Analysis of the accuracy of using ERA5 reanalysis data for diagnosis of evaporation ducts in the East China Sea

Abstract

Evaporation ducts are a phenomenon that occurs with extremely high frequency at the boundary between the atmosphere and the ocean. Because they directly affect the propagation of electromagnetic waves, it is necessary to study their various characteristics. Since it is difficult to conduct large-scale observations at sea, many researchers use reanalysis data for this task instead of observation data. However, there have been no studies verifying accuracy of this analysis method for the diagnosis of evaporation ducts. Therefore, in this work, observations of the low-altitude atmospheric refractivity profile were carried out over the East China Sea on board the research vessel Xiangyanghong 18 in April 2021. First, the differences between different evaporation duct models were examined based on the meteorological and hydrological data obtained at different heights. It was concluded that the diagnostic accuracy of the evaporation duct model is low in stable conditions and a low-wind-speed environment. Under the same conditions, the Naval Postgraduate School (NPS) model showed high diagnostic accuracy when compared with other models. Second, Taylor plots were used to verify the accuracy of the reanalysis data and the observation data. It was concluded that the single-parameter precision of the reanalysis data is relatively high, and there were strong correlations with the observation data. Finally, the observation and reanalysis data were used to compare and analyze the false-report rate, the missing-report rate, and the accuracy of the diagnosed evaporation duct height using the NPS model. The false-report rate and missing-report rate were found to be 1.93% and 1.52%, respectively. The average diagnosis deviation was 3.34 m. The Pearson correlation coefficient was found to be close to 1. The results indicate that it is basically feasible to analyze the characteristics of evaporation ducts based on the NPS model using reanalysis data.