Are observed wind speeds too high?

Comparisons of 10-m AGL numerical weather prediction model winds to point observations consistently show that daytime model winds are slow compared to observations, even after improvement of model physics and going to finer scales. We suggest that part of the discrepancy derives from the observed surface winds being faster than the area average because winds are typically measured in open areas. To document the bias in observed winds, we compare 10-m wind speeds from the nine International H_2O Project (IHOP_2002) surface flux stations, to winds extrapolated downward from 70 m by applying the Monin-Obukhov relationships to University of Wyoming King-Air aircraft data collected at 70 m and 1-3 other levels in the BL during its fair-weather Boundary-Layer Heterogeneity (BLH) missions. The BLH missions repeat straight-and-level legs several times, but calculations to the desired accuracy stretched the data to their limit, requiring removal of remaining heading-related wind and momentum-flux biases and assuring that the resulting data applied to common times. Another concern is how well M-O theory applies. It is suggested that M-O theory is appropriate on four days, since 70 m is less than 10% of the boundary-layer depth z_i, and the Obukhov length L < -70 m. On 2 additional days, the 70 m is within -2.5 L of the surface as well as <0.1 z_i, suggesting M-O theory probably works. As expected, the extrapolated winds are slower than the measured winds by 0.4 m/s on average for all six days and by 0.5 m/s for the days with –L>70 m. The estimated accuracy, assuming M-O theory applies, is ~0.2 m/s.