Atmospheric stability from numerical weather prediction models and microwave radiometer observations for onshore and offshore wind energy applications

Atmospheric stability controls the evolution of wind turbine wakes and thus the yield and performance of wind parks. For estimations of wind park power output and for improving analyses of wind park wakes, crucial parameters were found to be profiles of atmospheric temperature and stability metrics. Atmospheric temperature profiles are available from numerical weather prediction (NWP) models or are measured in situ by balloon-borne sensors but can also be estimated from the ground using radiometric observations. This paper reviews the stability metrics useful for monitoring wind park performances and provides a quantitative assessment of the value of NWP model data for estimating these metrics. This paper also extends previous work, quantifying the performances of microwave radiometer (MWR) observations to estimate stability metrics from surface-based observations in three climatological conditions (marine, continental, and polar) and with different instrument types, situated either on land or in the ocean. Two NWP systems (DOWA and NEWA) have been evaluated against temperature profiles measured by offshore met masts in the 30-100 m layer from the surface. The systematic differences are ∼0.3-0.5 K, with no clear dependence on the stability class. Conversely, both models show larger random differences in stable compared to unstable conditions. Root-mean-square (rms) differences were within 1 K for DOWA (Dutch Offshore Wind Atlas), while they exceeded 2 K for NEWA (New European Wind Atlas) in very stable conditions. For temperature gradients in the 50-100 m vertical layer, the mean absolute error (MAE) was ∼3.4-4.2 K km-1, with 5.8-8.4 K km-1 rms and 0.7-0.8 correlation. From the six datasets of MWR and radiosonde observations considered here, temperature profiles mostly agree within ∼ 0.5 K near the surface, increasing to ∼ 1.5 K at 2km. Substantial differences are found between MWR performances in retrieving temperature and potential temperature gradients (50-300 m) onshore and offshore. Onshore, potential temperature gradients agree with 2.1-3.4 K km-1 MAE and 0.7-0.9 correlation. Offshore, both MAE (0.9-1.9 K km-1) and correlation (0.3-0.4) are relatively lower, although performances tend to improve using elevation scanning retrievals. Considering all the datasets, the reported MAE values range from 0.9-3.4 K km-1 while rms values range from 1.2-5.1 K km-1. Thus, the uncertainty of MWR for temperature and potential temperature gradients in the 50-300 m vertical layer is ∼ 0.5-4.3 K km-1. The relatively lower performances offshore may be attributed to the training of the inversion method, which may under-represent the peculiar offshore conditions, and the ship movements, which can impact low-elevation observations. These considerations suggest that appropriate dedicated training and elevation scanning with ship movement compensation may be required for the MWR to better catch potential temperature gradients typical of offshore conditions.