Modeling Future Carbon Stock Predictions Based on Land Use

The considerable influence of extensive land use change on the increasing levels of carbon emissions has significant implications for the occurrence of a multitude of disasters. The objective of this research is to develop a predictive model of future carbon stocks based on land use type. The data set includes land use maps from 2014, 2018, and 2022, obtained through visual interpretation of Pleiades data and associated driving variables, including socio-economic, locational, physical, land, and spatial planning factors. To predict land use in relation to future carbon stock values, the Multilayer Perceptron Neural Network Markov Chain (MLPNN-MC) algorithm was employed. Research related to this modeling is capable of producing an accuracy rate of 98%. The results of the prediction demonstrate that by 2034, there will be a reduction in the area of land used with high to low carbon stock, with a decrease of 153.2 ha, which equates to a reduction in carbon stock of 9,050 tonnes C/ha. To reduce carbon emissions, it is essential to implement policies that regulate land use change, optimize forest management, and conserve mangrove ecosystems. The monitoring and prediction of future carbon stocks plays a pivotal role in climate change mitigation, enabling more targeted and measurable actions to be taken.