1. Introduction
Wind gusts, a turbulent phenomenon occurring at small spatial and temporal scales, are an important and potentially costly environmental hazard. Economic losses due to wind include forest and crop damage and are of great interest to the insurance industry (Welker et al. 2016; Valta et al. 2019; Jung and Schindler 2019; Usbeck et al. 2010). A single high-wind event in August 2020, for example, caused EUR 2,500,000 (∼USD 2.6 million) in apple orchard losses in northwest Italy (De Petris et al. 2021). During 1952–2006, the United States suffered losses in excess of USD 350 million annually from high-wind catastrophes (Changnon 2009).
Gusts are an important concern for a wide range of industrial and economic sectors. Within the energy sector, wind power is a significant driver of gust research (Jung et al. 2017; Suomi et al. 2013; Sinden 2007). Gusts are also of concern to electrical power generation, as strong winds affect the design and integrity of power distribution infrastructure (Teoh et al. 2019; Pinto and Belo-Pereira 2020; Wong and Miller 2010). Gust-related hazards within the transportation sector affect aircraft operations (Manasseh and Middleton 1999) and the stability of ground vehicles (Kim et al. 2020; Caban et al. 2021; Pirooz et al. 2021), trains (Zhang and Li 2021; Montenegro et al. 2020), and container cranes (Su et al. 2021; OSHA 2021). Gusts are a necessary design consideration for wind loads and water tightness for buildings, bridges, and other structures (Thalla and Stiros 2018; Sheridan 2018; Van Den Bossche et al. 2013) and are important to the aerodynamic performance and tracking of drones (Lei et al. 2021; Siti et al. 2019), the assessment of fire danger (Luchetti et al. 2020; Adame et al. 2018), and the resuspension of deposited radioactive particles (Giess et al. 1997).
Predicting peak wind gusts is a challenging component of weather forecasts. A limiting factor in forecasting peak gusts is the inability of existing measurement networks to spatially or temporally characterize the turbulent structures that produce strong gusts. These structures, which include microscale and mesoscale phenomena such as lee waves, small convective showers, and boundary layer eddies with typical horizontal length scales from 1 to 50 km (Hart et al. 2017; Hewson and Neu 2015; Grubisic et al. 2008), are moving closer to the resolution threshold of numerical weather prediction models utilizing increasingly fine grids (Sheridan 2018). Despite the availability of a variety of strategies that utilize dynamical, statistical, and empirical formulations to estimate peak gusts (Gutiérrez et al. 2020; Sheridan 2018), a reliable method suitable for operational purposes is still lacking (Yang and Tsai 2019). Furthermore, the sampling and reporting protocols for operational and archived wind measurements present significant obstacles for the verification of peak gust forecasts (Harris and Kahl 2017). These protocols require peak gust values to be reported only when certain criteria are satisfied, for example, that wind gusts must exceed lulls by at least 10 kt (1 kt ≈ 0.5 m s−1) (https://w1.weather.gov/glossary/index.php?word=gust), with the reported peak gusts corresponding to measurements made during only 10 min, rather than the full 60 min, of each hour (NOAA 1998).
Studies have shown that peak gusts and sustained wind speeds are related via the gust factor, which is the ratio of peak wind gust to sustained wind speed (Yang and Tsai 2019; Sherlock 1952). Gust factors are related to turbulence intensity and roughness of surrounding terrain, and are highly site-specific (e.g., Harper et al. 2010). When multiplied by a forecast wind speed, gust factors can provide a simple method for predicting peak wind gusts.
Wind conversion, the practice of converting between wind speeds averaged over different time periods for the purpose of standardizing or comparing wind impacts in applications such as tropical cyclones and wind energy, has long utilized gust factors (e.g., Masters et al. 2010; Harper et al. 2010). Gust factors have also been used to forecast peak wind gusts (e.g., Weggel 1999). The U.S. Air Force, for example, utilizes a standard gust factor of 1.4 for predicting nonconvective wind gusts (U.S. Air Force 2019). Other studies have found a mean gust factor for tropical cyclones ranging from 1.4 to 1.59 (Durst 1960; Krayer and Marshall 1992; Paulsen and Schroeder 2005) and a mean gust factor for extratropical cyclones of 1.35–1.44, depending on surface roughness (Paulsen and Schroeder 2005).
Recently, Harris and Kahl (2017) showed that gust factors are sensitive to various meteorological variables, and that gust factors stratified by wind speed (0–5, 5–10, 10–15, and >15 kt) and wind direction (30° bins) offer the best potential for predicting peak gusts. This stratification is referred to as “meteorologically stratified.” They introduced the meteorologically stratified gust factor (MSGF) model, in which these site-specific, wind speed– and wind direction–dependent gust factors may be combined with wind forecasts to predict peak gusts. The MSGF model was shown to be skillful in predicting peak gusts out to 72 h when combined with model output statistics (MOS) wind guidance (Kahl 2020). Site-specific gust factors at a large number of locations across the United States were subsequently made available to facilitate the operational use of the MSGF model (Kahl et al. 2021).
