Can ERA5 reanalysis data characterize the pre-storm environment?

Abstract

The atmospheric environment before the onset of rainfall, frequently obtained from reanalysis, oftentimes varies dramatically at time scales ranging from minute to hour. Nevertheless, the extent to which hourly reanalysis data can capture the pre-rainfall and pre-storm signals remains largely unknown. To this end, three warm-season (April to September of 2015–2017) records of high-resolution radiosonde measurements and the Fifth Generation Global Climate Atmosphere Reanalysis (ERA5) reanalysis data throughout the whole of China are used to derive the pre-storm environmental parameters, including convective available potential energy (CAPE), convective inhibition (CIN), precipitable water (PW), K index, LI index, storm-relative helicity (SRH) and vertical wind shear (WSR). For rainfall and storm events, we separately conducted a comprehensive comparative analysis of ERA5 against soundings at 120 radiosonde stations in China obtained 2 h before the rainfall onset. The CAPE before the onset of storm events is much greater than that for rainfall events, especially over the central and southern China, and the association between CAPE and rain rate for the storm events is much stronger than that for the rainfall events. Pre-rainfall environmental analysis suggests that ERA5 reanalysis tends to overestimate the thermodynamic parameters including CAPE, CIN, PW, and LI index, compared with soundings, and this overestimation exhibits a “west-high east-low” spatial pattern. In contrast, ERA5 reanalysis tends to underestimate the dynamic parameters like WSR and SRH. By comparison, the correlation coefficients between soundings and ERA5 for most pre-storm environmental parameters are lower than those of all rainfall events, suggesting ERA5 has a relatively better performance in characterizing pre-rainfall environment compared with in characterizing pre-storm environment. Besides, the pre-storm parameter discrepancy pattern may be more spatially heterogenous, compared with pre-rainfall counterpart. We also try to explore the evolution of the pre-rainfall atmospheric parameters with lead time that refers to the period between 20 min after the sounding balloon releases and the rainfall onset. Most pre-rainfall environmental parameters from ERA5 undergo more sudden changes within 10 min of the rainfall onset than those from soundings. Overall, ERA5 reanalysis cannot accurately capture the atmospheric convective environment before rainfall in most cases, especially in western China where the weather stations are scarce. Therefore, the findings highlight the need of integrating reanalysis with sounding measurements to better characterize pre-rainfall and pre-storm environment.