Climate Multi-Hazard Risk Classification System

This project successfully created an operationalized machine learning system for classifying regional vulnerability to climate hazards. The Climate Multi-Hazard Risk Classification System is a fully automated, explainable, and accessible platform that merges drought and forest fire risk into a single coherent classification framework — unlike most existing systems that treat these hazards separately. All elements of the pipeline were implemented using Python and standard machine learning libraries, making the solution fully reproducible. Physics-based data generation enabled scientifically grounded model training without costly real-world data annotation. Random Forest was selected as the final model after comparison, achieving over 94% test accuracy. The Flask-based web portal ensures that non-programming experts can use the system for research or climate risk decision-making, directly supporting UN SDG 13 on Climate Action. Key limitations include reliance on synthetic data and the current scope of only two hazard types. Future improvements include integrating actual geospatial data from sources such as WorldClim and NASA Earth Observation, expanding hazard coverage, exploring cloud-based deployment, and experimenting with gradient boosting or neural network architectures. In conclusion, the system demonstrates that machine learning can play a meaningful and practical role in multi-hazard climate risk assessment.

[1] I. Sommerville, Software Engineering, 10th ed. Harlow, U.K.: Pearson, 2017. [2] R. S. Pressman and B. R. Maxim, Software Engineering: A Practitioner's Approach, 8th ed. New York, NY, USA: McGraw-Hill, 2016. [3] L. Breiman, "Random Forests," Machine Learning, vol. 45, no. 1, pp. 5–32, 2021. [4] F. Pedregosa et al., "Scikit-learn: Machine Learning in Python," Journal of Machine Learning Research, vol. 12, pp. 2825–2830, 2021. [5] S. Heo, S. Park, and D. K. Lee, "Multi-hazard exposure mapping under climate crisis using random forest algorithm for the Kalimantan Islands, Indonesia," Scientific Reports, vol. 13, no. 13472, 2023. [6] H. R. Pourghasemi et al., "Assessing and mapping multi-hazard risk susceptibility using a machine learning technique," Scientific Reports, vol. 10, no. 3203, 2020. [7] T. G. Nachappa et al., "Multi-hazard exposure mapping using machine learning for the state of Salzburg, Austria," Remote Sensing, vol. 12, no. 17, p. 2757, 2020. [8] N. Khan et al., "Prediction of droughts over Pakistan using machine learning algorithms," Advances in Water Resources, vol. 139, p. 103562, 2020. [9] Y. Piao et al., "Forest fire susceptibility assessment using Google Earth Engine in Gangwon-do, Republic of Korea," Geomatics, Natural Hazards and Risk, vol. 13, no. 1, pp. 432–450, 2022. [10] C. B. Field (Ed.), Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation: Special Report of the IPCC. Cambridge University Press, 2018.