Short-Term Forecasting of Evaporation Ducts Based on GraphCastGFS

Evaporation ducts have an important impact on the propagation of electromagnetic waves near the sea surface. The short-term forecasting of regional evaporation ducts has important practical significance for radar, communication, etc. The development of machine learning technology has resulted in a wide range of applications in the field of meteorological forecasting. Recently, the National Centers for Environmental Prediction (NCEP) released the GraphCast Global Forecast System (GraphCastGFS), which is a new meteorological forecasting product based on the machine learning algorithm GraphCast. This study carried out a 168 h short-term forecast for the northern part of the South China Sea (SCS), based on the GraphCastGFS forecasting product and the NAVSLaM model. Taking ERA5 data as the benchmark value, a root-mean-square error (RMSE) analysis of the evaporation duct height (EDH) forecast results based on GraphCastGFS, and those of GFS, was performed. The results showed that the RMSE of the EDH forecast result based on GraphCastGFS was 3.0 m, and that of GFS was 3.4 m. The short-term forecasting accuracy of evaporation ducts based on GraphCastGFS was found to be marginally better than that of GFS, and the rapid data generation speed of GraphCastGFS further suggests its substantial potential in the field of evaporation duct forecasting.