Presentations

Abstract

Forests remain one of the most persistent challenges in wind resource assessment modelling, especially in complex terrain. Their impact on wind flow and turbulence is highly heterogeneous, non-linear, and both time and scale-dependent, which makes flow simulations particularly challenging. In practice, forests are often represented in overly simplified forms. Canopy datasets are frequently sparse or restricted to very local scales, while atmospheric dynamics act on kilometer scales. At the same time, computational cost forces compromises between model resolution, long term representativity, and the convergence of performance metrics.    We propose a practical and scientifically rigorous framework for modelling forests in complex terrain. The region of interest is simulated with large eddy simulations (LES) at high resolution to resolve canopy and orography features from high-resolution input data. This fine-scale LES is embedded within a coarser LES, with boundary conditions provided by a mesoscale model driven by ERA5 reanalysis data. Coarse-graining and global datasets ensure spatial continuity where detailed canopy data are unavailable. This nesting approach preserves spatial heterogeneity while allowing cost-effective inflow of realistic atmospheric states. Long-term representativity is ensured through a day selection routine based on uniform sampling of wind direction. Wind speed distributions and seasonality are conserved, which provides representative wind-climate statistics without unnecessary computational burden.    Within the context of this study, simulations were performed for 3 complex sites with forests. The simulations are validated in terms of bias and correlation in key metrics like turbulence intensity, speed-up ratios between measurements, and diurnal and directional wind patterns observed between met masts. Furthermore, the sensitivity of key metrics to canopy representation, day selection, dataset choice etc. are investigated.       This framework provides the industry with clear guidance on when high-resolution modelling of forests is worth the investment, balancing accuracy, representativity, and computational cost.