Tropical Storm Hermine Follow Up with Lessons Learned, Wind Gusts and Gust Factors


Four quadrant wind radii from the raw TCM guidance from the National Hurricane Center for Hermine issued at 11 AM EDT 01 September 2016.

To evaluate the performance of CSTAR related research to operations activities, members of the CSTAR Tropical Cyclone (TC) wind team recently discussed experiences with Tropical Storm Hermine (September 2016).  In addition, the sustained winds, wind gusts, and gust factors for Tropical Storm Hermine were examined across coastal and eastern Georgia, South Carolina, North Carolina and Virginia.

A conference call was conducted with team members from NWS offices in Charleston, Columbia, Newport, Raleigh, Wakefield, and Wilmington along with NWS Eastern Region SSD to examine the event in greater detail and to identify ways to improve the process over both the near and longer term. Hermine had some special circumstances that made the use of the TCM wind guidance more difficult than normal, including the TC’s weak intensity, the limited wind radii on the western semi-circle of the storm, and the fact that the storm was undergoing extratropical transition as it began to exit the region. The CSTAR TCM wind technique likely won’t be able to completely mitigate these issues. However, a couple of other issues were noted and some solutions are being developed to handle these items:
1)    Adjustments are being made to the CSTAR_Wind_from_WindReductionFactor smart tool so that the wind reduction is only applied to winds of 34 kts or greater (inside the TCM envelope).
2)    CSTAR TCM wind technique offices will explore the use of brief conference calls following the NHC coordination call, to discuss the PREAT guidance and ensure offices are using the technique appropriately and are well collaborated.

NWS Raleigh volunteer Victoria Oliva, examined the sustained winds, wind gusts, and gust factors for Hermine across coastal and eastern Georgia, South Carolina, North Carolina and Virginia. Hourly observations of winds and wind gusts from 48 regular ASOS or AWOS METAR locations impacted by the over land wind field associated with Tropical Storm Hermine were examined. The locations examined in this analysis extended from KSAV (Savanah, Georgia) northeast along and just inland of the coast to KWAL (Wallops Island, Virginia). The map below is a subjective analysis of the maximum wind gusts observed during Tropical Storm Hermine across North Carolina.


Subjective analysis of the maximum wind gusts (MPH) observed across North Carolina from 02 September to 03 September 2016 during Tropical Storm Hermine.

Only observations from routine hourly METARs were used (special observations and observations not at the top of the hour were excluded). In addition, gust factors were only calculated for sustained winds of 10 MPH or greater. For each observation, the hourly wind gust factor was computed. The gust factor is defined as the ratio between the wind gust of a specific duration to the mean (sustained) wind speed for a period of time.  A total of 860 gust factors were computed for Hermine.

In general, Hermine was characterized by modest sustained winds as it moved across the region although the strength of the wind gusts increased as the storm evolved during extratropical transition as it moved across North Carolina and off the coast. The maximum sustained wind contained in the Hermine data set was only 44 MPH and the maximum wind gust was only 62 MPH. The data set contained a large number of lower end sustained winds and wind gusts. Nearly 82% or 705 out of the 860 observations, contained in this data-set had sustained winds less than 25 MPH.  Only 64 observations, or less than 8% of all observations, had sustained winds of 30 MPH or more and only 3 observations had sustained winds of 40 MPH or more.


Scatter plot of sustained winds and gust factors for Tropical Storm Hermine for 48 METAR locations across coastal and eastern Georgia, South Carolina, North Carolina and Virginia

The chart to the right is a scatter plot of the sustained winds in MPH versus gust factors for the 860 observations included in the study along with a best fit regression curve (y = -0.375ln(x) + 2.6175). In general, the chart demonstrates an inverse relationship between the wind speed and gust factor. Not surprisingly, the gust factors with Hermine were rather variable at low sustained wind speeds and generally converged and decreased with increasing sustained wind speed. This chart is very similar to the database of 15 storms used to develop the CSTAR TCM wind technique.


