Introduction
A gust factor (GF) is defined as the ratio between the peak wind gust of a specific duration to the mean wind speed for a period of time. It is a simple statistic but is dependent on numerous inputs, including the roughness length (exposure), distance from an upstream terrain change, stability, height, and, potentially, the presence of convection. Wind speed fluctuations are associated with pressure and force fluctuations on a building, and result in fatigue loading on various structural components. Understanding differences in the structure of the wind, which may exist in various high-wind environments, is imperative for proper wind load design. Until recently, there was a lack of high-resolution wind speed data that are needed to calculate GFs from landfalling tropical cyclones, because anemometers and/or recording systems often fail in extreme conditions. Anemometers that do survive often do not record the required information from which specific GFs can be determined. Recently, several universities, including Texas Tech University (TTU), have developed programs to deploy instrumented towers in the paths of landfalling tropical cyclones. These efforts have collected high-resolution wind speed information that can be leveraged to determine various GFs, as well as other turbulence statistics.
Background
Several studies have been conducted to address the uncertainties regarding differences between winds that are generated by tropical cyclones and those that are generated by extratropical systems. Krayer and Marshall (1992) standardized tropical cyclone wind data to an open exposure, then compared it with extratropical wind data from an earlier study completed by Durst (1960). Krayer and Marshall found a mean 2-s to 10-min GF of 1.55 from hurricane winds, while Durst found a mean 2-s to 10-min GF of 1.40 for extratropical winds. Sharma and Richards (1999) proposed that the difference might lie in the stability of the environment in which the wind is generated. They used potential temperature profiles to show that tropical cyclones are convectively unstable and, therefore, have higher turbulence intensities than other situations where turbulence is generated primarily by mechanical shearing. On the other hand, Sparks and Huang (1999) found there to be little difference between tropical cyclone GFs and those found in extratropical systems. They proposed that with the exception of periods of intense convection, the difference in GFs from the two (the Krayer and Marshall, and the Durst) studies was related to differences in roughness exposure. Krayer and Marshall’s (1992) data came from sites with airport exposure, while the data used in Durst’s (1960) study came from a site in Cardington, United Kingdom, with a more open exposure. Sparks and Huang also proposed that during periods of significant convective activity, wind gusts could approach the wind speed above the top of the boundary layer. During these times, the roughness of the exposure may not affect the magnitude of the gust. They cite two cases—an extratropical cyclone and Hurricane Mitch—that produced similar high wind speeds and GFs, despite the different origins of the wind. Given these uncertainties and the availability of multiple high-resolution wind speed time histories obtained from both landfalling tropical cyclones and extratropical systems, this paper seeks to extend the current database of GF information and to explore the similarities and differences that exist between winds generated in tropical cyclones and extratropical environments.
Data sources
Two distinct types of wind data were used to complete this GF study. One set contains only information collected from landfalling tropical cyclones, while the other contains information gathered from an experimental site located near Lubbock in west Texas. While the collection location(s) and wind source for the two datasets is different, the same instrumentation was used to collect both sets of data.
Instrumentation
Some of the mobile-instrumented towers used in this study were originally designed to collect high-resolution wind speed and direction data from landfalling tropical cyclones. These towers (two) are ruggedized to withstand sustained winds of 67 m s−1 and feature anemometers at three to five levels (Conder et al. 1999; Schroeder and Smith 2003). They have been deployed in landfalling U.S. tropical cyclones since 1998 and 1999 and are named Wind Engineering Mobile Instrumented Tower Experiment (WEMITE) 1 and 2, respectively. Three more nonruggedized 10-m towers, each with wind instrumentation at one level, were added in 2001. All of the towers were capable of collecting relatively high resolution wind speed data (sampled temporally at 2–10 Hz), enabling the investigation of the turbulent fluctuations of the wind. In addition to wind speed data, the towers sampled barometric pressure (BP), temperature (T), and relative humidity (RH). A summary of the five towers from which data were collected and used in this study is provided in Table 1.
