The Relevance of Surface Roughness Data Qualities in Diagnostic Modeling of Wind Velocity in Complex Terrain: A Case Study from the Snieznik Massif (SW Poland)

Numerical modeling of wind velocity above complex terrain has become a subject of numerous contemporary studies. Regardless of the methodical approach (dynamic or diagnostic), it can be observed that information about surface roughness is indispensable to achieve realistic results. In this context, the current state of GIS and remote sensing development allows access to a number of datasets providing information about various properties of land coverage in a broad spectrum of spatial resolution. Hence, the quality of roughness information may vary depending on the properties of primary land coverage data. As a consequence, the results of the wind velocity modeling are affected by the accuracy and spatial resolution of roughness data. This paper describes further attempts to model wind velocity using the following sources of roughness information: LiDAR data (Digital Surface Model and Digital Terrain Model), database of topographical objects (BDOT10k) and both raster and vector versions of Corine Land Cover 2006 (CLC). The modeling was conducted in WindStation 4.0.2 software which is based on the computational fluid dynamics (CFD) diagnostic solver Canyon. Presented experiment concerns three episodes of relatively strong and constant synoptic forcing: 26 November 2011, 25 May 2012 and 26 May 2012. The modeling was performed in the spatial resolution of 50 and 100 m. Input anemological data were collected during field measurements while the atmosphere boundary layer parameters were derived from the meteorological stations closest to the study area. The model's performance was verified using leave-one-out cross-validation and calculation of error indices such as bias error, root mean square error and index of wind speed. Thus, it was possible to compare results of using roughness datasets of different type and resolution. The study demonstrates that the use of LiDAR-based roughness data may result in an improvement of the model's performance in 100 and 50 m resolution, comparing to CLC and BDOT10k. Furthermore, a slight improvement of these results can be accomplished if the LiDAR-based roughness calculation process includes the variable of prevailing wind direction. Qualities of both CLC and BDOT10k raw datasets (imposed land coverage classes, necessity of the roughness classes assignment) induce relatively high values of the modeled velocity error indices. Hence, these and other similar datasets need to be carefully analyzed (e.g. compared with aerial or satellite imagery) before they are used in the process of roughness length parameterization.