Conditional Warning Probabilities
Table of Contents
i. Background
ii. Methodology
iii. Results
1. Warning probability as a function of watch existence
2. Warning probability as a function of SPC categorical risk
3. Warning probability as a function of SPC hazard specific probabilities
4. Warning probability as a function of SPC significant severe areas
5. Warning probability as a function of forecasted watch probabilities
6. Flash flood warning probability as a function of WPC categorical risk
7. Warning probabilities for other hazards as a function of watch existence
iv. Conclusions
v. Appendix A: Tornado warning probability as a function of severe thunderstorm warning existence
vi. Appendix B: Warning probability as a function of mesoscale discussion existence
vii. Appendix C: Warning probability as a function of tropical cyclone warnings
viii. Appendix D: Fire warning probability as a function of SPC fire weather risk
Background
Within the realm of National Weather Service (NWS) products, there are three main categories of products (ordered from longest lead time to shortest lead time): outlooks, watches, and warnings. Generally speaking, "outlooks" provide a broad overview of what areas are at greatest risk for hazardous weather at lead times of 12+ hours. Depending on the hazard, watches can be issued 1-3 days in advance or a few hours in advance, and these alerts are intended to highlight more specific areas where conditions appear favorable for hazardous weather. In the case of severe thunderstorm watches and tornado watches (the so-called "convective watches"), the lead times and durations are usually 3-9 hours. Ultimately, the alerts that get the most attention are the "warnings", which indicate that life-threatening and/or destructive conditions are already occurring or about to occur.
While it is generally accepted that areas under a watch are more likely to experience hazardous weather than areas not under a watch, very little (if any) research has been conducted on how frequently a watch becomes a warning for any given location within the watch. The same could also be said about NWS outlooks; while locations in higher risk areas are supposedly more likely to experience hazardous weather than lower risk areas, the exact probabilities of witnessing a warning have not been established.
Since tornadoes, severe thunderstorms, and flash floods typically cause most weather-related casualties in the United States, the main focus of the proceeding analysis will be for these three hazards and their attendant products.
Methodology
For convective (tornado and severe thunderstorm) watches, the probability of a warning was determined by (1) rendering the watch polygon to a discrete 0.1° latitude × 0.1° longitude grid and (2) calculating the fraction of grid points that featured at least one warning being issued between the watch's issuance time and the watch's expiration time. The polygon (a parallelogram) was chosen because a forecaster has more control over the shape of the polygon than the counties that are included (counties are fixed areas and have a wide variety of sizes and shapes). Though, it is acknowledged that the polygon can include portions of counties that are not actually included in the watch and vice versa.
If a warning was issued without a watch being active, the warning is classified as an "unwatched warning". To estimate the probability of witnessing an "unwatched warning", the total number of days with unwatched warnings within each calendar year was then calculated and divided by the total number of days in a year (365 days for a normal year, 366 days for a leap year).
Estimating warning probabilities for outlook areas involved the same fraction of area calculation used for watches. Unlike watches, outlook areas are more clearly defined, so the estimated probabilities will be more accurate. Also like the watches, the outlook areas were rendered to a discrete 0.1° latitude × 0.1° longitude grid when calculating the fractional coverage of warnings.
Part I: Warning probability as a function of watch existence
The first result (presented in the below table) shows the estimated probability that a location will experience a warning based on what kind of watch (if any) that location is under.
| Warning Type | No Watch | Severe Thunderstorm Watch | Tornado Watch | Particularly Dangerous Situation (PDS) Severe Thunderstorm Watch | Particularly Dangerous Situation (PDS) Tornado Watch |
|---|---|---|---|---|---|
| Tornado Warning | 0.038% | 1.0% | 7.0% | 3.9% | 20.5% |
| Severe Thunderstorm Warning | 0.43% | 28.4% | 26.9% | 55.7% | 38.8% |
| Flash Flood Warning | 0.57% | 2.5% | 6.4% | 6.4% | 9.6% |
To read this table, find the watch type that your location is currently under. Within that column, you'll find the probability that your location will see a tornado warning before the watch expires (first row below the header), the probability that your location will see a severe thunderstorm warning before the watch expires (second row), and the probability that your location will see a flash flood warning before the watch expires (third row). As an example, if your location is under a normal tornado watch, there is a 7.0% chance that your location will see a tornado warning before the watch expires.
For those that aren't familiar, the Storm Prediction Center (SPC) can designate a watch as a "particularly dangerous situation (PDS)" watch if there is high confidence in a major severe weather event occurring within the watch area. For tornado watches, this means a relatively high likelihood of destructive (EF3+) tornadoes. For severe thunderstorm watches, this means a relatively high likelihood of destructive (80+ mph) wind gusts.
It should be noted that the column for "no watch" is unique in that it is the probability for a 24-hour timeframe and not the timeframe of a watch. So, if no watch is in effect, the probability that your location will see a tornado warning on any given day is about 0.038% (about 1 instance every 8 years), the probability that your location will see a severe thunderstorm warning on any given day is about 0.43% (about 1-2 instances per year), and the probability that your location will see a flash flood warning on any given day is about 0.57% (about 2 instances per year). It should be noted that this is a national average, and these particular probabilities are also influenced by what region of the country someone lives in. For example, a lot of severe thunderstorm warnings are issued in the southeast United States during the summer, but the SPC rarely issues watches for these storms because they're usually short-lived and disorganized.
One important conclusion is that the existence of a watch increases the probability that you'll see a warning by a factor of about 100-200. That's a huge increase, and it suggests that the issuance of a watch is objectively a big deal. This effect is not as pronounced for flash flood warnings, but tornado watches and severe thunderstorm watches are not intended to specifically capture flash flooding. Even so, the existence of a tornado watch or severe thunderstorm watch significantly increases the probability of experiencing a flash flood warning by a factor of about 5-15.
There are a couple of other interesting details in this finding that are worth discussing. One is that you're less likely to see a severe thunderstorm warning in a tornado watch vs a severe thunderstorm watch. This is likely due to the fact that tornado watches are usually issued when there is potential for isolated supercells to form within the watch area. Since these storms are isolated (low coverage), they tend to be more hit-or-miss. This is in contrast to a large line of thunderstorms (such as a quasi-linear convective system, QLCS), which impacts a much larger area and is more likely to prompt a severe thunderstorm warning at any given location.
Another important detail is the probability of a tornado warning within a normal severe thunderstorm watch. Usually, a severe thunderstorm watch is issued if the threats for straight-line winds and hail are expected to be more impactful than the threat of tornadoes, but the potential for tornadoes is not always 0. This is clearly reflected in the 1.0% probability of seeing a tornado warning within a severe thunderstorm watch.
Additionally, the issuance of a PDS watch is associated with a significantly higher probability of experiencing a warning before the watch expires. In fact, for any given location within a PDS tornado watch, there is a roughly 1 in 5 chance that a tornado warning will be issued for that location, which is approximately 3 times higher than the same outcome within a normal tornado watch.
Finally, flash flood warnings are more likely to be issued within a tornado watch than a severe thunderstorm watch (by about a factor of 2-3). This is likely due to the fact that many setups that favor tornadoes will involve multiple rounds of heavily precipitating storms (either multiple intense supercells impacting the same locations or a wave of supercells followed by a severe line of storms). Given that flash flood warning coverage tends to increase within PDS tornado watches, it is logical to suspect that high-end tornado events will also be more favorable for flash flooding.
Part II: Warning probability as a function of SPC categorical risk
The SPC can issue 7 different levels of risk on any given day: no thunderstorms, general thunderstorms (meaning thunderstorms are expected to occur, but they won't be "severe"), marginal risk, slight risk, enhanced risk, moderate risk, and high risk. Generally speaking, the higher level risks should correspond to higher probabilities of experiencing a warning. Sure enough, this expectation is confirmed in the below table:
| Warning Type | General Thunderstorms | Marginal Risk | Slight Risk | Enhanced Risk | Moderate Risk | High Risk |
|---|---|---|---|---|---|---|
| Tornado Warning | 0.0072% | 0.18% | 1.2% | 5.5% | 17.6% | 37.4% |
| Severe Thunderstorm Warning | 0.27% | 4.4% | 18.7% | 43.6% | 63.4% | 54.7% |
| Flash Flood Warning | 0.17% | 1.1% | 2.5% | 5.6% | 9.3% | 27.0% |
One pattern that is evident is the tendency to have higher warning probabilities when the SPC issues a higher risk. The one exception to this pattern is "severe thunderstorm warnings" within "high risks". One possible explanation for this artifact is likely due to manpower constraints within NWS offices on big risk days. Forecasters only have so much time to analyze a storm and issue a warning for it, and, in the case of a full-fledged tornado outbreak, it is not realistically possible to issue a severe thunderstorm warning on every single severe storm. Accordingly, some NWS offices will stop issuing severe thunderstorm warnings during a tornado outbreak, because the tornado threat is the highest priority. Since major tornado outbreaks are more likely on high risk days, this would explain why the probability of a severe thunderstorm warning actually decreases on high risk days.
