Assessment of nocturnal low-level jets in ERA5 and ERA-Interim reanalysis

Nocturnal low-Level Jets (NLLJ) are maxima in the vertical profile of the horizontal wind speed in the lowest hundreds of meters of the troposphere. NLLJ impacts have been noted in many research fields and applications, e.g., air traffic, forest fire propagation, aerosol transport, and wind power production. The present study assesses NLLJs in ERA5 reanalysis of ECMWF in comparison to the predecessor product ERA-Interim and a novel observational dataset for Lindenberg (Germany) developed by DWD. The observational reference data is a composite of measurements for the years 2014 to 2018 from a meteorological tower limited to levels below 100 m a.g.l. and from a Doppler Wind Lidar for data above 100 m. The NLLJs are evaluated in three (six) hourly data from ERA5 (ERA-Interim) against the observations by applying an automated detection algorithm for NLLJs. Specifically, the NLLJ frequency of occurrence, core height and wind speed are statistically analysed and compared. The results for 2014 so far show an improvement in the NLLJ occurrence in the most recent reanalysis, with probability of detection (POD) of 49% (33%) and false alarm rate (FAR) of 27% (42%) for ERA5 (ERA-Interim). Both ERA5 and ERA-Interim qualitatively reproduce the month-to-month differences in the NLLJ numbers. We see also an improvement of the NLLJ core wind speed in ERA5 compared to ERA-Interim, with mean differences of +0.5 m/s (+1.6 m/s) for ERA5 (ERA-Interim) relative to the composite data. The core of NLLJs is too high in both reanalyses, with mean differences of 63m (27m) for ERA5 (ERA-Interim). Furthermore, we compared ERA5 against the composite data for all five available years. The POD and FAR are similar to 2014, with no clear period of the year with a better performance. The year with the best reproduction of NLLJs was 2018, with a POD of 52% and FAR of 27%, and March was the best month, with POD 67% of and FAR of 27%. The mean wind speed difference in the NLLJ core assessed across all years was only 0.01 m/s lower in the ERA5 re-analyses compared to the observation, while the mean core height was overestimated by 69 m. Taken together, we see an improvement of the wind profile and the representation of NLLJs in ERA5 compared to ERA-Interim, although further work is needed for representing the NLLJ heights and frequency of occurrence with a higher precision. Ongoing work includes the development of a better understanding of the spatiotemporal differences in the occurrence and properties of NLLJs in Europe and beyond.