The Windstorms of 1995-2006:
Adjusting Peak 5-Second Wind
to Better Fit With Past Instant Gust Measures

compiled by

Wolf Read

Introduction

By the end of 1996, Automated Surface Observing System (ASOS) stations had been commissioned at all of the official National Weather Service (NWS) stations in the western Pacific Northwest. With the implementation of the automated equipment, arrived a new methodology for measuring wind. Before ASOS, a gust was an instantaneous value taken from an analog anemometer with an indicator dial and/or strip-chart recorder. After ASOS, a 5-second average from a digital anemometer became the new gust. Given a high variability in wind speed over small spans of time, a 5-second average, which flattens the short-period changes in wind speed, is likely to be considerably lower than an instant reading.

For a rigorous examination of windstorm climatology, this reality is not trivial. The 5-second gust is a completely different measure than the peak instant gust that was formerly used by the NWS. This issue has been covered in a number of webpages on this site, including the case studies for the December 12, 1995 and January 16, 2000 windstorms. This essay will examine the climatological problems that result from a 5-second wind in a more systematic manner. This is important, for the 5-sec gust measure nearly invalidates comparison between storms of history with those of today. For example, it is possible that the new 5-second peak "masked" some of the storms, due to a tendency for significantly lower wind-speed numbers compared to pre-ASOS events. In other words, a major storm have struck and not produced the expected peak gust signature owing to the difference in measure.

An even newer complication with gust measurement has occurred in the 2003-2007 timeframe. The NWS switched from cup-based anemometers to sonic versions in this period, with many being commissioned in the western Pacific Northwest during 2007. With sonic anemometers, gust measure is changed again. This time to a 3-second average, which is the international standard, and used in engineering circles to help develop building codes. A 3-second gust is somewhat different from a 5-second gust. Also, sonic anemometers are purely digital systems that have no moving parts; they are a dramatically different instrument for wind measure compared to the cup-based analog systems employed before ASOS, and a substantially different instrument compared to the cup-based digital systems employed during the first decade of ASOS.

The cup-based anemometers of yesteryear, relative to sonic systems, have inertial-response issues, both in terms of the cups spinning-up during a gust, and in the movement of the indicator needle or pen or strip-chart depending on what equipment was in use. Depending on the source, it is thought that an instant gust on the old anemometer systems is akin to a 1-sec to 3-sec average.

Nevertheless, the 3-sec gust on sonic anemometers is anticipated by the NWS to be more akin to the old instant gust readings. However, early evidence (based on observations during the 2007-08 storm season) points to the 3-sec gust on sonic anemometers perhaps being slightly high relative to the old instant gust. Gust ratios on modern sonic systems appear to be higher compared to the systems of old, especially during strong storms. These new sonic anemometers are clearly a new complication to comparative analysis of storm systems over time, one that will be explored in-depth hopefully soon.

To help clarify the issues, take some wind records for Astoria, OR, for example. On October 12, 1962, Astoria reported a peak instant gust of 96 mph on an analog cup-based anemometer; on March 3, 1999, Astoria reported a peak 5-second gust of 66 mph on a digital cup-based anemometer; and, on December 3, 2007, Astoria reported a peak 3-second gust of 94 mph on a digital sonic anemometer. Given the change in equipment, and gust measure, are these numbers in any way directly comparable?

The following discussion primarily focuses on what can be termed the "Five-Second Gust Era," which, for the majority of western Pacific Northwest stations occurred from 1996-2006, with some stations like Astoria being changed to ASOS as early as 1993. Examination will primarily be on the 5-second gust. The 3-second gust is also explored, but only peripherally. Note that the data presented on the 3-second gust was collected on a cup-based system, and the results are probably not applicable to sonic anemometer systems.

Discussion

A 5-second "gust" is something entirely different from an instant gust. The former is an average, so technically it should be called a 5-second "wind" (and usually is by the NWS and NCDC). An instant gust is a direct reading, though arguably on cup-based systems inertial response can be considered to be an averaging mechanism. A specific value for a 5-second wind can contain much higher instant readings within its duration. Figures 1 through 3, below, demonstrate the high-frequency variance in speed that can be present in wind.

Figure 1: Peak gust, Oregon City, OR, during the passage of the February 7, 2002 cyclone. Maximum speed for each averaging period indicated on the chart. Peak instant gust 27 mph, peak 3-second wind 23 mph and peak 5-second wind 21 mph. Same wind gust, different measure.

Figure 2: Peak gust, Albany, OR, for the December 5, 2003 passage of a strong open wave. Maximum speed for each averaging period indicated on the chart. Peak instant gust 38 mph, peak 3-second wind 33 mph and peak 5-second wind 30 mph.

Figure 3: Peak gust, Albany, OR, for the January 29, 2004 passage of a strong cyclone. Maximum speed for each averaging period indicated on the chart. Peak instant gust 44 mph, peak 3-second wind 39 mph and peak 5-second wind 37 mph.

The values and actual ratios are listed below in Table 1. Official readings taken during the December 12, 1995 windstorm are included. It is fortunate that the old direct-reading anemometer was still in operation then, and that the peak gust was noted by NWS personnel.

Table 1: Summary of known peak gust to 5-second wind measures. The NWS readings were from two different systems, the old direct-reading equipment and the ASOS sensor, a fact that confounds the estimation.

Table 2: Summary of known peak gust to 3-second wind measures.

The information in Table 1 can be used to make an instant-peak-gust estimation for windstorms that occurred during the five-second gust era. There is some evidence of a trend in the above data where the ratio decreases as the peak gust increases, suggesting that the higher the gust, the longer its duration. If this is true, using the straight average of the ratios would result in overestimation of peak gusts at the high end. For the following calculations, the 1.19 ratio from the official data will be used for the peak gust adjustments. At some point a sliding scale might be employed for finer-tuned estimations--this probably won't happen until more data is added to the above list, especially for faster than 27, 38 and 44 mph peak gust readings.

Outside the peak gust, a number of significant gusts are indicated in some of the charts in Figures 1 through 3. The January 29, 2004 cyclone produced a number of surges that were on par, or faster than the other storms. These data provide an opportunity to further scrutinize the 5-sec-gust / instant-gust ratio.

Looking specifically at the chart for the January 29, 2004 storm event, there are many waves in the data, some larger than others. Since the focus of the research presented on this website is peak gust, and there may be something unique about a peak gust relative to the other, slower gusts in a windstorm, logically an analysis of a continuous wind trace should concentrate on surges in wind speed that resemble the peak gust. Using the available data, a standard, or rule, should be developed for isolating legitimate gusts that also adequately describes the peak gust. Looking at the 5-second wind during the peak gust, the ratio of peak, 37 mph, to the preceding trough, 23 mph is about 1.61. Interestingly, this happens to be very close to the Golden Ratio of 1.618. Using a rounded version of the ratio as a guide, any change in wind speed with a peak-speed / preceding-trough-speed ratio of about 1.6 or greater counts as a legitimate gust.

When this methodology is employed, six gusts are identified in the January 29, 2004 record. These are shown in Figure 4, below. The ratios for all these gusts are above 1.6 save in the case of gust-3. For gust-3, the brief, and very shallow trough at about 80 seconds results in a ratio of about 1.3. However, given that there is a more significant trough at about 65 seconds, ahead of the brief stall in wind speed acceleration, gust-3 is considered valid as it results in a ratio of 1.7. Gust-5, which has a very shallow trough afterward on the way up to the peak gust, is perhaps more questionable than gust-3. Figure 4 is provided, in part, to let the reader decide on the veracity of these identified gusts.

Figure 4: In the high-resolution record of the January 29, 2004 peak wind period at Albany, OR, six gusts are identified. The basic gust criteria is to have a trough-to-peak ratio of approximately 1.6 (peak/trough), the value for the peak gust during this storm. The vertical blue lines indicating the various gusts are centered on the peak 5-second wind speed.

The data for the six gusts are provided in Table 3, below. The average ratio of instant-gust to 5-second gust is around 1.19, with a range of 1.15 to 1.29. In other words, instant gust--as determined by taking 0.5-second samples from the indicator dial of an analog cup-based anemometer during the highest wind phase of a single storm event--tends to be approximately 20% higher than 5-second gust. The average ratio of instant-gust to 3-second gust is around 1.12, with a range of 1.08 to 1.17. Instant gust tends to be approximately 10% higher than 3-second gust. These numbers probably delimit good rules-of-thumb: The conversion for 5-sec to instant is 1.2, and 3-sec to instant is 1.1.

The values given in Table 3 support the ratios from the other wind events that are listed in Tables 1 and 2. These data lend some weight to the idea of a ~1.2 conversion factor for 5-second gust to instant gust.

To be slightly more conservative than 1.2, the 1.19 adjustment factor supported by NWS equipment during the December 12, 1995 windstorm is used on some wind events from 1995-2006. The results appear in Table 4, below.

