Interactive Simulation Tool for Spatiotemporal Data of Climate-related Fires over Borneo
Furqon H. Muttaqien, Taufiq Wirahman, Ayu Shabrina, Arnida L. Latifah · 2023
Hazardous forest fires over Borneo in 2019 should be a concern that natural climate-related disasters have been increasing; they can be more frequent and extreme. Research about forest fire prediction and its impact has been abundant. However, there still needs to be more people’s awareness of the hazard and risks. This study introduces an interactive simulation tool to help society and scientists from other fields understand better the risk of climate-related fire in Borneo. The interactive simulator of forest fire prediction is created as a web-based application using the Flask framework and Bokeh library in Python. As the first development, it visualizes the historical spatial and temporal forest and land fire data over Borneo, computed by machine learning methods, namely Random Forest, Decision Trees, and Support Vector Machine. The prediction results show that the three models similarly predict burned areas. Meanwhile, Random Forest predicts the carbon emission slightly better than other models, with 4.751 and 0.008 for MAE and NRMSE, respectively. It also gives interactive features that the user can play to estimate the fires, given climate and environmental conditions.