Otherwise, this table is read in the same way as the table for watches. If your location is in a moderate risk, there is a roughly 1 in 6 chance that you will have to deal with a tornado warning at some point in the day. Or, put another way, about one-sixth of the moderate risk area will see tornado warnings. Such information is useful to emergency managers as it communicates approximately how much of their area of responsibility will see warnings. This can also be useful to the general public as it gives people a rough idea of what the probability of a warning is long before the severe weather event begins.
A concerning pattern is the tendency for the probability of flash flood warnings to significantly spike on high risk days. Here's why this is concerning: if a tornado warning is issued for your location, you should take cover in a small interior room on the lowest floor of a well-built structure. If a flash flood warning is issued for your location, you should move to higher ground. What do you do if you're under both a tornado warning and a flash flood warning at the same time? The recommended protective actions contradict each other, which makes for a major public messaging conundrum.
Part III: Warning probability as a function of SPC hazard specific probabilities
The SPC's categorical risk is largely determined by 25 mile (40 km) neighborhood probabilities for each of the three severe weather hazards (tornadoes, 58+ mph straight-line winds, and large hail). Therefore, it would be reasonable to expect a pattern between the SPC's tornado probability forecast and the probability of being included in a tornado warning. A similar pattern should also exist between the SPC's wind and hail probabilities and the probability of being included in a severe thunderstorm warning. Sure enough, this supposition is confirmed by the following three tables:
| Warning Type | 2% Tornado Risk | 5% Tornado Risk | 10% Tornado Risk | 15% Tornado Risk | 30% Tornado Risk | 45% Tornado Risk | 60% Tornado Risk |
|---|---|---|---|---|---|---|---|
| Tornado Warning | 0.71% | 3.6% | 12.4% | 23.9% | 36.9% | -- | -- |
| Severe Thunderstorm Warning | 13.0% | 26.4% | 42.7% | 55.2% | 54.9% | -- | -- |
| Flash Flood Warning | 2.2% | 4.7% | 9.9% | 12.7% | 27.1% | -- | -- |
| Warning Type | 5% Wind Risk | 15% Wind Risk | 30% Wind Risk | 45% Wind Risk | 60% Wind Risk |
|---|---|---|---|---|---|
| Tornado Warning | 0.28% | 1.2% | 4.9% | 18.2% | -- |
| Severe Thunderstorm Warning | 4.9% | 20.4% | 47.2% | 71.2% | -- |
| Flash Flood Warning | 1.1% | 2.1% | 5.8% | 9.9% | -- |
| Warning Type | 5% Hail Risk | 15% Hail Risk | 30% Hail Risk | 45% Hail Risk | 60% Hail Risk |
|---|---|---|---|---|---|
| Tornado Warning | 0.41% | 1.4% | 6.0% | 17.0% | -- |
| Severe Thunderstorm Warning | 7.4% | 22.6% | 45.9% | 68.4% | -- |
| Flash Flood Warning | 0.92% | 2.0% | 6.0% | 9.9% | -- |
The reason why some cells are blanked out in the above tables is due to small sample size. Tornado probabilities at or above 45% are extremely rare (issued about once every 5 years). This statement also applies to 60% wind probabilities (issued about once every 5 years) and 60% hail probabilities (issued about once every 10 years). Still, there is enough of a pattern present in the more common probabilities to extrapolate what the warning probabilities should be for these rare issuances.
Generally speaking, the tornado probability issued by the SPC is roughly equivalent to the probability of experiencing a tornado warning. The main exception to this rule is the 15% tornado risk. This is likely due to the large difference between 15% and 30% tornado probabilities. Since tornadoes are the rarest of the three severe weather phenomena, a 15% change in probability is more significant than having the same probability change for more common hazards like wind and hail. There are likely several days where the actual tornado probability is 20% - 25% and therefore not quite high enough to justify a 30% tornado probability area (which itself would prompt the rare and attention-getting "high risk").
Once again, we see the same concerning pattern noted above (flash flood warnings becoming significantly more likely within the highest tornado risks). Based on this finding, NWS forecasters probably should be especially conscious about what warnings they issue during major tornado outbreaks. For instance, should a tornado warning be issued on a weak looking velocity couplet in an area extremely susceptible to flooding? Probably not, because that might encourage people to seek shelter underground where they might drown. Should a flash flood warning be issued for low-end flooding while an EF4+ tornado is approaching a major city? Probably not, because that might discourage people from seeking out an underground shelter. This is, unfortunately, a conundrum with no simple solution. Ultimately, the forecasters issuing warnings may have to decide which threat is more likely to kill people and act accordingly.
Part IV: Warning probability as a function of SPC significant severe areas
Within the tornado, wind, and hail outlooks issued by the SPC, it is possible for SPC forecasters to issue a "hatched area". On the tornado outlook, this means there is a 10% or greater probability of an EF2+ tornado within 25 miles of any given location. On the wind outlook, this means there is a 10% or greater probability of a 65+ kt (74+ mph) wind gust within 25 miles of any given location. On the hail outlook, this means there is a 10% or greater probability of a 2.00"+ diameter hail event within 25 miles of any given location.
Since the hatched area is itself a 10% probability for a "significant event", it is unreasonable to issue a hatched area if the probability of the hazard itself is less than 10%. That is, it makes little sense to say, "there's a 10% chance you'll see 2.00"+ hail today, and there's a 5% chance you'll see 1.00"+ hail today". Accordingly, hatched areas are only issued if the hazard's normal probability is itself at least 10%.
If a hatched area is present, that implies the overall potential for severe weather is higher when contrasted to an outlook with no hatched area. Accordingly, it would be reasonable to expect higher warning probabilities when examining hatched areas.
When examining probabilistic tornado forecasts, there is, as a practical matter, only one potential point of comparison (10% unhatched areas vs 10% hatched areas). It is rare for the SPC to issue a 15% tornado probability without hatching since having lots of tornadoes increases the likelihood of an EF2+ tornado occurring (by the law of averages). There are only two possible scenarios when this might occur: a tropical cyclone prolifically producing weak tornadoes, and a line of thunderstorms producing numerous brief spin-ups along the leading edge. The SPC has never issued a 30% unhatched tornado probability, and the only way this could theoretically be issued is via one of the two aforementioned scenarios hosting a particularly extreme case.
Wind and hail forecasts have more practical points of comparison. The SPC has historically issued hatched and unhatched risks for all possible hail/wind probabilities of at least 15%, although the rarest combination is a 60% unhatched hail risk (issued only once on the 0100Z outlook for February 28, 2007) followed by a 45% unhatched hail risk (which has only ever been issued 6 times and hasn't been issued since 2009). Since 60% probabilities are rare regardless of hatching, it is difficult to draw any definitive conclusions regarding the effect of hatching on warning probabilities when a 60% probability is in effect.
With all that in mind, here are three tables showing the different warning probabilities based on the existence (SIG) or absence (non-SIG) of a hatched area:
| Warning Type | 10% (non-SIG) Tornado Risk | 10% (SIG) Tornado Risk | 15% (non-SIG) Tornado Risk | 15% (SIG) Tornado Risk | 30% (non-SIG) Tornado Risk | 30% (SIG) Tornado Risk | 45% (non-SIG) Tornado Risk | 45% (SIG) Tornado Risk | 60% (non-SIG) Tornado Risk | 60% (SIG) Tornado Risk |
|---|---|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 9.6% | 11.7% | -- | 23.9% | -- | 36.9% | -- | -- | -- | -- |
| Severe Thunderstorm Warning | 40.4% | 44.3% | -- | 55.2% | -- | 54.9% | -- | -- | -- | -- |
| Flash Flood Warning | 8.7% | 10.2% | -- | 12.7% | -- | 27.1% | -- | -- | -- | -- |
| Warning Type | 15% (non-SIG) Wind Risk | 15% (SIG) Wind Risk | 30% (non-SIG) Wind Risk | 30% (SIG) Wind Risk | 45% (non-SIG) Wind Risk | 45% (SIG) Wind Risk | 60% (non-SIG) Wind Risk | 60% (SIG) Wind Risk |
|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 1.1% | 2.4% | 4.6% | 5.5% | 17.4% | 18.7% | -- | -- |
| Severe Thunderstorm Warning | 19.8% | 39.9% | 43.2% | 52.3% | 56.3% | 71.4% | -- | -- |
| Flash Flood Warning | 2.1% | 2.3% | 5.8% | 5.5% | 10.7% | 10.4% | -- | -- |
| Warning Type | 15% (non-SIG) Hail Risk | 15% (SIG) Hail Risk | 30% (non-SIG) Hail Risk | 30% (SIG) Hail Risk | 45% (non-SIG) Hail Risk | 45% (SIG) Hail Risk | 60% (non-SIG) Hail Risk | 60% (SIG) Hail Risk |
|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 1.3% | 1.9% | 5.2% | 6.2% | -- | 17.0% | -- | -- |
| Severe Thunderstorm Warning | 21.2% | 29.3% | 38.3% | 46.7% | -- | 68.4% | -- | -- |
| Flash Flood Warning | 2.0% | 2.7% | 4.9% | 5.7% | -- | 9.9% | -- | -- |
The table with SPC tornado probabilities show the expected pattern; tornado warnings are more likely to be issued in locations where the SPC has highlighted a significant tornado risk. The same could be said for severe thunderstorm warning probabilities and flash flood warning probabilities.