Histogram of gust factors from Tropical Storm Hermine for 48 METAR locations across coastal and eastern Georgia, South Carolina, North Carolina and Virginia.

A histogram of the frequency of gust factors for Hermine is shown to the right.  The average gust factor for Hermine was 1.58 which is somewhat higher than the average of 1.53 for the database of 15 tropical cyclones used to develop the CSTAR TC wind technique. The histogram data noted that the gust factors were most frequently noted in both of the bins between 1.4 and 1.5 as well as 1.5 and 1.6. Given Hermine’s relatively weak winds, it’s not surprising that the distribution of gust factors for Hermine is shifted to the right when compared to the 15 storm database, while the overall pattern and character of the distribution for Hermine is very similar to the 15 storm study.


Comparison of gust factor regression equations associated with Tropical Storm Hermine and the 15 storm database used to develop the CSTAR TCM wind technique.

We generated a regression equation using the gust factor data set associated with Tropical Storm Hermine and compared it to the regression equation used in the 15 storm database used to develop the CSTAR TCM wind technique. The figure above  compares the Hermine regression equation to the 15 storm equation. They show a similar trend and generally match each other well, but given the modest wind speeds with Hermine, it’s not surprising that the Hermine curve is steeper than the larger dataset.

To summarize, participants from the six WFOs using the CSTAR TCM Wind technique discussed the techniques performance during Tropical Storm Hermine, and while the feedback received noted that the process seemed to work well and forecasts were generally accurate and well collaborated, some issues were identified. Adjustments to the CSTAR_Wind_from_WindReductionFactor smart tool are being made so that the wind reduction is only applied to winds of 34 kts or greater.  In addition, CSTAR TCM wind technique offices will try to use brief conference calls to discuss the PREAT guidance and ensure offices are using the technique appropriately and are well collaborated. Finally, despite the modest strength of the tropical storm and its extratropical transition as it exited the region, the observed gust factors matched the 15 storm database used to develop the CSTAR TCM wind technique reasonably well and the tools provided good guidance.


Posted in CSTAR, TC Inland and Marine Winds, Uncategorized | Leave a comment

HSLC/SHERB Presentation to NWS Central Region Southeast SOO Group

On August 31st, 2016, I was invited to attend and speak at the NWS Central Region (CR) Southeast SOO Group Meeting at WFO Indianapolis on the topic of our (WFO ILN — Wilmington, OH) experiences/research with non-supercell HSLC tornadoes and the use of the SHERB parameter in our operations to help diagnose environments more prone to producing HSLC tornado and damaging wind events.  The CR Southeast SOO Group consists of SOOs from the following offices in Central Region:

  • Indianapolis
  • Jackson (KY)
  • Louisville
  • Lincoln (IL)
  • Paducah
  • St. Louis
  • Springfield (MO)
  • Kansas City

Before diving into SHERB and environment awareness, I wanted the group to see some of the radar-based research that has been done here at WFO ILN by one of our meteorologists, Andy Hatzos.  Andy has taken  this research to the Ohio State Severe Weather Symposium and the 2016 NWA Annual Meeting.  I presented about half of Andy’s slide deck to the SOOs showing some of the rotational velocities, reflectivity characteristics, and other commonalities and challenges we see in the Ohio Valley, particularly during the cool season. Much of Andy’s work focuses on the use of FAA (TDWR) radar and some of the benefits and challenges of this data.  This slide deck is available here in PDF format.  This slide deck — and the challenges/questions it brings — is important to understand in context of what makes research like SHERB so important. The more situationally aware a meteorologist is to environments that can produce HSLC tornado and damaging wind events, the more effective the messaging will be in all formats — including social media, outlooks, decision-support briefings, and most importantly, warnings.

Andy’s presentation was an excellent segue into the importance of SHERB use in operations. Graciously, Keith Sherburn provided me a wealth of background slides on SHERB development and verification, to which I added some brief case events from the WFO ILN area.  The slide deck I put together from several of Keith’s presentations is available here in PDF format.

