Avalanches are among the most important natural hazards in mountainous regions and forecasting them is extremely challenging due to their pronounced spatial and temporal variability. Existing forecasts are mostly given at the regional level, which limits their applicability to specific microlocations in the field. The purpose of the thesis is to develop an information solution for microlocation forecasting of avalanche risk based on open data and to include the use of artificial intelligence.
The research included a review of national and international literature and an analysis of available open data sources related to avalanche risk, weather conditions and terrain characteristics. On this basis, spatial and temporal data processing was performed, and a forecast model was developed using machine learning methods. The solution is implemented as a web-based information system. This presents forecasts to the user in the form of microlocation risk maps in a web application.
The results show that it is possible to create a model that enables assessment of the relative risk of avalanches at the microlocation level through the appropriate use of open data and artificial intelligence. The developed solution complements existing regional forecasts and provides a more detailed insight into the spatial variability of the risk.
The practical contribution of the work is reflected in the improved accessibility and comprehensibility of avalanche hazard data for users in mountainous areas. The solution has potential utility for ski touring, rescue services and infrastructure managers. Limitations are related to the availability, quality and quantity of input data and the limited range of training samples, and further work may include expanding data sources, upgrading the model and expanding the geographical scope of the forecast.
|