The main instrument used to collect wind speed and direction data on each tower was an R. M. Young Wind Monitor Model 05106. The instrument is a propeller vane–type anemometer that yields measurements of both wind speed and direction. It has a distance constant of 2.7 m for 63% recovery. Because the anemometer design includes moving components with inherent inertia, the sensor’s ability to measure small-scale (high frequency) energy of the wind is compromised (Schroeder and Smith 2003). For this study, which is focused on resolving 2-s peak and 10-min mean wind speeds to determine GFs in high-wind environments, this limitation does not compromise the resulting statistics. For a more complete discussion of the response limitations of the instrument used in this study, see Schroeder and Smith (2003).
Tropical cyclone wind speed data
The tropical cyclone wind speed data used in this study were obtained during the 1998–2002 Atlantic hurricane seasons. A list of storm/platform names and deployment locations is provided in Table 2. While the vast majority of the storms listed in Table 2 were classified as hurricanes at the time of landfall, most of the collected datasets only contain tropical storm–force winds. On most occasions, an open/airport exposure was sought for the deployment site to limit upwind obstructions from all wind directions. However, this was not always possible, and the dataset contains information from a variety of exposures, including marine and rough exposures as classified by Wieringa (1993). The dataset also contains transitional flow regimes. Wind speed and direction data were collected at 2–10 Hz.
Extratropical wind speed data
In May 2002, an array of seven towers was deployed along a runway at Reese Technology Center (a decommissioned Air Force base) west of Lubbock as part of an experiment to observe and document the kinematic and thermodynamic structure of thunderstorm outflows (Gast and Schroeder 2003). Seven of TTU’s mobile hurricane towers were used in the experiment, including each of the towers represented in Table 1. Wind data, along with other meteorological parameters, were recorded at 2 Hz throughout May and June. The towers were arranged in a linear array from north to south. While the exposure of the experimental site at Reese Technology Center is mainly open, some of the data yielded roughness lengths representative of rougher regimes.
Each record was closely examined before it was included in the study. Instrument-induced data spikes (representing short-lived values sufficiently high that they could not represent actual wind speeds) were removed by dividing the record into two records—one before the spike and another after the spike. The records were also examined for signs that they might include data from a thunderstorm outflow. The mechanisms that generate extreme thunderstorm outflows are relatively small scale; therefore, thunderstorm outflow winds are often highly nonstationary and transient (even within a 10-min segment) relative to tropical cyclone winds. For that reason, data from days when significant thunderstorm outflows occurred were removed for this study. If a particular wind speed record had a sudden strong peak in wind speed that lasted less than 10 min and was accompanied by an abrupt change in wind direction, it was considered thunderstorm outflow data. Written records of thunderstorm outflow occurrences were also kept, and data from those days were not included. Most of the data, however, were from wind generated by large-scale surface pressure gradients and dryline passages. Wind records generated by these event types were included in this study.
An example of an extratropical nonthunderstorm high-wind producer
Although the data from significant thunderstorm outflow events were removed from this study, wind data from other localized high-wind events were included. During the late spring and early summer, the dryline—a boundary separating air with high dewpoints from air with low dewpoints—frequently moves across the area where the tower array was arranged. Passage of the dryline usually results in strong, gusty winds in the Lubbock area. Passage of the dryline is associated with a sharp change in wind direction and an increase in wind speed (Conder et al. 2003). Figure 1 shows wind speed and direction time histories from 10 June 2002. The passage of the dryline is evident on these plots by the abrupt change in wind direction, from approximately south to approximately west, and the sharp increase in wind speed from a mean near 6 m s−1 to a peak 2-s gust of 16.41 m s−1. Changes in temperature and relative humidity also accompanied the passage of the dryline.
Analysis methodology
Each individual wind speed time history represents data recorded using one instrument at a specific height on a specific tower within an individual event. These time histories were divided into 10-min segments (fully segmented with no overlap). The individual 10-min segments were then used to determine various wind flow characteristics, including GFs, turbulence intensities (TI), and roughness lengths (Z0). The resultant statistics were assimilated into two separate databases—one focused on tropical cyclones and the other on extratropical data. Because GFs are dependent on height, only those statistics generated from data recorded by instruments located at heights between 9.15 and 10.67 m (∼10 m) AGL were used in this study.
Gust factor determination
Roughness length determination
The profile method requires simultaneous data to be collected from multiple anemometer heights, which is not always available or possible. The TI method is widely used for sites with instrumentation available at only one level. Given the differences that can exist between the determination methods, both are included in this study.
Stratification
Wind speed
After assimilating all of the statistics into the tropical database, the GFs were plotted against the mean wind speed (Fig. 3), and appear to include increasing amounts of scatter with decreasing wind speed. This result is not unexpected; environments with lower average wind speeds are much more conducive to free convection (thermals), which can introduce additional turbulence in the record and increase deviations from the mean. This set of low wind speed data represents information collected from the outskirts of the various tropical cyclones, and to enhance interpretation of the resulting statistics under higher wind environments, segments with a mean wind speed of less than 5 m s−1 were removed.
Exposure/roughness length
Because GFs vary by upstream terrain conditions (roughness), calculated roughness length values were used to stratify the remaining dataset into the various roughness regimes (Schroeder et al. 2002) that are shown in Table 3. Without stratification by roughness, relevant comparison between and assimilation of GF statistics generated from different deployment locations (with different exposure characteristics) would be troublesome to make and difficult to interpret. One complication is that there can be differences in the resulting Z0 value, depending on which method is employed for its determination. Hence, two sets of stratifications were performed for the tropical dataset—each using one of the roughness length determination methods (TI and profile) employed in the study.
Figure 4a, a scatterplot of GFs versus TI-derived Z0 from the tropical dataset, demonstrates the sensitivity of GFs to Z0. The plot shows GFs increasing approximately linearly with Z0. This relationship is not evident when defining Z0 using the profile method, as shown in Fig. 4b. These differences reinforce the fact that the resulting statistics may vary, depending on the method employed to determine the roughness length.
Figure 5 shows GFs versus mean wind speed with the data points stratified by the roughness regime (TI method). The data points are evenly distributed horizontally (almost independent of mean wind speed), but appear to be layered vertically by the roughness regime. The bottom layer, associated with the lowest mean GF, represents the smooth regime and its associated lowest roughness length range. The next three layers moving upward represent data classified as open, open to roughly open, and roughly open to rough. Each corresponds to a higher mean GF and roughness length.
Results and discussion
Tropical cyclone gust factors
A summary of the tropical cyclone GF statistics, including the mean GF, standard deviation, maximum GF, minimum GF, and the number of observations, is shown in Table 4. The table is organized by roughness regime and method of roughness length calculation. Prior to stratifying the tropical database into roughness regimes, there were 1811 observations with a mean gust factor of 1.59, a maximum gust factor of 2.94, a minimum gust factor of 1.18, and a standard deviation of 0.24. The mean Z0, using all 1811 observations, was 0.0538 m (0.0376 m) as was determined by using the TI (profile) method of calculation. This comparison underscores the differences that can be encountered in using these two Z0 determination methods, especially when transitional flow regimes (in terms of mean and fluctuating components) are inherent to the dataset.
The mean GF for the entire tropical dataset, 1.59, is higher than both the mean GF value found by Krayer and Marshall (1992) of 1.55 and the Durst (1960) mean GF of 1.40. This result is not unexpected, because the entire tropical database contains a variety of roughness regimes, including many that are not as “smooth” as an open exposure. Krayer and Marshall’s GF information was generated from data that were adjusted to a roughness length of 0.03 m, which corresponds to an open roughness regime. If only the wind speed segments in the database with Z0s ranging from 0.02 to 0.0499 m (open roughness regime) are examined, the average GF value becomes 1.49 (1.55), as was determined by using the TI (profile) method. Figure 6a shows a GF histogram for the entire dataset with wind speeds greater than 5 m s−1. The resulting distribution is heavily skewed to the right. When the data were stratified into roughness regimes using the TI-derived Z0, the GF distributions are much more symmetrical. Figure 6b shows the GF histogram for the roughly open to rough dataset (0.09 m ≤ Z0 ≤ 0.1899 m) that was calculated using the TI method.
Several differences have been identified, depending on whether the roughness values employed for comparison are generated using the TI or the profile method. These differences are quite extreme, because 81% of the data segments examined in this study would be categorized into different roughness regimes as a result of using the two different methods. This result may not be all that surprising because many of the deployment locations include various terrain conditions and transitional flow regimes (upwind of a nearby roughness change). It has been shown that the mean wind profiles come into equilibrium faster than the turbulent (Deaves 1981) fluctuations and peak gusts as a new internal boundary layer forms, following a change in terrain. Because the GF is dependent on the fluctuating component (peak values) of the wind, the TI method of determining Z0 yields a stronger relationship with the GF, regardless of whether full equilibrium has been reached. This point is discussed in more detail in section 5d.
Extratropical gust factors
The summary statistics for the extratropical gust factors are provided in Table 5. As expected, the mean GF values for the dataset, stratified using the TI method, increase with increasing roughness; however, the dataset that is stratified using the profile method indicates the GF remaining relatively stable with increasing roughness. This result is contrary to what is expected. Increasing roughness lengths, as identified by any method, would seemingly indicate more mechanical mixing and higher GFs.
The collection site for the extratropical dataset is open country in almost every direction for a significant distance, but there are some slight changes in roughness that occur. Various buildings are located in the far distance (1–2 km) to the east-southeast of the site, with a few low-rise building located closer to the site. Open, flat fields and runways are also nearby. Patchy grass and extremely small brush and shrubs mark the immediate area (100 m) surrounding the instrumentation. Prior to stratification by roughness, there were 5975 10-minute segments of data with wind speeds greater than 5 m s−1 (the threshold value used for this study). Only 2427 GF observations were acquired by using platforms with multiple anemometer heights, which would enable the calculation of Z0 via the profile method. Once stratified by the roughness length (using the profile method), the total number of observations within the extremities of the included roughness regimes (0.005 m ≤ Z0 ≤ 0.1899 m) was reduced to 1869 GF observations. The vast majority (82.6%) of the 558 unused observations represented a “smoother” upstream terrain than any roughness regime used within this study (Z0 < 0.005 m). When stratified using the TI method for determining roughness length, the same result occurs with 95.6% of the 3464 unused observations representative of a smoother regime.
The mean GF for the entire extratropical dataset (mean wind speeds > 5 m s−1) was 1.35, which is lower than both of the values found in previous studies. This result was expected because the vast majority of the extratropical data comes from a fairly open roughness regime, which is representative of an extremely smooth approach. After stratification by Z0 (using the TI method), the mean extratropical GF from the open roughness regime was 1.44, which falls between the values found in previous studies but below the 1.49 value determined from the tropical dataset.
In comparison with the tropical GF distributions, the extratropical distributions are similar in shape regardless of whether the TI or the profile stratification techniques are employed. Histograms for the open roughness regime are shown in Fig. 7. The reason for the additional symmetry relative to the tropical dataset is unknown; however, the site did contain a longer fetch of relatively uniform roughness in comparison with some of the tropical deployments. This additional fetch would allow the profile and turbulence characteristics of the wind to come closer to equilibrium values than most of the tropical data. While the extratropical dataset contains a significant amount of data, the stratified results are still confusing. Even with only minor changes in upstream terrain conditions at substantial distances from the observation sites, there are seemingly significant effects in some of the resulting wind flow statistics.
Comparison of extratropical and tropical gust factors
Within the same roughness regimes, the mean GFs for tropical cyclone data were consistently higher than those for extratropical data. Although the percent difference between the extratropical and tropical GFs is small, as can be observed in Table 6, it increases steadily with increasing roughness. The same effect can be observed in Fig. 8, which is a set of four histograms comparing the extratropical and tropical GFs within the same roughness regime (as was determined by using the TI method). In the case of the smooth regime, the histograms for the two distributions follow each other fairly closely. For the open regime, the two distributions begin to show some distinct differences, including different mean values. The effect is most pronounced in the roughly open–to-rough regime, which contains the largest roughness lengths. This observation agrees well with Table 6, where the percent difference between the mean GF from the extratropical and tropical datasets is also greatest for the roughly open–to-rough regime at 6.13%.
Table 6 also presents the mean Z0 (TI method) determined for both extratropical and tropical cyclone roughness regime–stratified datasets. The percent difference in the mean value of Z0 between the extratropical and tropical datasets varies from 22.81% for the smooth roughness regime to 0.23% for the roughly open–to-rough regime. Interestingly, based on the mean value of Z0, the exposure (indicated by the roughness lengths) for the two datasets becomes more similar with increasing roughness, while the GF distributions diverge and become substantially different. Hence, the largest percent difference in mean GF between the extratropical and tropical datasets corresponds to the roughness regime (roughly open to rough) with the most similar mean roughness length values. The differences in mean GF values are significant at a 1% test level for all roughness regimes, except for the smooth regime—the difference in the smooth roughness regime is significant at a 7% test level. This result suggests that other parameters beyond observation height and roughness, such as stability and convective-scale motions, may be playing an important role in the final gust factor statistics. Table 7 summarizes the statistics based on using the profile method for Z0 determination.
Comparison of the associated wind speed histograms
Regardless of the approach employed to determine Z0, the GF is solidly linked to the TI and the spread of the associated distribution of the wind speed fluctuations about the mean. Differences in the resulting mean GFs indicate a discrepancy in the associated wind speed distributions between the tropical and extratropical datasets. Specifically, it identifies a discrepancy in the extension of the right tail of the two distributions. Figure 8 shows a plot of the distribution about the mean of the ratio of two different peak lengths and the 10-min mean for both the tropical and extratropical regime. The plot was generated by finding the ratio of two peak wind speeds (2 and 60 s) to the 10-min mean wind speed. Only data collected in an open exposure, as was determined through comparison of the TI-derived Z0 values against the stated values (Table 3), were included in the histograms. The resulting ratios were assimilated using a histogram algorithm, and the frequencies were normalized by the total number of samples available. The 2-s extratropical and tropical histograms compare well, while the right-hand tail of the tropical distribution extends further right, enabling a slightly higher GF for the dataset. At 60-s (longer peak durations), the tropical distribution is more consolidated around the mean value of 1. In this case, the 2- and 60-s values were obtained using a fully segmented approach (not a moving average). As a result, the maximum tropical wind speed ratio value observed in Fig. 9 is slightly different from the GF value shown in Table 4.
Conclusions and limitations
High-resolution wind speed data were collected from tropical cyclones during the last five Atlantic hurricane seasons. The data were compiled in a database and then stratified by wind speed and exposure (roughness length) for examination. Comparisons were made with an extratropical dataset collected near Lubbock, Texas, with the same instrumentation. Last, the associated wind speed histograms were compared and contrasted. Conclusions from this study include the following:
Stratification of the tropical and extratropical GF datasets into various roughness regimes (exposures) must be performed in order to make relevant comparisons between the datasets. Without stratification, based on the values of Z0, the resulting GF statistics and histograms are not well behaved. Once stratification occurs, the GF histograms become much more symmetrical.
The mean tropical cyclone GF determined for “open exposure” was 1.49 when the open exposure classification was made using a Z0 value that was determined by using a TI method. If the profile method was employed to determine the Z0 values and make roughness classifications, the mean GF was 1.55 for the tropical dataset.
The mean tropical cyclone GF determined within any specific roughness regime (using the TI method) was always higher than its extratropical counterpart. The largest percentage difference between the extratropical and tropical datasets was determined in the “roughly open to rough” exposure (0.09 m ≤ Z0 ≥ 0.1899 m) where the mean tropical GF was 7.41% higher than its extratropical counterpart. Results using stratifications based on the profile method were even more diverse.
The resulting wind speed histograms generated from the extratropical and tropical datasets show some significant differences, including the presence of higher-magnitude short-duration wind speed peaks in the tropical dataset, while the extratropical dataset yield a flatter histogram at longer peak durations.
As shown within this study, transitional flow regimes complicate the GF analysis greatly. The profile and TI methods of determining Z0 produced different results in many situations, and while the mean wind profile may come in equilibrium relatively quickly, the turbulent fluctuations take additional time and distance. This issue complicates the stratification of the data with respect to roughness. In most cases, the field deployments of instrumented towers are not placed in locations with fully developed flow, even though best efforts are being made to do so, resulting in transitional flow regimes that are inherent to the dataset. True equilibrium flow would demand kilometers of unaltered exposure, which is rare, if not impossible, to find in most cases along the U.S. coastline. The only way to further evaluate these transitional effects is to conduct an in-depth study site by site with recent aerial photographs to evaluate transitions that occur within 5–10 km. This effort is planned for some of the tropical deployment sites, as well as the extratropical deployment site.
Even with an abundance of higher-resolution wind speed data as compared with previous studies, the root cause for the differences between the extratropical and tropical GF statistics are not fully understood. Whether the disparity in statistics is due to differences in boundary layer stability or the presence of convective-scale motions that can modify the boundary layer is difficult to determine through the examination of only surface level wind speed data. If the underlying reason for the difference in GF statistics is relatively vigorous convection, then these differences would most certainly exist in precipitating extratropical cyclones as well. In fact, one can easily conceptualize that thunderstorm downdrafts could easily modify the boundary layer from above with far greater efficiency than most tropical systems. From a tropical cyclone perspective, the questions then become how widespread is this “convective” effect within the general extent of the tropical cyclone wind field, and to what extent are the GFs found within these convectively active regions different from those found in other regions of the tropical cyclone. Analysis of surface wind speed data alone cannot answer these questions. Rather, it must be coupled with other data sources, such as radars, to evaluate the presence and location of convection.
Another limitation of this study is the minimal amount of extreme wind speed cases found in the database. This problem is true even within the tropical cyclone dataset used for this study. Major hurricanes result in a vast increase in wind damage relative to weak tropical cyclones due to the squared relationship between wind speed and wind load. However, even with all of the field experimentation conducted in tropical cyclones at landfall over the 5-yr period, the database of surface level wind speed information from which to draw conclusions about major hurricanes is dreadfully inadequate and almost nonexistent.
Acknowledgments
The authors acknowledge NIST (Department of Commerce NIST/TTU Cooperative Agreement Award 70NANB8H0059) and NSF (ATM-0134188) for supporting this study. We thank all of the Texas Tech University graduate and undergraduate students who contributed to the collection of the tropical cyclone wind data over the 5-yr period; they are too numerous to list. We specifically thank Rob Howard for his dedication to the hurricane deployment efforts and Kirsten Gast for sharing all of the extratropical data used in this study. The authors also acknowledge the effort of three anonymous reviewers who improved the quality of this manuscript.
REFERENCES
Beljaars, A. C. M., 1987. The measurement of gustiness at routine wind stations—A review. KNMI Scientific Rep. WR87-11.
Conder, M. R., R. E. Peterson, J. L. Schroeder, and D. A. Smith, 1999. Comparison of hurricane wind data from Hurricane Bonnie: The Texas Tech Wind Engineering Mobile Instrumented Tower Experiment (WEMITE) and the NWS Wilmington ASOS. Proc. 13th Symp. on Boundary Layers and Turbulence, Dallas, TX, Amer. Meteor. Soc., 484–487.
Conder, M. R., R. E. Peterson, J. L. Schroeder, and D. A. Smith, 2003. An analysis of wind characteristics during the ERODE experiment, April-May 2002. Preprints, 11th Int. Conf. on Wind Engineering, Lubbock, TX Int. Assoc. for Wind Engin. and Amer. Assoc. for Wind Engin., 2029–2036.
Deaves, D. M., 1981. Terrain-dependence of longitudinal R.M.S. velocities in the neutral atmosphere. J. Wind Eng. Ind. Aerodyn. 8:259–274.
Durst, C. S., 1960. Wind speeds over short periods of time. Meteor. Mag. 89:181–187.
Gast, K. D., and J. L. Schroeder, 2003. Supercell rear-flank downdraft as sampled in the 2003 Thunderstorm Outflow Experiment. Preprints, 11th Int. Conf. on Wind Engineering, Lubbock, TX, Int. Assoc. for Wind Engin. and Amer. Assoc. for Wind Engin., 2233–2240.
Krayer, W. R., and R. D. Marshall, 1992. Gust factors applied to hurricane winds. Bull. Amer. Meteor. Soc. 73:613–617.
Schroeder, J. L., and D. A. Smith, 2003. Hurricane Bonnie wind flow characteristics as determined from WEMITE. J. Wind Eng. Ind. Aerodyn. 91:767–789.
Schroeder, J. L., M. R. Conder, and J. R. Howard, 2002. Additional insights into hurricane gust factors. Preprints, 25th Conf. on Hurricanes and Tropical Meteorology, San Diego, CA, Amer. Meteor. Soc., 39–40.
Sharma, R. N., and P. J. Richards, 1999. A re-examination of the characteristics of tropical cyclone winds. J Wind Eng. Ind. Aerodyn. 83:21–23.
Sparks, P. R., and Z. Huang, 1999. Wind speed characteristics in tropical cyclones. Wind Engineering into the 21st Century—Proceedings of the Tenth International Conference on Wind Engineering, A. Larson, G. L. Larose, and F. M. Livesey, Eds., A. A. Balkema, 343–350.
Wieringa, J., 1993. Representative roughness parameters for homogeneous terrain. Bound.-Layer Meteor. 63:323–363.
(a) Wind speed (U) and (b) wind direction (WD) time histories from a day when the dryline passed over the tower array (extratropical data).
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
The 10-min wind speed segment, as collected by WEMITE 1 from Hurricane Bonnie, is shown. The location of the maximum 2-s wind gust is identified. The segment provides for a 2-s peak to 10-min mean gust factor of 1.64.
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
Comparison of 2-s peak to 10-min mean (U) GF vs mean wind speed (tropical dataset) is given, before stratification by wind speed (tropical data). The vertical line marks 5 m s−1, the lowest wind speed included in the study.
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
Scatterplot of 2-s peak to 10-min mean GF vs Z0, determined using (a) the TI method and (b) the profile method (tropical data, mean wind speed > 5 m s−1).
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
Scatterplot of 2-s peak to 10-min mean GF vs mean wind speed (tropical data, Z0 calculated using the TI method, mean wind speed > 5 m s−1).
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
(a) The 2-s peak to 10-min mean GF frequency (f) for the tropical dataset with wind speeds >5 m s−1and (b) the dataset stratified by wind speed and including only data collected in a roughly open to rough (0.09 m ≤ Z0 ≤ 0.1899 m) exposure.
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
The GF histogram for the open extratropical dataset when data are stratified by (a) the TI method and (b) the profile method.
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
Normalized frequency (F) of extratropical and tropical cyclone 2-s to 10-min GFs in each roughness regime (roughness determined using the TI method).
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
Histograms of 2-s to 10-min and 60-s to 10-min wind speed ratios for the tropical and extratropical datasets.
Citation: Journal of Applied Meteorology 44, 2; 10.1175/JAM2199.1
Details of the towers used to collect high-resolution wind speed data for this experiment.
Storm names and associated deployment locations (1998–2002) employed for this study.
Roughness regimes and associated roughness length values.
Summary of 2-s peak to 10-min mean GF data (tropical dataset).
Summary of 2-s peak to 10-min mean GF data (extratropical dataset).
Comparison of statistics from the tropical cyclone and extratropical wind speed datasets (TI method).
Comparison of statistics from the tropical cyclone and extratropical wind speed datasets (profile method).
