How do multiple kernel functions in machine learning algorithms improve precision in flood probability mapping?

Abstract

With climate change, hydro-climatic hazards, such as floods in the Himalayas regions, are expected to worsen, thus likely to overwhelm humans and socioeconomic system. Precisely, the Koshi River basin (KRB) is often impacted by floods over time. However, studies on estimating and predicting floods are still scarce in this basin. This study aims at developing a flood probability map using machine learning algorithms (MLAs): Gaussian process regression (GPR) and support vector machine (SVM) with multiple kernel functions including Pearson VII function kernel (PUK), polynomial, normalized poly kernel, and radial basis kernel function (RBF). Historical flood locations from available (topography, hydrogeology, and environmental) datasets were further considered to build a flood probability model. Two datasets were carefully chosen to measure the feasibility and robustness of MLAs: the training dataset (flood locations between 2010 and 2019) and the testing dataset (flood locations of 2020) with thirteen flood influencing factors. Validation of the MLAs was performed with statistical indices such as the coefficient of determination (r2: 0.546 –.995), mean absolute error (MAE: 0.009 –373), root mean square error (RMSE: 0.051–0.466), relative absolute error (RAE: 1.81–8.55%), and root-relative square error (RRSE: 10.19–91.00%). Results showed that the SVM-Pearson VII kernel (PUK) yielded better prediction than other algorithms. The resultant map from SVM-PUK revealed that 27.99% area with low, 39.91% area with medium, 31.00% with high, and 1.10% area with very high probabilities of flooding in the study area. The flood probability map, derived in this study, could add great value to the effort of flood risk mitigation and planning processes in KRB.