Roughness maps from gridded land cover information

Background

Satellite images, e.g. LANDSAT, or land-cover databases, e.g. CORINE, might provide useful information for roughness maps.

Preliminary work has been made, e.g. in the EO-WINDFARM project sponsored by the European Space Agency (ESA). The report Risø-R-1497 (Chapter 3) discusses automatic conversion to the WAsP map format. Roughness information is slightly more difficult to handle than digital elevation models, but we hope to include these routines in the WAsP Map Editor at a later stage.

EO-WINDFARM partner Armines offers CORINE based information at www.dataforwind.com. However, the roughness information is deduced on basis of a look-up table, which in principle will always be subject to dispute as explained below under “General Considerations”.

Read more on satellite information at www.risoe.dk/vea-atu/remote

General considerations

• The basic information is land cover, e.g. vegetation type, which must be translated to typical aerodynamic roughness length with a look-up table. It should be stressed, that such a translation can never be unique, but will involve a large element of subjective judgment. E.g. one may always dispute whether a grass farm field should be assigned a roughness length of 0.03 or 0.02 m. So, the translation of land cover to roughness length will always be the responsibility of the user, and one should consequently select a data source for land cover based roughness information where it is possible to exercise this responsibility. As a minimum, one should be able to inspect the look-up table, but it would be natural to require also the possibility to change or prepare the look-up table according to one’s own judgment. Ideally, it is recommended to check the land-cover-to-roughness translation on a site inspection trip.
• The data resolution might be too coarse, e.g. in the first version CORINE pixels were 250´250m.
• The data resolution might be too fine, e.g. LANDSAT TM pixels are 20´20m producing too many points for WAsP to handle – unless we simplify the grid by a majority filter, see Risø-R-1497.
• Shelterbelts can contribute significantly to terrain roughness, but this information might be absent in land-cover files.
• Satellite images are snapshots, but for wind resource estimation we need annual average roughness lengths.
• The roughness information should correspond to the time period in question: for projected wind farms the roughness information should represent the most recent (and maybe even future) conditions, but for analyses of measured wind data, the roughness length should represent the observation period.

Back to Frequently Asked Questions page