92nd American Meteorological Society Annual Meeting (January 22-26, 2012)
Wednesday, 25 January 2012: 8:45 AM
The Potential Perils of Averaging Wind Speed and Wind Direction, Using Data From a Variety of Data Providers
Room 239 (New Orleans Convention Center )
Sytske Kimball, Univ. of South Alabama, Mobile, AL

In order to obtain a complete picture of the spatial and temporal wind speed and direction variability in the Mobile Bay area, it is necessary to use wind data from a variety of providers. Providers measure wind for a wide range of applications and, hence, a wide variety of different instruments, mounting heights, site selection criteria, averaging and recording intervals, and averaging methodologies due to the vectorial nature of wind, is employed. This study investigates the sensitivity of wind speed and direction observations to these factors.

Due to technical, budgetary, and application considerations wind data is averaged anywhere from 1-minute to 15-minutes. In most, but not all, cases, the averaging and recording intervals coincide. So if data is averaged over an X-minute period, and recorded every Y minutes, X and Y would be identical. However, that is not always the case. Sometimes Y exceeds X (e.g. data is averaged over 10-minute intervals, but only recorded at 30-minute intervals, i.e. X=10 and Y=30), which leads to data gaps. Furthermore, many studies and applications require longer term averages (e.g. hourly or monthly), therefore, since X is usually shorter than 1 hour, a second averaging process has to be applied. This will be termed “2-pass averaging”. Two-pass averaging would not be a problem if 1) the averaging process was purely arithmetical (as it would be for a scalar quantity) and 2) X and Y were equal.

Since wind is a vector quantity (it has sped and direction), its averaging methodology is not straightforward. There are 2 averaging methods for both speed and direction. Firstly, the wind speed can be considered a scalar and simply averaged arithmetically yielding a scalar mean wind speed. Secondly, all wind vectors in an averaging period can be added together with the length of the resultant vector providing the vector mean wind speed. Similarly the direction of the vector would give the vector mean wind direction. Alternatively, the unit-vector mean wind direction is obtained by considering all wind vectors in the averaging period to be of length 1, the direction of the resultant vector then provides the wind direction. The different methods are useful in different applications. This study will explain the differences between the methods and investigate the magnitude of the differences in mean wind speed and direction obtained from these methods.

The effect of varying X on hourly mean wind speeds and directions will be examined, as will the impact of Y > X on the 2-pass averaging process. The sensitivity of wind speed and direction measurements due to different anemometer mounting heights and site locations will be quantified.

Using 1-minute average wind data from 3 University of South Alabama Mesonet stations, the impact from these various factors will be quantified. These 3 stations were chosen for their proximity to Mobile Bay, their fine (1-minute) temporal averaging/recording resolution, the fact that both scalar and vector wind speed are archived, and the fact that wind speeds are measured at both 2 and 10 m. By considering the 1-minute averaged observations to be instantaneous, larger time period mean wind speeds and directions using both averaging methodologies can be derived and compared. The 1-minute average data can be used for direct comparison of wind data at different heights and different sites.

The 2-pass averaging processes using various values of X and Y do not produce substantially different results as long as X and Y are identical. Differences become larger as Y exceeds X, but depend on their relative values. Averaging methodologies yield generally small differences in values. Mounting height differences produce slightly larger differences in wind speed and direction values, however, differences due to different physical locations outweigh all other factors considered in this study. This is not unexpected given the wide spatial variation in obstructions (shrubs, trees, building, etc.), dynamic forcing, atmospheric stability, and turbulent structure. In all cases, wind direction variability increases as the wind speed decreases. Lower wind speeds allow small scale turbulent motions to dominate the flow regime.

Supplementary URL: