Avalanche forecast using numerical weather prediction in Indian Himalaya

The integration of a nearest-neighbours method based avalanche forecast model with a mesoscale weather forecast model (MM5) has been attempted for avalanche forecasting in Indian Himalaya. The MM5 model simulates weather parameters up to day-4 over the entire western Himalaya. The paper describes the methodology of using MM5 model predictions and some empirical relations, to find the probability of avalanche occurrence up to day-4 at a spatial resolution of 5 km by applying the nearest-neighbours method. The nearest-neighbours model uses Euclidean weighted distance metric to find 10 nearest neighbours from the past data in terms of snow and weather parameters. Based on the avalanche occurrences associated with nearest-neighbours, an a priori probability of avalanche occurrence is derived. This approach has been tested for forecasting of avalanches in Chowkibal-Tangdhar road axis in Indian western Himalaya.