The presentations were well received with good dialog and discussion – and I was thrilled to be able to present and discuss SHERB use outside the CSTAR footprint where it was developed.  A hearty thank you to Keith and Andy for allowing me the opportunity to present their outstanding work.      — Seth Binau (WFO ILN SOO)

Posted in Convection, High Shear Low Cape Severe Wx | Leave a comment

Forecasts of Winds and Wind Gusts Associated with Tropical Cyclone Hermine Using CSTAR Research

The approach of Tropical Cyclone (TC) Hermine provides WFOs in the Carolinas and Virginia with another opportunity to use and evaluate a technique to forecast tropical cyclone winds and wind gusts. A Collaborative Science, Technology, and Applied Research (CSTAR) project with North Carolina State University and over a half dozen WFOs in the Southeast, examined ways to add science and to improve inland wind and wind gust forecasts associated with TCs. The CSTAR TC wind technique consists of three primary improvements including a bias correction of the TCM wind vortex, using collaborated wind reductions over land, and using collaborated wind gust factors for wind gust grids across the domain.


One of the primary difficulties forecasters at the local Weather Forecast Office (WFO) experience during TCs is with forecasting wind and wind gusts. This largely originates from the need to downscale the limited spatial and temporal resolution of the guidance provided by the National Hurricane Center (NHC) to the 2.5km resolution gridded forecast that the local WFOs create for the NDFD. The image to the right is a plot of the guidance provided by the NHC that includes wind radii for the four quadrants of the storm for 34kt, 50kt, and 64kt winds. This forecast for Hermine is from 15 UTC on 01 September, 2016 and is derived from the NHC TCM wind product.

The CSTAR TC wind methodology was utilized by several WFOs in the Southeast beginning as early as Tuesday. The new methodology is used in the Gridded Forecast Editor (GFE) and uses two new forecast grids: the Wind Reduction Factor and the Wind Gust Factor. By providing forecasters with the opportunity to vary these two grids both spatially and temporally across the forecast area and allowing the values to be collaborated across WFO boundaries and maintained from shift to shift, improved forecasts are expected. An example of the Wind Reduction Factor (left) and Wind Gust Factor (right) grids used during Hermine at 20 UTC on 31 August, 2016 valid at 02 or 03 UTC on 03 September, 2016 is shown below.hermine.reduction.factor.gust.factor.4pm.0831

Despite some initial technical and procedural issues , the subjective quality and consistency of the wind and wind gusts forecast from the participating WFOs has been quite good during the days leading up to the arrival of Hermine. Initial feedback from forecasters is that the process is improved over the standard methodology through the utilization of more science and much more collaboration. An example of the improved dialog between WFOs is shown in the AWIPS Collaboration tool chat from around 0330 UTC on 31 August in which forecasters are working together to ensure a consistent forecast more than 3 days in advance of the storm.


Finally, I wanted to pass along a couple of images showing the end result of the forecast process. The image below shows the 33-hour forecast of wind (left) and wind gusts (right) from the NDFD valid at 09 UTC on 03 September. In these images, the wind and wind gusts are collaborated across 6 WFOs using the CSTAR methodology. These WFOs had to adjust the forecast for the high bias in the TCM product, inland decay of the TC, varying boundary-layer air masses, a pre-existing frontal zone, complex land-sea boundaries including the NC Sounds and the Chesapeake Bay while collaborating across 8 CWA boundaries. The end result appears to be a well collaborated forecast resulting from the hard work of the forecast staff on duty.


A more formal evaluation of the process ad verification of the wind and wind gust forecast is planned after the storm. We’ll share those results on a future CIMMSE post.

Posted in CSTAR, TC Inland and Marine Winds, Uncategorized | Leave a comment

New Paper Comparing ASOS Near-Surface Winds and WSR-88D-Derived Wind Speeds in Landfalling Tropical Cyclones

Map of ASOS and WSR-88D sites along with tropical cyclone tracks used in the study. 

A new article in Weather and Forecasting by Krupar et al. entitled “A Comparison of ASOS Near-Surface Winds and WSR-88D-Derived Wind Speed Profiles Measured in Landfalling Tropical Cyclones” may be of interest to some collaborators. The researchers made 22 comparisons during landfalling TC events in the Gulf of Mexico to compare low-level WSR-88D wind data with ASOS surface wind and wind gust observations within 10 km of the RDA. The goal was to develop an empirical relationship to relate the TC boundary layer wind to near surface winds. They found that radar based, site-specific linear regression equations using a 0–200-m layer average wind speed can be used to predict the ASOS 10-m standardized mean wind speed while a non-site-specific linear regression model using a VAD 0–500-m mean boundary layer wind can be used to predict ASOS 10-m nonstandardized gust wind speeds.

Posted in TC Inland and Marine Winds, Uncategorized | Leave a comment

GOES-14 will be in Super Rapid Scan Operations Through August 25th


GOES SRSOR visible satellite imagery from 1558 UTC 15 August 2016 in AWIPS

GOES-14 Super Rapid Scan Operations for GOES-R (SRSOR) began on 9 August and will continue for through 25 August 2016. Super Rapid Scan Operations (SRSO) will provide 1-minute imagery to support multiple research and GOES-R/S user readiness activities. The SRSO domain is usually selected a day or two in advance. The domain schedule along with selected imagery from prior days is available at:  Additional background information including training and links to online imagery is available at:

This will be a great opportunity to view the data over our region. NWS forecasters will be able to view some of this data in real-time in AWIPS-2. An example of GOES SRSOR imagery from AWIPS is shown in the image above.

Imagery including visible, infrared, and water vapor is available on the web at the links below…

Posted in Satellite | 2 Comments

Mesonet Data Captures Impressive Nocturnal Surface Theta-E Rise Immediately Ahead of QLCS EF1 Tornado in Ohio (June 23rd, 2016)

During the nighttime hours of June 22nd/23rd, a cluster of severe thunderstorms expanded (QLCS) as it moved from Indiana into Ohio ahead of a seasonably strong shortwave trough shifting through the Great Lakes.


This cluster of storms had previously been responsible for significant/high-end wind gusts across portions of northern Indiana earlier in the night, evolving within and through the heart of a well-predicted SPC Moderate Risk threat area.  As the line moved into the National Weather Service (NWS) Wilmington, Ohio (WFO ILN) forecast area, rapid line-embedded mesovortex evolution began in northern Warren County and continued across much of Clinton County, moving very close to both the WFO ILN WSR-88D, and a Ohio Department of Transportation Road Weather Information System (RWIS) site.  A 20+ mile EF1 tornado was surveyed beginning near Waynesville, OH, to just south of Wilmington, OH, and included a near-miss of both the WFO and the RWIS site, but allowed for some interesting findings.


Detailed data from the RWIS station was sent to WFO ILN by the ODOT RWIS coordinator, Timothy Boyer, and comes courtesy of ODOT.  This data revealed an anomalous mid-summer nocturnal rise in boundary layer theta-e in the hours preceding the QLCS arrival in Warren/Clinton counties.


As background, in our HSLC CSTAR group, Jessica King presented to us the results of her study of numerical simulations of severe weather environments at night in the Ohio Valley/Southeast. The results of her study of severe vs. non-severe events showed that rapid changes in SBCAPE immediately ahead of the line (within the 1-3 hours previous to severe report) was an excellent discriminator vs shear and cooling aloft.  Below is a screen capture of her results – notice in the severe-vs.non-severe plots how SBCAPE increases dramatically in the hour(s) previous to severe events. In seemingly most (>90%) of her simulations, SBCAPE increased by 200-600 J/kg within 2 hours of the severe event.



Courtesy: Jessica King

Here is another image from Jessica’s research showing the surface theta-e field ahead of an idealized convective boundary from her study of these severe cases.  Notice the narrow plume of significant implied advection in theta-e relative to the convective boundary.