The same general pattern shows up on the table with SPC wind probabilities, but there are a few unexpected results that are worth discussing. One is the significant difference in severe thunderstorm warning probabilities between a normal 15% wind risk and a 15% wind risk with hatching. This could partially be explained by the fact that significant wind areas tend to be small, so the fractional coverage could be inflated because of that. Still, when the wind probability is 15%, there is a two-fold increase in severe thunderstorm warning probability if a hatched area is also present for wind.
The other detail worth highlighting on the wind probability table is the slight decrease in flash flood warning probability when comparing 45% unhatched wind risks to 45% hatched wind risks. This is likely due to the fact that higher-end wind events typically correlate with faster-moving storms, which limits the potential for rapid heavy rainfall.
Part V: Warning probability as a function of forecasted watch probabilities
Whenever the Storm Prediction Center issues a tornado watch or a severe thunderstorm watch, the forecaster issuing the watch also includes probabilities for 7 different potential outcomes within the watch area. These potential outcomes include:
- 2 or more tornadoes of any intensity occurring somewhere within the watch area
- 1 or more significant (EF2+) tornadoes occurring somewhere within the watch area
- 10 or more severe (58+ mph) straight-line wind events occurring somewhere within the watch area
- 1 or more significant (75+ mph) straight-line wind events occurring somewhere within the watch area
- 10 or more severe (1.00"+) hail events occurring somewhere within the watch area
- 1 or more significant (2.00"+) hail events occurring somewhere within the watch area
- 6 or more combined severe straight-line wind and severe hail events occurring somewhere within the watch area
The probabilities for each outcome are largely influenced by the categorical risk and the hazard specific probabilistic forecasts issued by the SPC (discussed in prior parts). Based on this information, a natural expectation would be for tornado warnings to be more likely in watches with higher tornado probabilities and for severe thunderstorm warnings to be more likely in watches with higher wind and hail probabilities. This general pattern does show up in the following tables, but some of the trends are not as consistent as with other results.
Starting with the tornado probabilities:
| Forecasted probability of at least 2 tornadoes | 5% | 10% | 20% | 30% | 40% | 50% | 60% | 70% | 80% | 90% | 95% |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 0.3% | 0.6% | 1.5% | 4.0% | 5.8% | 7.0% | 9.4% | 11.5% | 15.7% | 17.0% | 24.5% |
| Severe Thunderstorm Warning | 29.4% | 28.6% | 28.1% | 20.0% | 25.7% | 29.2% | 34.1% | 33.0% | 35.5% | 42.3% | 39.2% |
| Flash Flood Warning | 1.7% | 2.3% | 2.9% | 5.5% | 5.5% | 6.9% | 7.7% | 7.6% | 9.0% | 7.2% | 12.7% |
As expected, there is a fairly steady correlation (r = 0.96) between forecasted tornado probability and tornado warning probability. However, the pattern with severe thunderstorm warnings shows more of a V-shape with a minimum value when the forecasted tornado probability is 30%. The minimum value probably occurs at 30%, because the 30% tornado probability seems to only be used when a forecaster believes conditions are marginally favorable for tornadoes and only a few storms are expected. If only a few storms form, relatively few locations will experience a severe thunderstorm warning.