Storm watchers, pay close attention to the modified readings for the Columbus Day Storm. That's the 5-sec gust signature of a "storm-of-the-century" event. The extreme instant gust values are reduced considerably, with the storm's average falling close to the unmodified value for the November 14, 1981 windstorm. With peak gusts depicted in estimated 5-second averages, the Columbus Day Storm stands out somewhat less. Gusts in the 60s and 70s don't seem as spectacular as 80s, 90s, and 100s. Nevertheless, even with the reduction, some stations still have a strong showing, including Portland, OR, with an 87 mph 5-second wind. Of course, the 104 mph was an estimate--a necessity as the storm knocked out power early and shut down the direct-reading wind equipment. Using the measured value of 106 mph at Troutdale, we get 89 mph, which does stand out, and is a close match to an 88 mph "fastest mile" (88 mph 41-second wind) at Portland that was recorded near the beginning of the storm. In any event, a Columbus-Day type of event would show up numerically less strongly in a 5-second wind regime, but, owing to some stations being subjected to particularly strong winds (and the physical reality of the kind of damage such a storm would cause), will still stand out as something unusual.

For an extreme, but not Columbus-Day-Storm-extreme, event like November 14, 1981, the story is different. Save for at North Bend, the peak wind values do not stand out much. Many storms in history have produced 50 to 60 mph instant gusts. The average of 55.4 is still in major category, but is right at the borderline. This clearly demonstrates how peak 5-second gusts can "mask" a rare windstorm. This table suggests that the 5-second gust signature for a major event is 50 to 60 mph gusts, not 60 to 75.

There's another twist to the story being examined here: What happens to the frequency of high wind warnings between the pre and post 5-second-gust eras? Did more than one standard for a high wind event, by NWS definition, change?

High Wind Warnings in the Five-Second Gust Era

National Weather Service criteria for a high wind warning (HWW) event is for gusts of 50 knots, or 58 mph, to occur within the warning area. Gusts meeting this criterion are shown in red in Tables 2 and 3. This definition has been around some time; minimum HWW criteria during the pre-ASOS era certainly wasn't higher. I recall a few high wind warnings for 55 mph gusts in the Seattle area during the 1980s.

If the NWS bases its wind forecasts on how the official stations respond, then, quite possibly the frequency of HWW might have decreased since the inception of ASOS. Note that three storms in Table 4 did not produce HWW criteria "gusts" under the official peak 5-second wind record of ASOS: February 5, 1999, February 6, 1999 and December 3, 2003. Yet, based on the 1.19 adjustment factor, it appears that these three events could have achieved gusts of 58 mph or higher at coastal stations. For the last storm, this point is kind of moot, for a HWW was issued for the coast (one that was verified by stations not included in the table).

High wind warnings for the interior sections, like the Willamette Valley, are rare. The difference between peak 5-second wind and peak instant gust is probably more significant than on the coast, where gusts to 60 and 70 mph are regular occurrences during the winter season. The January 1, 1997, March 3, 1999 and December 27, 2002 windstorms, which generally didn't have a strong showing in the interior for 5-second wind measures (save SEA and BLI in 1999), meet HWW criteria with the peak instant gust adjustment, either at one point, or over a wider portion of the inland region.

Another way of looking at this is with Table 6, especially the November 14, 1981 windstorm. The number of stations meeting HWW criteria drops from 10 to 3 when the peak instant gusts are adjusted to 5-second wind! The storm appears to barely meet the high wind warning standard, yet it was one of the biggest in history.

It seems that the standard for high wind was automatically changed by employing a new peak 5-second "gust" at official stations. Instant gusts of 69 mph are a reasonable expectation when 5-second winds are approaching 58. So, the new ASOS era appears to have changed the HWW criteria to approximately a minimum of 70 mph for instant gust. The other way to look at this is that peak 5-second winds of 48 to 50 mph (implying potential instant gusts of 57 to 60 mph) at ASOS stations would have likely met the minimum HWW criteria pre-ASOS. It seems that 5-second "gusts" of 43 knots, approximately 50 mph, might have been a reasonable new criteria for high wind warnings, certainly for interior sections of the Pacific Northwest (there's good reason to distinguish between coastal and interior regions when providing wind advisories and warnings, a topic in its own right).

Since it's doubtful that 58 mph instant gusts have changed in force during the implementation of the 5-second gust, it's interesting that the NWS has kept the 58 mph minimum for HWW when using a 5-second wind to "verify" warnings. Skywarn spotter reports probably reduce this issue to some extent. A number of personal anemometers still report instant gust in analog format (Maximum equipment being one example), though digital stations that show short-duration averages, like 2.5-seconds on Davis systems, are also common, watering down the effect.

The bottom line: The 5-second gust era of 1996-2006 was different from the periods both before and after. A 5-second average on a cup-based anemometer system is a unique kind of wind measure. This should be kept in mind when trying assess the strength of a particular wind event relative to other storms in history.

Footnotes

[Footnote 1] Certain stations, like Cape Arago and Cape Blanco, of course, have recorded significantly faster 5-second winds, but these are among the most wind-favored locations in the Northwest and aren't included in Table 2.

Last Modified: November 29, 2008
Page Created: December 9, 2003

| Back |