Infometrics - Consulting Engineers
Evaluation of wind flow computational models using multi-resolution remote sensing datasets in a high complexity terrain domain.
SPIE Remote Sensing 2012 - Conference 8538A Earth Resources and Environmental Remote Sensing/GIS


Wind flow estimation over complex terrain is a crucial procedure for several applications such as prediction of wind energy resources, pollution dispersal estimation and bridge design. Numerous flow models of different type and complexity have been developed in order to reconstruct a steady-state wind field based on initial experimental data. These computational models are divided into two main categories: a. linearized models and b. mass-consistent models. Linearized models are based on simplified steady-state solutions of the Navier-Stokes equations for boundary layer wind profile. On the other hand, mass-consistent models are based on adjustment of interpolated field of wind flows to satisfy mass conservation equations. In both cases the accuracy of wind flow estimation strongly depends on the complexity of terrain domain and consequently on resolution and accuracy of topographical input data, that describe terrain characteristics. In this work we evaluate three wind flow computational models (MS-3DJH/3R, WaSP and 3D-RANS) in a complex terrain domain in Southern Greece using three remote sensing datasets of different spatial resolution. The study area is a mountainous region with frequent changes of terrain slope and land cover. The evaluation is based on a pair-wise intercomparison of results derived from two meteorological stations. The first anemometer mast is used as a predictor and the second is used for evaluation of wind speed predictions. For the simulation of the wind field, terrain characteristics are described by a terrain roughness map and by a digital elevation model. Object-oriented segmentation techniques have been applied to Landsat, ASTER and IKONOS images in order to derive terrain roughness maps with spatial resolution of 30 meter, 15 meter and 1 meter respectively. Furthermore, two DEMs with global coverage (SRTM and ASTER GDEM) and a high-resolution DEM, derived from digital photogrammetric techniques, paired with corresponding terrain roughness maps produce three different resolution input data to wind flow computational model. The influence of the topographical input data on the accuracy of wind flow estimation over complex terrain has been tested with three computational models: MS-3DJH/3R, WaSP and 3D-RANS. From the results, it has been found that the spatial resolution of topographic input data is a crucial factor in achieving accurate wind flow estimation over complex terrain. Moreover non-linear 3D-RANS computational model with topographical input data derived from high resolution remote sensing images results to higher accuracies in wind flow estimation. An interesting conclusion is that wind speed predictions from the simplest linear model MS-3DJH/3R with high resolution input data are more accurate than results from the sophisticated 3D-RANS solver with low resolution input data.

Tel: 0030 211 770 0998 · Fax: 0030 211 780 9899 · Email: