On Estimation of Surface Roughness for Wind Energy Resources Assessment
风能资源评估中地表粗糙度的研究

There are two commonly used techniques, i.e., fitting to logarithmic wind profile and the Davenport roughness classification system, in wind energy resources assessment, which were used for evaluation of surface aerodynamics roughness length of a steppe in Inner Mongolia. It was expected that the grassland location would have roughness lengths between the Davenport classes of the Fifth (0.25 m) to the Third (0.03 m). Data, collected from May 21, 2009 through May 14, 2010 and measured by cup anemometers and wind vanes (1Hz) deployed at four levels (10, 30, 50 and 70 meters above the ground), were analyzed. If one of the levels had missing data for a time period, all records from that time period were removed from the analysis. Then, original data were averaged over 10 minute intervals. Time periods with wind speed at 50 m height or above 6 m/s, were chosen for further analysis, which was because if wind speed at 50 m height is under 6 m/s, most large turbines would generate little usable power. This approach is useful for wind resource assessment in the absence of stability calculation and can largely reduce errors of calculation of roughness length. Ten-minute averaged wind profiles were classified into categories of twelve directions. The liner least-squares method was used to fit the averaged logarithmic wind profiles. It was found that the roughness length had monthly and seasonal variations. In summer-fall months when grass was lush, the average roughness length was measured to be 0.138 m and in winter-spring months when grass was shriveled, it was measured to be 0.088 m. The minimum value of the monthly-average roughness length was 0.041 m appearing in January of 2010, which may have reflected that the land was covered by snow. The hourly-average roughness lengths showed a diurnal variation, with the maximum value occurring at midnight and the minimum value at noon. The roughness lengths computed based on southwest and west wind measurements were also shown to be larger than these computed based on northeast and north wind measurements in all time periods. Minor topography changes could influence the roughness lengths. The higher topography, despite being several kilometers far away from tower, was shown to impact the roughness lengths. In general, the magnitude of roughness lengths calculated from the two techniques seems to be consistent. To estimate wind power densities, we extrapolated a 10 m/s wind speed at 50 m to an 80 m hub height and used the estimated roughness lengths based on above two techniques. Difference in estimated roughness length between the two techniques is about 2%, which means that the Davenport roughness classification system appears to be a practical method to estimate roughness length for flat terrain with moderated roughness.