Courtesy: Jessica King

With data from the RWIS, and the timeliness of a 05Z special sounding taken at WFO ILN, augmentation of the 05Z surface conditions to what was in place at 07Z (when the tornado touched down and moved through the area) revealed key thermodynamic changes to the environment occurred in the 2.5 hours between 0430Z sounding launch, and tornadogenesis immediately west of Wilmington.  While it’s easy to get caught up with the fact that the RWIS measured an 80 mph gust (image below) as the tornado passed very close by, the surface temperature/dewpoint trends in the hours prior to the tornado are key and critical.  There is a steady and impressive rise in surface theta-e as measured by the site. This is in the depth of night (3 AM EDT) — in summer — when rises in surface temperature/dewpoint are very difficult in comparison to strong advection regimes typically seen in the cool season QLCS scenario. The fact this advection occurred in late June is  very significant.   From a temperature/dewpoint of 71F/69F at sounding release (balloon actually released at 0430Z) to 76/73 at 07Z when a tornado was on the ground very near the RWIS and sounding location, this well-timed and well-located data provided verification to the simulations observed in Jessica’s work.  This may seem incidental at first look – only a 5F (temperature) and 4F (dewpoint) rise in 3.5 hours – but for nocturnal QLCSs in mid-summer this equates to a rapid change in near-storm environment and threat.



Wilmington RWIS Time Series (Courtesy Timothy Boyer, ODOT)


Using the same SBCAPE/SBCINH computing methods as the 05Z sounding linked via SPC above, augmenting that sounding used the observed surface temperature and dewpoint (subjectively maintaining a similar mixing ratio and temperature lapse rate immediately above the surface), the environment felt by tornadic QLCS near Wilmington featured an SBCAPE rise from 456 (with SBCINH of -269 suggesting elevated storms) to > 2000 J/kg with SBCINH reduction to -25 (suggesting surface based storms).  That’s a huge change in threat/environment over the period of 3 hours — at night.

05Z KILN Special Sounding Augmented With 07Z RWIS Surface Conditions

Granted this event was far from a traditional HSLC event which was the crux of Jessica’s study, but as we’ve seen time and time again with nocturnal QLCS events in the Ohio Valley (and elsewhere) amidst strong low level shear, if your boundary layer theta-e advection is aggressive/strong – warning meteorologists should anticipate resultant increase in impacts via damaging wind and/or mesovortex tornadoes provided ambient low level shear/cold pool balance is optimal.


Below is the evolution of the tornadic circulation via KILN WSR-88D (top images) and FAA Terminal Doppler Weather Radar (TDWR) at Dayton, OH (bottom images) which was surveyed to be a 20+ mile EF1.  There was even very subtle TDS (reduction in correlation coefficient co-located with the SRM velocity couplet) on KILN from Waynesville to just west of WFO ILN before noise/clutter near the radar masks the signal.  For the WSR-88D imagery, upper left quadrant is 0.5 degree reflectivity (with MESO-SAILS 3 invoked), upper right is 0.5 degree SRM.  For the TDWR imagery, the bottom left is 0.3 degree reflectivity, and the bottom right is 0.3 degree SRM.

KILN (top) and Dayton TDWR (bottom) reflectivity and velocity loop

From a wind shear perspective, this event possessed the ingredients seen in other high-impact QLCS nocturnal wind events in the Ohio Valley (and elsewhere), with strong 0-1km effective SRH (owing to a strong west-southwesterly low level jet).

07Z SPC 0-1km Effective SRH (Courtesy SPC and Jonathan Blaes)

Also of interest, is the SPC Mesoanalysis clearly indicating a large area positive nocturnal surface theta-e advection with relative maximum in advection centered very near WFO ILN in southwest Ohio.



07Z SPC Surface Theta E (Green Contours) and Advection (Purple Contours) — Courtesy SPC and Jonathan Blaes

Posted in Uncategorized | 5 Comments

Webinar on the Environmental Conditioning of Cool Season, Low Instability Thunderstorm Environments in the Tennessee and Ohio Valleys and Southeastern U.S.

