Posters

Abstract

Wind resource assessment often relies on reanalysis data or meso-scale models due to challenges and costs linked to direct measurements. Validating these datasets requires careful consideration of the data aggregation method, given the varied spatial and temporal resolutions of synthetic datasets. Additionally, these datasets offer instantaneous samples spatially averaged within grid cells. This study proposes a validation methodology to effectively compare differently resolved modeled data against point measurements. We utilize wind speed outputs from two reanalysis datasets and two WRF simulations to explore a meaningful aggregation method with point measurements. In addition to the well-known ERA5 reanalysis dataset, we used the Copernicus European Regional ReAnalysis (CERRA), developed using ERA5 for lateral boundary conditions and the HARMONIE model for data assimilation, optimized for the European region. CERRA provides outputs on a finer grid (5.5 km x 5.5 km) and a temporal resolution of 3 hours. For meso-scale models, we employ our in-house Weather and Forecasting (WRF) simulation and the New European Wind Atlas (NEWA) dataset. Both models use ERA5 data for boundary conditions, down-scaling to finer spatial (up to 1 km) and temporal (up to 10 mins) resolutions. While other modeled data are familiar in wind energy literature, CERRA is relatively new, warranting a detailed verification in this study. We benchmarked the CERRA wind speeds against ERA5, WRF and NEWA, validating all of them with lidar measurements collected at 5 offshore locations in the Duch part of the North Sea and the FINO3 platform. CERRA outperforms ERA5 in long-term statistics, including Weibull distribution parameters, power spectral densities, wind direction distribution, and average bias. While ERA5 generally underestimates the wind speed with a total bias of 0.15 to 0.55 m/s, CERRA is more accurate with a negligible total bias, when the temporal resolution is increased up to the resolution of ERA5 using forecasts. The performance improvements can be attributed to the smaller grid size of CERRA, which captures events specific to the measurement site. Our study reveals that the CERRA (ERA5) dataset exhibits an improved correlation with wind speed measurements when these measurements are averaged over 5 hours (6-7 hours). For meso-scale models, a 9-hour averaging window provides a better correlation with measurements, surpassing the time scale associated with the horizontal grid, considering the mean wind speed. This suggests that the model's ability to capture finer wind scales may be constrained by assumptions and simplifications in estimating the turbulent terms. This analysis aims to uncover the underlying factors contributing to the optimal window duration. Notably, models incorporating assimilated observations demonstrate an improvement in resolving finer wind scales, evident in the requirement for a smaller optimal averaging window of the measurements. Among meso-scale data, simulations with finer temporal and spatial resolution necessitate a smaller time window for alignment with the measurement. The outcome has the potential to enhance the wind resource assessment calculations and improve the results of long-term extrapolations using the MCP method.