Research on ensemble prediction of visibility during the ice cover period of mountain power lines based on WRF-LES-RF

This study proposes a visibility ensemble forecasting method based on WRF-LES-RF for the fog process in the mountainous areas of Guizhou from January 20th to 24th, 2024. Firstly, the WRF model is used to simulate the large-scale meteorological background field, and the large eddy simulation (LES) is nested to capture the local small-scale turbulent motion. Then, three traditional visibility algorithms, SW99, FSL and AFWA, are combined for integrated forecasting. Finally, the random forest (RF) model is introduced to correct the forecast results. The results show that traditional algorithms have obvious limitations in the complex terrain of mountainous areas. However, the random forest model, by learning the complex nonlinear relationships in a large amount of historical data and effectively combining multiple meteorological factors, can predict the rapid changes of visibility in mountainous areas more accurately. This method significantly improves the accuracy of winter visibility forecasts in complex mountainous areas, provides more reliable meteorological services for departments such as transportation and aviation, effectively reduces the adverse effects of low visibility weather on the social economy and residents' lives in mountainous areas, and at the same time helps the power department take measures in advance to prevent the risk of power line icing. It has significant scientific value and application prospects.