Evaluation of ERA5 reanalysis in convective parameters estimation of vertical wind shear and lifted index using radiosonde data in Iran

The variety of produced variables at the surface and atmospheric pressure levels, appropriate resolution, and global coverage of the ERA5 reanalysis data have led its consideration both in numerous climate research studies and for predicting atmospheric parameters. The initial step in using reanalysis data involves its verification using observational data. Despite the scattered nature of measuring stations, observational data remains a practical dataset for this purpose, particularly in investigating thunderstorms. In this research, we have verified the accuracy of ERA5 reanalysis data in estimating of two convective parameters: the Lifted Index (LI) and Vertical Wind Shear (WVSH). To achieve this, we analyzed radiosonde data from nine stations across Iran (including Tabriz, Mashhad, Tehran, Kermanshah, Isfahan, Ahvaz, Kerman, Shiraz and Zahedan stations) in the period of 1990-2020. Statistical indicators were employed for comparison between the reanalysis data and observational data. Several constraints were applied to the data. For instance, both temperature and dew point profiles should be measured simultaneously. Profiles that terminated below the 6 km above the ground or provided data at fewer than 10 pressure levels were excluded. Additionally, some constrains were utilized to quality control wind and temperature gradients. Specifically: (I) Profiles were removed if the lapse rate in the mid-troposphere exceeded 9 K/km, (II) Profiles were excluded if the lapse rate in the low-troposphere exceeded 11 K/km, (III) Profiles were discarded if VWSH-1000 exceeded 35 m/s, (IV) Profiles were omitted if VWSH-3000 exceeded 45 m/s, and (V) Profiles were removed if VWSH-6000 exceeded 70 m/s. 
The VWSH was calculated across three layers at altitudes of 1000, 3000 and 6000 meters from the surface. Investigations were conducted on daily, monthly, seasonal and long-term time scales. On a monthly scale, the minimum (maximum) root mean square error (RMSE) for VWSH-1000, VWSH-3000, and VWSH-6000 was approximately 3 (8.5), 3.36 (9.84), and 4 (20) m/s, respectively. The results showed that the ERA5 reanalysis data consistently underestimated the value of VWSH-1000 across all stations (except Ahvaz station in recent years). The estimation of VWSH-3000 and VWSH-6000 parameters exhibited both overestimation and underestimation in different months. Notably, the highest error in ERA5 data for VWSH-6000 occurred during January. Across most stations, the largest errors were observed during cold months (particularly for the VWSH-6000 parameter), while the smallest errors occurred during warm months. In conclusion, the results suggest that as the height of the investigated layer increases, the performance of ERA5 in generating the considered VWSH parameters at the stations improves, especially in recent years.
A comparison between reanalysis-LI and observational-LI indicated that the highest (lowest) error occurs during warm (cold) months of the year. Throughout the study period, the reanalysis data produced an error of at least 10 K (at Zahedan station) and up to 15 K (at Tehran station) in LI estimation. Except for Ahvaz station, LI was consistently underestimated across all stations. The monthly mean of reanalysis-LI reflected more unstable conditions, whereas the observed values indicated a more stable atmosphere. Consequently, reanalysis-LI may not be a suitable metric for distinguishing stability and instability in the considered stations. However, in Mashhad and Tehran stations, there is a consistency between the trend of annual average values from reanalysis and observational data. In other stations, this agreement becomes evident in recent years. However, in some stations, the annual average value of reanalysis LI has overcome the observations, while in others, it is the opposite.