and through interviews concerning economic factors. In contrast, secondary data on agroclimate suitability comprised altitude, air temperature data, air humidity, rainfall, wind speed, and duration of sunlight. FINDINGS: Land uses that tended to expand were for plantations/gardens, settlement areas, and Shrubs. Meanwhile, rice fields and mixed vegetation tended to experience a reduction in area over the years. There were fluctuations in forest land use, which saw an increase in 2018 but a subsequent decrease in 2022. Furthermore, the land use prediction for 2030 in Pakpak Bharat Regency showed that land use for forests, rice fields, and Shrubs decreased. Conversely, projections indicated that the use of land for plantations, housing developments, and farming would expand. The widest durian land suitability class was quite suitable (S2), and thus, durian was recommended to be developed to maintain forests, reduce land damage, and for its high economic value. CONCLUSION: Land use changes in the Pakpak Bharat Regency from 2014 to 2022 were relatively slow. Between 2014 and 2018, as well as from 2018 to 2022, the land use for forests experienced fluctuations, while the land use for plantations and settlements consistently rose. The 2022 land use modeling results had an excellent level of accuracy, and thus, the model could be used to predict land use in 2030. Additionally, the results revealed that durian could establish itself as the foremost regional commodity, supported by scientific evidence demonstrating its suitability for farming and profitability.