On the diurnal cycle and variability of winds in the lower planetary boundary layer: evaluation of regional reanalyses and hindcasts


To accurately calculate the impact of renewables on power production in complex electric power grids, high-resolution and ideally seamless data within the planetary boundary layer are required. Therefore, the quality of different regional reanalyses and hindcasts is evaluated with respect to the representation of the planetery boundary layer and related sub-daily processes. On the one hand, high resolution regional reanalysis from the UERRA (UE-SMHI, UE-UKMO) and a similar project (COSMO-REA6) are considered. On the other hand, two hindcasts based on the COSMO-REA6 configuration are included in this study, i.e. a simulation with perfect boundaries and a simulation additionally utilizing spectral nudging. The focus of the evaluation is on measurements at four flux towers that are not part of any assimilation procedure. In this paper, we will show that the model’s quality depends on both the complete model system – assimilation method, resolution and physical parameterization – as well as on the performance measure. The daily cycle is best depicted by the hindcasts and even COSMO-REA6 hardly introduces spurious variability. UE-SMHI (3D-Var) suffers from spin-up in particular visible at the elevated levels, whereas the spin-up is damped in UE-UKMO (4D-Var). Investigation of atmospheric stability reveals that diurnal variation of stratification is for the most part well reproduced, but strong deficits were found for all COSMO simulations in reproducing strong stratification and corresponding wind speed gradients. Moreover, an overestimation of superadiabatic lapse rates and corresponding overly weak turbulent mixing is found for UE-UKMO. Furthermore, a combination of ramp statistics and contingency tables is utilized to detect a clear advantage of sophisticated assimilation systems over hindcasts. The evaluation framework presented underpins the importance of ramp statistics and vertical measurement profiles, especially with respect to assessing long-term simulations.
QR Code: Link to publication