Abstract The Kathmandu Valley, Nepal, situated within a seismically active Himalayan belt, faces high earthquake risk due to its proximity to major thrust faults and dense population. While previous studies have focused on tectonic mapping and probabilistic hazard assessment, the application of spatial analysis techniques like Kernel Density Estimation (KDE) for visualizing seismic hotspots remains limited in Nepal. A catalog of 6532 earthquakes (1970 to 2025) from USGS, ISC, and IRIS was declustered to 1337 mainshocks using the Gruenthal algorithm in ZMAP. Three kernel functions, Gaussian, Epanechnikov, and Biweight with a bandwidth of 0.05° 0.10°, and 0.15°, were tested to model spatial epicenter density, and their performance was evaluated using Likelihood Cross Validation. Results show strong spatial clustering with the Gaussian kernel yielding the best.A comprehensive sensitivity analysis examining bandwidth variation, grid resolution (20 × 20, 25 × 25, 50 × 50), and edge effects (10 km buffer) confirmed the robustness of hotspot identification, with hotspot displacement 12) in the eastern and southeastern parts of the Valley, particularly around Bhaktapur and Lalitpur, which align closely with the traces of the Main Frontal Thrust and Main Boundary Thrust. In contrast, cold spots were located in the western and northern fringes. The findings confirm that KDE based heatmaps enhanced by sensitivity analysis and magnitude weighting, identify persistent seismic hotspots, providing a strong spatial tool for seismic risk assessment and urban planning in the Kathmandu Valley.
Dibyashree Poudyal Lohani (Mon,) studied this question.