ESTIMATION OF SECTOR ROUGHNESS LENGTHS AND THE EFFECT ON PREDICTION OF THE VERTICAL WIND-SPEED PROFILE

An estimate of roughness length is required by some atmospheric models and is also used in the logarithmic profile to determine the increase of wind speed with height under neutral conditions. The choice of tec hnique for determining roughness lengths is generally constrained by t he available input data. Here, we compare sets of roughness lengths de rived by different methods for the same site and evaluate their impact on the prediction of the vertical wind speed profile. Wind speed and direction data have been collected at four heights over a three-year p eriod at the North Norfolk Wind Monitoring Site. Wind speed profiles w ere used to generate sector roughness lengths based on the logarithmic profile formula. This is the only direct way of determining roughness lengths. The simplest and cheapest method is to use maps with publish ed tables giving roughness.length estimates for different terrain type s. Alternatively Wieringa (1976, 1986) and Beljaars (1987) give formul ae for determining roughness lengths from wind speed gusts or standard deviations. The four sets of estimated roughness lengths vary conside rably. They were used to estimate 34 m wind speeds from 12.7 m observa tions. The profile-derived roughnesses are used simply as a check on t he prediction of the wind speed profiles. The terrain-derived roughnes s lengths give reasonable results. Gust-derived and standard deviation roughnesses both predict wind speeds which are lower than the observe d ones. The error is greater in the case of standard deviation roughne sses. If stability corrections are applied in the prediction of the ve rtical wind speed profile, the results are considerably improved.