A forecasted tornado probability of 20% or lower is typically assigned to severe thunderstorm watches. Generally, severe thunderstorm warnings cover a larger percentage of a severe thunderstorm watch than a tornado watch. This is likely why the severe thunderstorm warning probabilities attain relatively high values when the tornado probability is lower.
When the forecasted tornado probability exceeds 60%, the severe thunderstorm warning probability begins to fluctuate. This is likely an operational anomaly that can occur when a major tornado outbreak occurs within a National Weather Service's area of responsibility (the County Warning Area, CWA). During a major tornado outbreak, severe thunderstorm warnings become a lower priority, so forecasters dedicate their time toward issuing or updating tornado warnings instead of worrying about severe thunderstorm warnings.
As far as flash flood warning probability is concerned, we see a generally positive correlation between forecasted tornado potential and flash flood warning probability.
| Forecasted probability of at least 1 significant (EF2+) tornado | 2% | 5% | 10% | 20% | 30% | 40% | 50% | 60% | 70% | 80% | 90% | 95% |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 0.6% | 1.3% | 2.8% | 5.9% | 7.9% | 9.6% | 11.6% | 17.8% | 20.2% | 20.6% | 24.7% | 28.0% |
| Severe Thunderstorm Warning | 28.7% | 27.7% | 27.3% | 25.7% | 30.5% | 35.8% | 36.0% | 41.0% | 43.0% | 38.6% | 46.4% | 34.8% |
| Flash Flood Warning | 2.0% | 2.8% | 4.6% | 5.9% | 6.3% | 8.0% | 6.6% | 10.1% | 10.7% | 7.5% | 9.3% | 25.1% |
Like above, the probability of a tornado warning generally increases as the forecasted significant tornado probability increases (r = 0.99). However, there are a couple of fluctuations that occur when the forecasted probability is 70% and 80%. The steady upward trend is slightly disrupted around these values, suggesting that 80% probabilities are being issued when lower probabilities are warranted.
The severe thunderstorm warning probability exhibits a similar pattern as in the previous table, except that the minimum value (though less pronounced) occurs when the forecasted significant tornado probability is 20%. There are also fluctuations in severe thunderstorm warning coverage for the higher-end forecasted tornado probabilities despite the fact that tornado warning coverage generally increases.
Once again, we see a generally positive correlation between forecasted tornado probabilities and observed flash flood warning probabilities. This correlation is probably due to the fact that high-end tornado events can involve multiple intense storms producing heavy rainfall in same locations. What's especially noteworthy in this particular table is how similar the tornado warning and flash flood warning coverage is when the forecasted significant tornado probability is 95%. This implies a relatively high likelihood of some people having to deal with both a flash flood warning and a tornado warning (possibly simultaneously).
| Forecasted probability of at least 10 severe wind events | 5% | 10% | 20% | 30% | 40% | 50% | 60% | 70% | 80% | 90% | 95% |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 6.8% | 5.5% | 2.6% | 2.7% | 2.0% | 2.5% | 3.5% | 4.6% | 7.0% | 7.3% | 7.2% |
| Severe Thunderstorm Warning | 3.4% | 10.3% | 16.5% | 22.6% | 27.6% | 30.1% | 34.1% | 40.3% | 44.3% | 47.6% | 46.9% |
| Flash Flood Warning | 13.3% | 10.1% | 3.7% | 3.7% | 3.0% | 3.2% | 3.9% | 4.3% | 5.7% | 5.8% | 5.7% |
When examining the table with forecasted severe wind probabilities, an interesting pattern emerges for the tornado warning probability. When the forecasted probability of severe wind is below 20%, the probability of a tornado warning within the watch area is above 5%. The probability of a tornado warning then fluctuates when the forecasted wind probability is 20% or 30% before experiencing a more consistent positive correlation with the forecasted severe wind probability at or above 40%.
When the forecasted severe wind probability is below 20%, the forecaster most likely believes that storms will mainly be isolated. Since an isolated storm mode is more favorable for long-lived supercells and therefore more favorable for tornadoes (and less favorable for widespread straight-line wind damage), it does make sense that the tornado warning probability would be higher when the forecasted wind probability is low.