It remains to be seen, however, how well the MSGF model performs during specific types of gust-producing weather conditions. In the present study, we extend the work of Kahl (2020) by evaluating MSGF model performance during several types of gust-producing weather phenomena. Specifically we address the following research questions. Is MSGF model performance sensitive to the meteorological scenario under which the gusts were produced? How is model performance for several specific, gust-producing weather conditions affected by MOS type, forecast projection, and site location? Our analysis refines the performance characteristics of the MSGF model, with implications for its operational use in predicting peak wind gusts.
2. Data and methods
The data and methods utilized to conduct this study were adapted from Kahl (2020) and are reviewed briefly here, with additions noted.
a. Wind and gust data
The 1-min meteorological data were retrieved from the U.S. Automated Surface Observing System (ASOS) for 2010–17 at 16 locations (Fig. 1). The 16 ASOS locations reflect a wide array of climates, elevations, and topographic influences (Table 1). From the 1-min ASOS data, mean wind, wind direction, and peak wind gust, reported in integer values of knots (kt, defined in section 1), were assessed and compiled into hourly records. Standard hourly ASOS reports were not considered due to restrictive sampling and reporting protocols limiting wind speed to only 2 min and wind gust, if reported, to only 10 min of the hour, thus ignoring wind data outside of those periods (Kristensen et al. 1991; Harris and Kahl 2017). The 1-min data underwent quality control assurances prior to being added to the hourly dataset, including removal of repeated entries, ensuring chronological order, removal of erroneous measurements (example: unrealistically positive or negative wind speeds), and records that lacked necessary fields. Remaining meteorological data were then compiled into an hourly dataset, requiring a minimum of 54 quality-controlled 1-min records for each hour. The hourly ASOS records were utilized to prepare meteorologically stratified gust factors (section 2b) for use in the MSGF model and as verification for predicted peak gusts (section 2e).
Locations of ASOS stations utilized to evaluate the meteorologically stratified gust factor model.
Citation: Weather and Forecasting 37, 8; 10.1175/WAF-D-21-0201.1
Location and database details. Locations assessed are those from Kahl (2020), with a second site in Wisconsin included because of local interest to the authors. Here, ID indicates the station identifier code and MSL indicates above mean sea level.
b. Meteorologically stratified gust factors
Meteorologically stratified gust factors for KATL, one of the sites utilized in the present study, are shown in Fig. 2. The KATL ASOS station is situated on the southwest corner of the Atlanta International Airport in Georgia. For winds greater than 10 kt (the outermost two rings in the gust web diagram), the smallest gust factors of 1.55–1.71 are observed from 300° clockwise to 120°, mostly due to flow over the large, relatively smooth surface of the airport, while larger gust factors occur outside of this sector. Largest gust factors of 1.80–1.86 for winds greater than 10 kt occur from the southeast and southwest, a result of topographic features and surface roughness in these upwind sectors.
Gust web diagram depicting MSGFs for KATL based on 65 299 hourly mean wind and peak gust records from 2010 to 2017. The radial lines represent 30° directional boundaries. The rings represent mean hourly wind speed ranges (0–5 kt in the center, 5–10 kt, 10–15 kt, and 15 kt or greater on the outermost ring). The numbers shown are mean gust factors for each wind speed and wind direction combination. Symbols are utilized to denote the frequency of mean wind speed and directional combination. Gust webs for all sites analyzed here are provided in Kahl et al. (2021).
Citation: Weather and Forecasting 37, 8; 10.1175/WAF-D-21-0201.1
c. MOS forecasts
The MOS is a statistical method that uses regression-based techniques to downscale numerical weather prediction forecast for specific locations. Wind speed and direction forecasts were obtained from two widely utilized MOS products, Global Forecast System (GFS) and North American Mesoscale Model (NAM) MOS, as various types of MOS have proven to be effective in objective weather forecasting (Taillardat 2021). The GFS and NAM MOS guidance are initialized every 6 and 12 h, respectively, and include forecast projections at 3-h intervals from 6 to 60 h, as well as 66 and 72 h. The intention of this study was not to evaluate specific MOS forecast products, but rather the performance of the MSGF model when coupled with commonly utilized MOS wind forecasts.
d. Weather type identification
To identify weather conditions that present particular challenges in forecasting peak gusts, an informal survey was distributed to 22 professional meteorologists at selected universities and National Weather Service (NWS) forecast offices, asking under which meteorological situations gust forecasting is the most difficult or important. Fourteen meteorologists from 11 NWS offices and 2 universities across several states responded. While we acknowledge the limitation of the small sample size of survey responses, those who replied cited that wind gust forecasting was most difficult during convective storms, winter storms, low wind speed environments, low-level inversions, nighttime mixing and cooling, rain, and snow showers. Survey responses further indicated that wind gust forecasting was considered most important during blowing snow, strong cyclones, days with elevated fire danger, when wind gusts are near safety limits, and for aviation, marine, and truck transportation.
The responses to this informal survey informed the selection of weather types to include in the study. The seven weather types selected for analysis are listed in Table 2, including one additional gust-producing weather type (high pressure) added after the survey. Hourly Local Climatological Data (LCD), obtained from the U.S. National Centers for Environmental Information (www.ncei.noaa.gov/data/local-climatological-data/access), were utilized to ascertain when the criteria for each weather type were met at each of the 16 sites analyzed. For all hours identified as meeting the criteria of the specific weather types, ASOS wind and gust data were merged with MOS wind speed and wind direction forecasts at projections ranging from 6 to 72 h. The number of hours at each site for which merged data were available for the 24-h forecast projection is illustrated in Table 3. A minimum of 100 h of merged data were required for subsequent analysis. The high pressure, rain without thunder, and night weather types presented sufficient sample sizes at most sites, while the snow and rain with thunder events occurred relatively infrequently.
Weather types and identification criteria. The 17-kt wind speed criterion is intended to be sufficiently windy to represent winter storms without having an overly negative impact on sample size.
Number of hours with merged ASOS wind and gust observations, and NAM and GFS 24-h wind forecasts, i.e., the 24-h projection sample size, for each weather type.
e. Application and evaluation of the MSGF model for specific weather types
Peak gust climatology at KATL, stratified by hour and season during 2010–17. Peak gust climatologies for all sites analyzed here are provided in Kahl et al. (2021).
Citation: Weather and Forecasting 37, 8; 10.1175/WAF-D-21-0201.1
Verification metrics of bias (gustfcst − gustobs) and absolute error (|bias|) were utilized to evaluate the MSGF model, where gustobs is the observed hourly peak wind gust determined from the 1-min ASOS data, and gustfcst is the predicted gust according to Eq. (2). These metrics were found for the MSGF model using NAM and GFS MOS wind forecasts at all projections, as well as for PPROG and climatology forecasts, during all hours for which an identified weather type occurred.
Skill was evaluated by comparing absolute error distributions of peak gusts predicted using MOS wind forecasts with those of climatology forecasts, via the sign test (Mendenhall et al. 1990). Differences were considered statistically significant when the null hypothesis, that there is no difference between the two distributions, was rejected at the 0.05 confidence level. A model was considered skillful if its absolute error distribution was significantly different than that of climatology, with a smaller mean value.
3. Results and discussion
a. Climatology forecasts
Climatology forecasts of peak wind gusts represent the benchmark against which MSGF model skill is assessed. MAEs for the no-skill climatology forecasts are presented in Table 4. MAEs were smallest for the high pressure and night clear weather types, ranging from 2 to 5 kt, and are generally consistent with those reported by Kahl (2020) whose analysis did not discriminate among specific weather types. Climatology forecast errors were particularly small for KLAX where the MAE was near 2 kt for these weather types, likely due to frequent onshore, steady winds (Kahl 2020). The largest MAEs were observed for the rain and windy snow weather types, with MAEs generally 5–10 kt and as large as 16.4 kt for windy snow at KGFK.
Mean absolute errors (kt) in climatological gust forecast for all sites. Here, em dashes indicate that the sample size was insufficient to be considered for the weather type at the given location. Values in parentheses represent the percent frequency of weather type occurrence at each location.
Irrespective of MAE magnitude, climatology forecast error was often directionally dependent. For example, the MAE in climatology forecasts at KCDC for the night overcast weather type was 5.1 kt, with much larger errors in the 180°–240° wind direction sector than in other sectors (Fig. 4a). During rain with thunder events at KCNK, where the MAE was nearly 2 times as large at 9.7 kt, little directional dependence in climatology forecast performance was noted (Fig. 4b). We note also that some weather types, particularly rain with thunder and the two snow types, only occur a small fraction of the time (Table 4) and thus are not likely to be adequately represented by the climatology forecasts.
Bias (kt) in climatology peak gust forecasts for (a) night overcast at KCDC and (b) rain with thunder at KCNK.
Citation: Weather and Forecasting 37, 8; 10.1175/WAF-D-21-0201.1
b. Perfect prog model performance
The perfect prog (PPROG) model uses observed wind speed and direction data in Eq. (2) rather than to forecast wind variables. Note that the wind direction appears indirectly in Eq. (2), as it informs the selection of the appropriate meteorologically stratified (i.e., wind direction- and wind speed-dependent) gust factor. PPROG model performance illustrates the performance potential of the MSGF model in predicting peak gust provided that accurate wind speed and direction forecasts are available (Harris and Kahl 2017).
MAEs for PPROG forecasts for all weather types except rain with thunder ranged from 1.1 to 2.6 kt (Table 5), demonstrating that the MSGF model is a viable means of predicting peak wind gusts given the availability of accurate wind forecasts. These small MAEs are observed at all sites, consistent with previous work showing upstream terrain to be the dominant driver for nonconvective wind flow characteristics (Harper et al. 2010; Masters et al. 2010). MAEs for the rain with thunder weather type were larger, as high as 4.7 kt at all but one site. These larger MAE values are likely associated with gust-producing processes not associated with terrain roughness, such as convection, evaporative cooling and downdrafts.
Mean absolute errors (kt) in gust forecast using PPROG wind forecast for all sites. Em dashes have the same meaning as in Table 4.
Overall, the small PPROG MAEs are consistent with those reported by Kahl (2020), implying that gustiness is largely determined by turbulent structures associated with roughness of the surrounding terrain. The higher MAEs for the rain with thunder weather type suggest that the MSGF model may be unable to resolve the mechanisms that govern gustiness associated with convective precipitation.
c. Peak gust forecast performance at different forecast projections
The MSGF model was evaluated at each site, for each of the seven gust-producing weather types, using NAM and GFS MOS wind guidance at projections ranging from 6 to 72 h.
We first present MAE results for the MSGF model coupled with NAM and GFS wind forecasts, during all-weather types and at projections from 6 to −72 h, at station KATL (Fig. 5). Also included in Fig. 5 are MAEs corresponding to the PPROG and climatology forecasts. MAEs were smallest for the high pressure, night clear, and night overcast weather types, ranging from 2.5 to 4 kt, with errors generally increasing with longer forecast projections (Fig. 5). The MSGF peak gust forecasts for these weather types were skillful, with MAEs smaller than no-skill climatology, with statistical significance, at the full range of forecast projections. The largest MAEs were observed for the rain with thunder weather type throughout the entire range of forecast projections and remained relatively consistent in magnitude at 6–7 kt. Statistical analysis failed to identify significant differences between MAE distributions of the MSGF model when paired with GFS versus NAM guidance.
Mean absolute error of MSGF peak gust forecasts at KATL, using GFS and NAM wind guidance at projections from 6 to 72 h, for all nonsnow weather types. Perfect prog (PPROG) and climatology (CLIM) forecast MAEs are indicated by dashed and dotted lines, respectively.
Citation: Weather and Forecasting 37, 8; 10.1175/WAF-D-21-0201.1
Model evaluation results comparable to those presented in Fig. 5 at other sites (not shown) similarly demonstrated the smallest MAEs for high pressure, night clear and night overcast weather types, with MAEs generally increasing with longer forecast projections.
MSGF model performance for the less windy snow weather type is shown in Fig. 6, for the 9 sites that met the minimum data requirements. (Windy and less windy snow weather types rarely occurred at KATL, a southern U.S. location, and are not included in Fig. 5 or 6) Overall, most sites illustrate MAE ranging between 2 and 5 kt. Mean errors generally increase with longer forecast projections but remain less than climatology out to 72-h projections at most sites. MAEs were largest for KCDC and KLND, with values of 5–7 kt, and were not skillful relative to climatology. Results at KMDH were limited to the GFS guidance through the 33rd projected forecast hour because of sample size constraints.
Mean absolute error of MSGF peak gust forecasts at all sites, using GFS and NAM wind guidance at projections from 6 to 72 h, for the less-windy snow weather type. Perfect prog (PPROG) and climatology (CLIM) forecast MAEs are indicated by dashed and dotted lines, respectively.
Citation: Weather and Forecasting 37, 8; 10.1175/WAF-D-21-0201.1
d. 24-h forecast skill
We now present MSGF model performance statistics at one particular forecast projection, 24 h, as a means to evaluate and compare peak gust predictions at all sites across all weather types examined.
MAEs at all sites ranged from 2 to 6 kt, for all weather types except rain with thunder (Table 6). Smallest values were associated with weather types of night clear, night overcast, and high pressure. MAEs during rain with thunder were larger, generally 6–8 kt, and at most sites were not significantly better than climatology.
Mean absolute errors (kt) for each weather type for MSGF peak gust forecasts at 24 h using wind forecasts provided by NAM and GFS MOS guidance. Boldface entries denote skill at the 95% confidence level when compared with climatology. Values are only shown when least 100 h of merged forecast and observed gusts were available.
There is no clear difference in performance when the MSGF model is coupled with NAM versus GFS MOS wind guidance. In a previous evaluation, Kahl (2020) noted that the MSGF model performed better when coupled with GFS MOS winds. The Kahl study did not distinguish between different weather types, so model evaluations reported in its findings were based on larger sample sizes. The relative frequencies of specific weather types for several study locations may account for the observed consistency in performance between the two MOS types.
The MSGF model coupled with NAM and GFS MOS 24-h wind forecasts generally demonstrate skill at most sites for all weather types except rain with thunder. Exceptions were noted at KLND, a site located in mountainous terrain where larger errors were consistently observed across all weather types (Table 6). As noted in the earlier evaluation by Kahl (2020), peak gusts may be poorly predicted by the MSGF model in areas with intricate topographical influences. Model skill was also generally lacking at KLAX, where more accurate climatology forecasts (Table 4) render skill more difficult to attain, and at KROW, a particularly gusty site (Kahl et al. 2021) with complex terrain, where PPROG model errors are larger (Table 5).
MAEs for the rain with thunder weather type, where the MSGF model was generally not skillful, were highest at KCNK (Table 6). Despite the larger errors at this site, the MSGF model displayed skill due to large climatology errors (Table 4) for this weather type. A similar result of a relatively large MAE nevertheless showing skill was observed at KMDH when the MSGF model was coupled with GFS MOS wind guidance, and at KBUF and KGFK for the windy snow weather type.
MSGF model performance was also analyzed relative to wind speed. For some sites there was a slight negative bias at larger wind speeds (not shown), although this was not uniformly observed across all sites and weather types.
e. Longest forecast projection
The longest forecast projection (LFP) is the projected forecast up to which the MSGF model shows skill at all previous forecast projections (Kahl 2020). The LFP is presented as a means of assessing the quality of peak gust predictions over the full range of available forecast projections. LFPs for the high pressure, night clear, and night overcast weather types were 72 h at 8 of the 16 sites analyzed (Table 7), indicating skill in peak gust forecasts at all projections from 6 to 72 h. At the other sites the MSGF exhibited skill for a shorter range of forecast projections, with LFP values ranging from 6 to 60 h.
Longest forecast projection (h) for the MSGF model coupled with NAM and GFS MOS wind guidance at each site during specific weather types. LFP is the hour at which the model demonstrates skill for the indicated forecast projection as well as for all previous forecasts projections prior to the LFP. Here, an em dash indicates insufficient sample size (N < 100) to perform the analysis. Blank entries indicate a lack of skill at the smallest projection (6 h).
For the rain without thunder and less windy snow weather events, the LFP ranged from 33 to 72 h at all but three sites (KSFF, KLAX, and KCDC), where LFP values were as low as 6 h. Model evaluation during these weather events was hampered by sample size constraints (Table 3), particularly for less windy snow. Reduced sample sizes during the remaining weather types, rain with thunder and windy snow, further limited analysis to only four sites, exhibiting inconsistent LFP results of 6, 66, and 72 h.
The LFP results are largely consistent with those presented in the earlier analysis of MSGF performance without attention to weather type (Kahl 2020). However, model evaluation results restricted to specific gust-producing weather types (the present study) provide evidence that peak gusts can be skillfully predicted by the MSGF model at multiday forecast projections during high pressure, night clear, night overcast, rain without thunder, and less windy snow conditions. The LFP results suggest that the model does not perform skillfully during rain with thunder events. Results are promising for windy snow events, but evaluations were sparse because of limited sample size.
4. Summary and conclusions
Previous research has shown that the MSGF model demonstrates skill in forecasting peak wind gusts (Kahl 2020), and the site-specific gust factors necessary for operational use of the MSGF model have been provided for a large number of locations across the United States (Kahl et al. 2021). Model performance during specific gust-producing weather types, however, had not been investigated. To remedy this, the present study investigated how the meteorologically stratified gust factor method, coupled with MOS wind guidance, performed during seven gust-producing weather types at 16 diverse locations throughout the United States.
Using the MSGF model, peak wind gusts were predicted by multiplying a site-specific, wind speed- and wind direction-dependent gust factor by a wind speed forecast. A perfect prognosis (PPROG) approach, which uses observed wind speed and directional data in the MSGF model (rather than wind forecasts), yielded small MAEs, generally 1–2 kt for all sites and most weather conditions. The PPROG results confirm the potential of the MSGF model as a viable, operational method of forecasting peak wind gusts during these weather types, provided that accurate wind speed and direction forecasts are available. An exception was noted for PPROG performance during rain with thunder events, where MAEs at most sites were larger at 3–4 kt.
MSGF model predictions coupled with actual wind forecasts, specifically NAM and GFS MOS guidance at forecast projections ranging from 6 to 72 h, yielded MAEs ranging from 2.5 to 4 kt for high pressure, night clear, and night overcast weather types. These predictions were generally skillful relative to climatology at all projections. Model performance was substantially reduced during rain with thunder events, with larger, nonskillful MAEs of 6–7 kt. Differences in performance were typically not observed when the MSGF model was coupled with NAM versus GFS MOS wind forecasts.
Detailed evaluations of model performance using wind guidance at the 24-h forecast projection yielded the following findings:
-
MAEs in peak gust predictions ranged from 2 to 6 kt at all sites for all weather types except rain with thunder, where mean errors as large as 8.4 kt were observed.
-
The MSGF model performed best (smallest MAEs) for night clear, night overcast, and high pressure weather types.
-
The largest MAE, indicating the poorest model performance, was observed for the “rain with thunder” weather type.
-
The MSGF peak gust forecasts were generally skillful for all weather types except rain with thunder.
-
No clear difference in model performance was observed when the MSGF model was coupled with NAM versus GFS wind guidance.
A further evaluation of the MSGF model investigated the LFP, the longest projection up to which the MSGF model consistently shows skill at all projections. LFPs for the high pressure, night clear, and night overcast weather types were 72 h at 8 of the 16 sites analyzed. At the other sites and for the other weather types, the LFP ranged from 6 to 60 h. LFP values for the two snow weather types are promising (6–72 h), but because of limited sample size, these results were only available at a minority of sites analyzed.
Although wind gust prediction continues to be a challenge, this analysis confirms the earlier findings of Kahl (2020) that the MSGF model is a viable tool for the operational prediction of peak wind gusts out to projections of 72 h. The present evaluation provides important context for performance characteristics when the model is applied during specific types of gust-producing weather phenomena. Determination of site-specific gust factors at additional locations beyond those provided by Kahl et al. (2021), along with incorporation of these data into model postprocessing software, would make the method more accessible to operational meteorologists. Moreover, future studies can provide additional operational context by evaluating MSGF model performance within different spatial sectors or evolutionary stages of specific synoptic flow regimes or regime transitions.
Acknowledgments.
Support for Brandon Selbig was kindly provided by the University of Wisconsin–Milwaukee’s Office of Undergraduate Research.
Data availability statement.
All 1-min ASOS wind observations utilized during this study are openly available from the National Centers for Environmental Information dataset 6405 (ftp.ncdc.noaa.gov/pub/data/asos-onemin) as cited in Kahl (2020). Model output statistics (MOS) data utilized in this study are openly available from Iowa State University (https://mesonet.agron.iastate.edu/mos/) as cited in Kahl (2020). All Local Climatological Data (LCD) utilized in this study are openly available from the U.S. National Centers for Environmental Information (www.ncei.noaa.gov/data/local-climatological-data/access). Gust web diagram datasets utilized for this research are included in Kahl et al. (2021).
REFERENCES
Adame, J. A., L. Lope, P. J. Hidalgo, M. Sorribas, I. Gutiérrez-Álvarez, A. del Águila, A. Saiz-Lopez, and M. Yela, 2018: Study of the exceptional meteorological conditions, trace gases and particulate matter measured during the 2017 forest fire in Doñana Natural Park, Spain. Sci. Total Environ., 645, 710–720, https://doi.org/10.1016/j.scitotenv.2018.07.181.
Caban, J., A. Nieoczym, and L. Gardynski, 2021: Strength analysis of a container semi-truck frame. Eng. Fail. Anal.,127, 105487, https://doi.org/10.1016/j.engfailanal.2021.105487.
Changnon, S. A., 2009: Temporal and spatial distributions of wind storm damages in the United States. Climatic Change, 94, 473–482, https://doi.org/10.1007/s10584-008-9518-6.
De Petris, S., F. Sarvia, M. Gullino, E. Tarantino, and E. Borgogno-Mondino, 2021: Sentinel-1 polarimetry to map apple orchard damage after a storm. Remote Sens., 13, 1030, https://doi.org/10.3390/rs13051030.
Durst, C., 1960: Wind speeds over short periods of time. Meteor. Mag., 89, 181–187.
Giess, P., A. J. H. Goddard, and G. Shaw, 1997: Factors affecting particle resuspension from grass swards. J. Aerosol Sci., 28, 1331–1349, https://doi.org/10.1016/S0021-8502(97)00021-9.
Grubisic, V., and Coauthors, 2008: The Terrain-Induced Rotor Experiment: A field campaign overview including observational highlights. Bull. Amer. Meteor. Soc., 89, 1513–1533, https://doi.org/10.1175/2008BAMS2487.1.
Gutiérrez, A., C. Porrini, and R. G. Fovell, 2020: Combination of wind gust models in convective events. J. Wind Eng. Ind. Aerodyn., 199, 104118, https://doi.org/10.1016/j.jweia.2020.104118.
Harper, B. A., J. D. Kepert, and J. D. Ginger, 2010: Guidelines for converting between various wind averaging periods in tropical cyclone conditions. WMO/TD-1555, World Meteorological Organization, 64 pp., https://library.wmo.int/doc_num.php?explnum_id=290.
Harris, A., and J. Kahl, 2017: Gust factors: Meteorologically stratified climatology, data artifacts, and utility in forecasting peak gusts. J. Appl. Meteor. Climatol., 56, 3151–3166, https://doi.org/10.1175/JAMC-D-17-0133.1.
Hart, N. C. G., S. L. Gray, and P. A. Clark, 2017: Sting-jet windstorms over the North Atlantic: Climatology and contribution to extreme wind risk. J. Climate, 30, 5455–5471, https://doi.org/10.1175/JCLI-D-16-0791.1.
Hewson, T. D., and U. Neu, 2015: Cyclones, windstorms and the IMILAST project. Tellus, A67, 27128, https://doi.org/10.3402/tellusa.v67.27128.
Jung, C., and D. Schindler, 2019: Historical winter storm atlas for Germany (GeWiSA). Atmosphere, 10, 387, https://doi.org/10.3390/atmos10070387.
Jung, C., D. Schindler, A. Buchholz, and J. Laible, 2017: Global gust climate evaluation and its influence on wind turbines. Energies, 10, 1474, https://doi.org/10.3390/en10101474.
Kahl, J. D. W., 2020: Forecasting peak wind gusts using meteorologically stratified gust factors and MOS guidance. Wea. Forecasting, 35, 1129–1143, https://doi.org/10.1175/WAF-D-20-0045.1.
Kahl, J. D. W., B. R. Selbig, and A. R. Harris, 2021: Meteorologically stratified gust factors for forecasting peak wind gusts across the United States. Bull. Amer. Meteor. Soc., 102, E1665–E1671, https://doi.org/10.1175/BAMS-D-21-0013.1.
Kim, S.-J., J.-H. Shim, and H.-K. Kim, 2020: How wind affects vehicles crossing a double-deck suspension bridge. J. Wind Eng. Ind. Aerodyn., 206, 104329, https://doi.org/10.1016/j.jweia.2020.104329.
Krayer, W. R., and R. D. Marshall, 1992: Gust factors applied to hurricane winds. Bull. Amer. Meteor. Soc., 73, 613–618, https://doi.org/10.1175/1520-0477(1992)073<0613:GFATHW>2.0.CO;2.
Kristensen, L., M. Casanova, M. Courtney, and I. Troen, 1991: In search of a gust definition. Bound.-Layer Meteor., 55, 91–107, https://doi.org/10.1007/BF00119328.
Lei, Y., Y. Huang, and H. Wang, 2021: Effects of wind disturbance on the aerodynamic performance of a quadrotor MAV during hovering. J. Sens., 2021, 6681716, https://doi.org/10.1155/2021/6681716.
Luchetti, N. T., K. Friedrich, and C. E. Rodell, 2020: Evaluating thunderstorm gust fronts in New Mexico and Arizona. Mon. Wea. Rev., 148, 4943–4956, https://doi.org/10.1175/MWR-D-20-0204.1.
Manasseh, R., and J. H. Middleton, 1999: The surface wind gust regime and aircraft operations at Sydney Airport. J. Wind Eng. Ind. Aerodyn., 79, 269–288, https://doi.org/10.1016/S0167-6105(97)00293-6.
Masters, F. J., P. J. Vickery, P. Bacon, and E. N. Rappaport, 2010: Toward objective, standardized intensity estimates from surface wind speed observations. Bull. Amer. Meteor. Soc., 91, 1665–1682, https://doi.org/10.1175/2010BAMS2942.1.
Mendenhall, W., D. D. Wackerly, and R. L. Schaeffer, 1990: Mathematical Statistics with Applications. 4th ed. PWS-Kent Publishing, 818 pp.
Montenegro, P. A., R. Heleno, H. Carvalho, R. Calçada, and C. J. Baker, 2020: A comparative study on the running safety of trains subjected to crosswinds simulated with different wind models. J. Wind Eng. Ind. Aerodyn., 207, 104398, https://doi.org/10.1016/j.jweia.2020.104398.
NOAA, 1998: Automated Surface Observing System (ASOS) user’s guide. NOAA Tech. Doc., 74 pp., https://www.weather.gov/media/asos/aum-toc.pdf.
OSHA, 2021: Standard 1917.45—Cranes and derricks. U.S. Department of Labor, Occupational Safety and Health Administration, accessed 24 February 2022, https://www.osha.gov/laws-regs/regulations/standardnumber/1917/1917.45.
Paulsen, B. M., and J. L. Schroeder, 2005: An examination of tropical and extratropical gust factors and the associated wind speed histograms. J. Appl. Meteor., 44, 270–280, https://doi.org/10.1175/JAM2199.1.
