Factors Affecting the Prediction Process

The combined WAsP Analysis and Application procedures is actually a transfer function model linking the wind speeds at the reference site with those at the predicted site. A high correlation between the reference and predicted sites is therefore an essential condition for an accurate prediction by the WAsP model. 

1.  Atmospheric Conditions
   
   It is known that errors in the prediction due to non-standard atmospheric conditions that may affect the behaviour of the wind flow can be very significant. The climatic conditions that affect the accuracy of the prediction by WAsP model may be sorted into two categories:
   
   - The two sites have different wind regimes.
   - The two sites have the same weather regime, but the prevailing atmospheric conditions are not neutrally stable.
   
   WAsP assumes a standard, neutrally stable climate. If the climate of a site wanders from this standard climate, the accuracy of prediction will be affected. These climate effects could vary from being rather unimportant to having a very extensive influence on the prediction. However, some effects can be corrected thanks too a stability correction factor. Prediction of wind flow over hilly terrain in non neutrally stable conditions are not common due to their complexity and the wide range of possible atmospheric conditions of interest. The assessment of WAsP predictions under strong thermal effects has also been studied for the hilly terrain of Sardinia. But the individual effects that each type of climate phenomenon has on the WAsP prediction performance are too diverse to be studied in details in this project.
   


2. Orography
   
   Factors associated with the terrain conditions of both the reference and potential sites have also a substantial influence on the WAsP prediction performance. Errors could be caused by the following factors:
   - Individual site ruggedness
   - Extensive flow separation
   - Topographic features beyond the terrain map considered by WAsP
   - Site elevation
   - Hill height relative to the boundary-layer height
   - Effective surface roughness length due to the ruggedness of terrain
   - The degree of turning due to large-scale terrain effects and the resulting changes in the frequencies of occurrence in each sector.
   Usually a minimum area of at least 6x6 km2 around the sites is recommended for the mapping of the zone, depending on site complexity. Orographic effects are the main focus of this page and are discussed more fully.
   
   
3. Wind Speed Records
   
   Wind speeds are generally recorded as 10-minute mean, maximum and minimum values. While predicting the wind resource, WAsP assumes that these 10-minute wind speeds are perfectly correlated between the two sites and are related by a simple transfer function. However, it is very unlikely that the 10-minute wind regime would cover both sites unless they are located very close to each other. Usually, the reference and the potential sites are separated by a significant distance (at least 50 km generally). A longer averaging time (1 hour for example) should be more appropriate to enable a particular wind speed to physically envelope the two sites considered. Unfortunately, only a small improvement has been observed using 1 hour-average rather than 10 minute-average. This issue therefore was not pursued further and remains unsolved. 
   The record length is another crucial factor in the magnitude of the resulting correlation coefficients. Observations of wind speed indicate that monthly, seasonal and yearly variations of the wind significantly affect the correlation values. Relatively short-term measurements therefore generate a significant variation in the transfer function, which depends on the prevailing atmospheric conditions at the time of the measurements.
   
   
4. Weibull Frequency Distribution
   
   The WAsP Analysis procedure uses the standard Weibull frequency distribution as a tool to represent the frequency distribution of wind speed in each directional sector. Actually, the WAsP model creates the generalised wind data by forcing the measured data to fit a standard Weibull distribution. However, the measured distributions do not always perfectly fit the Weibull distribution, which can generate errors in the prediction because of the transformation necessary. The magnitude of this error is indicated in the self-predictions when the reference and predicted sites are the same.
   
   
5. Wind Direction
   
   We know that the wind rose is generally divided into 12 equal direction sectors. However, steep and oblique ridges may affect the direction of the incident flow. This turning could cause the wind direction to fall into an adjacent directional sector, which is different to the one at the reference site. 

 < Previous | Next >