A comparative study of Quantum and classical  Support Vector Machines in mapping forest fire severity

Our work presents an in-depth evaluation of machine learning techniques applied to environmental remote sensing, through a comparative study of classical versus quantum Support Vector Machines for mapping wildfire severity. As environmental monitoring faces growing challenges associated with large-scale spatial data, traditional algorithms often struggle to identify the complex non-linear relationships linked to geographic coordinates. To address this, our study examines the potential of quantum machine learning by embedding spatial attributes into high-dimensional quantum state spaces using advanced kernel techniques. The primary objective is to compare the performance of classical and quantum machine learning models using specialized Python tools, directly contrasting traditional and quantum hyperparameters to evaluate performance metrics, training stability, and spatial prediction efficiency. Furthermore, the project applies these predictive models to the spatial domain to generate a high-resolution, three-dimensional map of wildfire severity across the affected landscapes. The study results show that the use of quantum hyperparameters with support vector machines enables better capture of complex, non-linear environmental correlations within spatial datasets, thereby offering superior classification performance for severe burn areas. Specifically, the proposed quantum machine learning model achieved an overall classification accuracy of 96.8% and an F1-score of 0.95, outperforming the classical Support Vector Machine baseline which yielded an overall accuracy of 93.5 % and an F1-score of 0.91 under identical environmental feature parameters.