Pinto, P., and M. Belo-Pereira, 2020: Damaging convective and non-convective winds in southwestern Iberia during Windstorm Xola. Atmosphere, 11, 692, https://doi.org/10.3390/atmos11070692.
Pirooz, A. A. S., S. Moore, R. Turner, and R. G. J. Flay, 2021: Coupling high-resolution numerical weather prediction and computational fluid dynamics: Auckland Harbour case study. Appl. Sci., 11, 3982, https://doi.org/10.3390/app11093982.
Sheridan, P., 2018: Current gust forecasting techniques, developments, and challenges. Adv. Sci. Res., 15, 159–172, https://doi.org/10.5194/asr-15-159-2018.
Sherlock, R. H., 1952: Variation of wind velocity and gusts with height. Proc. Amer. Soc. Civ. Eng., 78, 1–24; 79, 1–20.
Sinden, G., 2007: Characteristics of the U.K. wind resource: Long term patterns and relationship to electricity demand. Energy Policy, 35, 112–127, https://doi.org/10.1016/j.enpol.2005.10.003.
Siti, I., M. Mjahed, H. Ayad, and A. El Kari, 2019: New trajectory tracking approach for a quadcopter using genetic algorithm and reference model methods. Appl. Sci., 9, 1780, https://doi.org/10.3390/app9091780.
Su, N., S. Peng, and N. Hong, 2021: Stochastic dynamic transient gusty wind effect on the sliding and overturning of quayside container cranes. Struct. Infrastruct. Eng., 17, 1271–1283, https://doi.org/10.1080/15732479.2020.1809465
Suomi, I., T. Vihma, S.-E. Gryning, and C. Fortelius, 2013: Wind-gust parametrizations at heights relevant for wind energy: A study based on mast observations. Quart. J. Roy. Meteor. Soc., 139, 1298–1310, https://doi.org/10.1002/qj.2039.
Taillardat, M., 2021: Skewed and mixture of Gaussian distributions for ensemble postprocessing. Atmosphere, 12, 966, https://doi.org/10.3390/atmos12080966.
Teoh, Y. E., A. Alipour, and A. Cancelli, 2019: Probabilistic performance assessment of power distribution infrastructure under wind events. Eng. Struct., 197, 109199, https://doi.org/10.1016/j.engstruct.2019.05.041.
Thalla, O., and S. C. Stiros, 2018: Wind-induced fatigue and asymmetric damage in a timber bridge. Sensors, 18, 3867, https://doi.org/10.3390/s18113867.
U.S. Air Force, 2019: Air Force Handbook 15-101; Weather Meteorological Techniques. Air Force Handbook AFH15-101, 248 pp., https://static.e-publishing.af.mil/production/1/af_a3/publication/afh15-101/afh15-101.pdf.
Usbeck, T., T. Wohlgemuch, C. Pfister, R. Volz, M. Beniston, and M. Dobbertin, 2010: Wind speed measurements and forest damage in Canton Zurich (central Europe) from 1891 to winter 2007. Int. J. Climatol., 30, 347–358, https://doi.org/10.1002/joc.1895.
Valta, H., I. Lehtonen, T. K. Laurila, A. Venäläinen, M. Laapas, and H. Gregow, 2019: Communicating the amount of windstorm induced forest damage by the maximum wind gust speed in Finland. Adv. Sci. Res., 16, 31–37, https://doi.org/10.5194/asr-16-31-2019.
Van Den Bossche, N., M. A. Lacasse, and A. Janssens, 2013: A uniform methodology to establish test parameters for watertightness testing. Part I: A critical review. Build. Environ., 63, 145–156, https://doi.org/10.1016/j.buildenv.2012.12.003.
Weggel, J. R., 1999: Maximum daily wind gusts related to mean daily wind speed. J. Struct. Eng., 125, 465–468, https://doi.org/10.1061/(ASCE)0733-9445(1999)125:4(465).
Welker, C., O. Martius, P. Stucki, D. Bresch, S. Dierer, and S. Bronnimann, 2016: Modelling economic losses of historic and present-day high-impact winter windstorms in Switzerland. Tellus, 68A, 29546, https://doi.org/10.3402/tellusa.v68.29546.
Wong, C. J., and M. D. Miller, Eds., 2010: Guidelines for Electrical Transmission Line Structural Loading. ASCE, 204 pp.
Yang, T.-H., and C.-C. Tsai, 2019: Using numerical weather model outputs to forecast wind gusts during typhoons. J. Wind Eng. Ind. Aerodyn., 188, 247–259, https://doi.org/10.1016/j.jweia.2019.03.003.
Zhang, Y., and L. Li, 2021: Crosswind stability of metro train on a high-pier viaduct under spatial gust environment in mountain city. KSCE J. Civ. Eng., 25, 4661–4670, https://doi.org/10.1007/s12205-021-0706-5.
