Evaluation of NWP Models Using Scanning Lidar Measurements in Complex Terrain during the WFIP2 Experiment: Lessons Learned

Numerical weather prediction (NWP) forecast models are widely used in wind energy (WE) operations as foundational or base models for wind resource assessment and wind-power forecasting. During the WFIP2 2015-2018 experiment, a significant effort was made to improve model forecasts in complex terrain by improving model physics, parameterization schemes, and horizontal grid resolution. These improvements were evaluated by measurements from various remote sensing instruments including scanning Doppler lidars, 915-MHz wind-profiling radars, sodars, and wind-profiling lidars. This paper presents results of model validation by scanning Doppler-lidar measurements at three sites, to assess model skill in capturing the spatial and temporal variability of wind-flow profiles and to quantify model errors in forecasting winds in the first 1 km above sea level (ASL), with particular emphasis on turbine hub-height wind speed. In addition to seasonal variability, model errors are evaluated for periods of interesting meteorological events, such as cold pools and gap flows. Characterizing model errors for frequently occurring phenomena generated in complex terrain by regular wind-flow systems in the lower troposphere is shown to be an important aspect of both diagnosis of model problems and improvement of model skill.

The paper discusses how model validation results depend on specific wind-flow conditions at the location of each research site and on the uncertainties of measurements by each instrument type.