That being said, at a certain point, a sufficiently intense line of storms can produce both widespread wind damage and numerous embedded tornadoes. The intensity of a line is often implied by the forecasted wind probability (higher wind probabilities reflect a forecaster's belief that a more intense line of thunderstorms will occur). With this in mind, it makes sense why the tornado warning probability does eventually increase once the forecasted wind probability exceeds a certain amount (~40%).
Since more intense lines of storms often produce more widespread wind damage, more severe thunderstorm warnings can be expected when the line of storms is more intense. Therefore, if a forecaster issues a high wind probability (with the expectation of an intense line of storms), it would make sense for more severe thunderstorm warnings to be issued for the line, which itself also impacts a large area. Sure enough, a strong positive correlation (r = 0.99) exists between the forecasted wind probability and the probability of a severe thunderstorm warning. The only discernible inconsistency is between the 90% and 95% probabilities, and this could be due to smaller sample size.
The flash flood warning probability shows a very similar pattern to the tornado warning probability, which further supports the conclusions proposed above about how high-end tornado events seem to correlate with high-end flooding events.
| Forecasted probability of at least 1 significant severe wind event | 5% | 10% | 20% | 30% | 40% | 50% | 60% | 70% | 80% | 90% | 95% |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 3.0% | 2.6% | 2.4% | 3.1% | 4.1% | 5.1% | 6.9% | 10.0% | 13.4% | 9.0% | 12.9% |
| Severe Thunderstorm Warning | 14.3% | 25.1% | 28.5% | 27.1% | 36.4% | 42.2% | 44.6% | 56.1% | 43.1% | 59.4% | 61.9% |
| Flash Flood Warning | 6.3% | 3.9% | 3.1% | 3.7% | 4.2% | 4.0% | 4.2% | 6.7% | 9.4% | 6.1% | 6.8% |
In the above table, the pattern for tornado warning probabilities is similar to the pattern for tornado warning probabilities in the forecasted severe wind probability table. There is a relatively high likelihood of a tornado warning when the forecasted probability of a significant severe wind event is very low (less than 10%). Above 10%, the probability of a tornado warning generally increases as the forecasted significant severe wind probability increases, but there is a considerable fluctuation when the forecasted probability exceeds 70%. It is not entirely clear why the tornado warning probability drops at 90% and then increases again at 95%, though one potential explanation could be small sample size.
Otherwise, like with the forecasted severe wind probability, the correlation between severe thunderstorm warning probability and forecasted significant severe wind probability is strong (r = 0.95), but there is a notable fluctuation when the forecasted significant severe wind probability is 80%. This pattern suggests that 80% probabilities are being issued when lower probabilities are warranted.
| Forecasted probability of at least 10 severe hail events | 5% | 10% | 20% | 30% | 40% | 50% | 60% | 70% | 80% | 90% | 95% |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 4.8% | 3.4% | 2.4% | 1.9% | 2.9% | 2.6% | 3.8% | 4.2% | 6.8% | 6.7% | 10.1% |
| Severe Thunderstorm Warning | 22.9% | 29.0% | 29.3% | 26.4% | 28.0% | 28.0% | 31.6% | 30.8% | 34.3% | 33.6% | 37.5% |
| Flash Flood Warning | 7.0% | 3.2% | 2.9% | 3.4% | 3.6% | 3.5% | 4.0% | 3.5% | 5.3% | 5.2% | 5.9% |
Ideally, higher forecasted probabilities for hail should be associated with a higher probability of seeing a tornado warning. High-end hail events require an isolated storm mode to sustain long-lived and intense supercells, which also are associated with greater tornado potential. While this expectation does vaguely appear in the above table, there is a great deal of noise evident, and this is reflected in one of the weakest correlation coefficients obtained (r = 0.72). A slightly stronger correlation (r = 0.87) is present between forecasted hail probability and severe thunderstorm warning probability, but there are fluctuations in the severe thunderstorm warning probability as the forecasted hail probability increases. This result suggests that the accuracy of hail probability forecasts can be significantly improved.
| Forecasted probability of at least 1 significant severe hail event | 5% | 10% | 20% | 30% | 40% | 50% | 60% | 70% | 80% | 90% | 95% |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Tornado Warning | 3.5% | 2.6% | 1.8% | 2.9% | 3.2% | 4.9% | 5.2% | 6.2% | 10.0% | 12.4% | 21.8% |
| