A Rapidly Updated Numerical Model for Power Grid Rainstorm Forecast

Disaster forecast is the first line of defense for power grid disaster prevention and reduction. In this study, we firstly built a numerical weather forecast model with a resolution of 3 km in south China, continuously update and iterate the grid precipitation forecast for the next 24/96 hours by rolling the forecast every three hours. The model’s performance is evaluated with rainstorm on 21th July 2021 in Henan province, which cause major disasters in the Henan power grid. The forecast results for Henan indicate that the model can predict the spatial distribution characteristics of heavy precipitation 48 hours in advance, but the maximum precipitation intensity is relatively small and the precipitation center is located to the west. In contrast, forecasts made 24 hours in advance can better predict the location of precipitation centers, enhance precipitation intensity, and be more consistent with observations. The forecast results of the model is also compare with the ECMWF data, which is a state-of-art model for weather forecast. The precipitation prediction data of this model can be used to further drive the geological disaster and flood disaster model to carry out rainstorm secondary disaster prediction, and the disaster prediction can be iteratively updated based on the latest precipitation prediction results, which can buy valuable time for power grid disaster prevention.