In late April, Jessica King, presented a summary of much of her CSTAR research simulating severe and non-severe high shear, low CAPE convective events. The presentation focused on the examination of the rapid destabilization that occurs in the few hours leading up to severe convection in the simulated events and the mechanisms responsible for the rapid changes in CAPE. Links to the PowerPoint slides and the recorded webinar are available at the bottom of the post. Some notes from the presentation are shared below.

fig1Low instability severe thunderstorms are heavily concentrated in the Southeast especially in the Ohio, Tennessee, and Mississippi Valleys. These events tend to occur in the cool season, often in the overnight and early morning hours, and frequently with a convective mode of mini-supercells or QLCSs. The occurrence of severe convection in low instability environments can often be explained by the presence of synoptic scale forcing and mid-latitude cyclones. The research goal was to determine mechanisms by which environmental conditioning occurs in severe and non-severe high shear, low CAPE thunderstorm events.

Simulated Environments
fig1.5Jessica conducted real-data simulations of high shear, low CAPE events when there was at least a “slight” risk for severe storms and the SfcOA (SPC mesoanalysis) CAPE ≤ 1000 J kg-1 and 0-3 km shear ≥ 18 m s-1. She developed a 3-hour time series analysis for separate points in a 7×7 grid for the three hours prior to severe convection. The 0-1km wind shear showed some discrimination between severe and non-severe events with values higher in severe events, especially at night. The wind shear tended to remain relatively steady over time while the CAPE increased over time.

Calculating Contributions to CAPE
fig3Advection of high theta-e air often leads to an increase in CAPE. A plot of 3-hour change in SBCAPE up to the event time shows a much larger increase in SBCAPE prior to the event for severe events with a smaller increase for non-events. An increase in CAPE can be realized by 1) increasing surface temperature 2) increasing surface moisture or 3) decreasing temperature aloft. The mechanisms for destabilization varied significantly among all environments with significant destabilization occurring in the 3 hours prior to severe events. The change in surface moisture was a positive contribution to increase in CAPE for all cases. Warming near the surface was important in destabilizing all of the severe events as well, and cooling near the surface may be detrimental in the non-severe events. The increase in surface temperature was noted for both daytime and nocturnal cases.

Synoptic Forcing for Ascent
fig4Forcing for ascent can be driven by processes such as warm advection, lifting by a front or boundary, and cyclonic vorticity advection aloft. These processes can lead to the release of potential instability. The release of potential instability through layer lifting occurred prominently in 4 of the simulated severe events. In these 4 cases, the 3km vertical velocity increases significantly and the 0-3km lapse rate became less negative in the hour before the severe event as the synoptic forcing approaches.

The SHERBS1 and to a lesser extent the SHERBS3 were discriminators between the severe and non-severe events. The skill of the SHERBS1 was likely a result of the 0-1km shear serving as a good differentiator. Work is underway to modify the SHERB parameters to perhaps include a term to account for the release of potential instability.

Future Work and Looking Ahead
• Investigate larger scale observations and climatologies to identify recurring synoptic-to-mesoscale patterns in cool season, low instability events in the Southeast.
• Examine the vertical distribution of CAPE.
• The recognition of patterns in high-resolution model data and observations that indicate a low CAPE environment has the potential to evolve into a severe-weather producing environment.
• The availability of datasets and tools to examine data in the highest temporal resolution possible is critical. In many events, some of the higher resolution data doesn’t show a supportive environment 3 hours in advance of the event but a rapid change in low-level moisture or temperature can occur just ahead of the line that provides the needed buoyancy. Sub-hourly time scales or other perspective may provide an opportunity to anticipate these events.
• There may be an R2O pathway with the near-storm situational awareness environment tool that is under development which provides a means for detailed monitoring of the environment.
• There was also an interest in requesting the SPC produce some 1 or 2 hour change fields for surface temperature and moisture.

PowerPoint slide deck
Video recording of webinar


Posted in Convection, CSTAR, High Shear Low Cape Severe Wx, Uncategorized | 1 Comment