Evaluation of the performance of ERA5 reanalysis data in estimating multiple types of CAPE and CIN convective parameters in upper-air stations in Iran

Estimation of thunderstorm characteristics is important worldwide. Due to scattered nature of upper-Air soundings, reanalysis data is used as another approach. However, using reanalysis data without any evaluation process can lead to increased uncertainty. Iran with its diverse climate conditions, experiences thunderstorms in different parts of the country in different seasons. Estimation of thunderstorm characteristics is important all over the world. Since the stations measuring the vertical profile of meteorological parameters are scattered, meteorologists use reanalysis data as another approach. Using reanalysis data without any evaluation process can increase the possible uncertainty. Iran as a country with various climate conditions suffers from thunderstorms in its different parts and also in different seasons. In this research, around 90,000 sounding measurements were used to evaluate the accuracy of ECMWF Reanalysis v5 (ERA5) in determining all types of the two convective parameters of Convective Available Potential Energy (CAPE) and Convective Inhibition (CIN). The investigation area limits to nine upper-air stations located in various climate regions including dry, coastal, mountainous and urban areas. The stations were in Tabriz, Mashhad, Tehran, Kermanshah, Esfahan, Ahwaz, Kerman, Shiraz and Zahedan. The analysis was done over a 31-yr period (from the beginning of 1990 to the end of 2020). Data measured at both 00:00 and 12:00 UTC were used. Four calculated types of CAPE parameter were (a) CAPE, (b) surface-based convective available potential energy (SB-CAPE), (c) 0–500 m mixed layer convective available potential energy (ML-CAPE) and (d) most-unstable convective available potential energy (MU-CAPE). Four computed types of CIN parameter were (a) CIN, (b) surface-based convective inhibition (SB-CIN), (c) 0–500 m mixed-layer convective inhibition (ML-CIN) and (d) most-unstable convective inhibition (MU-CIN). The main difference between various types of each convective parameter is referred to the focused parcel. The analysis was done using the statistical indices of correlation coefficient (R), mean error (ME), absolute mean error (AME) and root mean square error (RMSE). To filter incomplete and unreal observational profiles some criteria were imposed on the observational data to quality control them. The criteria were as (a) both profiles of temperature and dew point temperature should be measured, (b) the sounding should pass the 6-km height above the surface, (c) the profiles should contain measurements at more than 10 pressure levels, (d) lapse rate in mid-troposphere should be less than 9 K/km, and (e) lapse rate in low-troposphere should be less than 11 K/km. Some criteria were imposed after the calculation of the convective parameters including (a) MU-CAPE values should be less than 8000 J/kg, (b) ML-CAPE values should be less than 6000 J/kg, and (c) CIN values should be more than -1000 J/kg. The results showed that the two parameters of ML-CAPE and ML-CIN in most stations produced the highest values of correlation coefficient for calculated convective parameters using observational and reanalysis data. Based on ME and MAE indices, ML-CAPE, ML-CIN, and SB-CIN parameters generated the least error in most stations. The RMSE index showed that ML-CAPE and ML-CIN produced the lowest values of error in most stations. In a conclusion, the obtained results indicated that the two convective parameters of CAPE and CIN calculated using air mass in the mixed layer (ML-CAPE and ML-CIN) from ERA5 reanalysis data provided the most reliable values over most stations compared with the observational data. Hence, it is suggested that the last mentioned type for the two studied convective parameters be considered for future research studies, especially in simulation of thunderstorms.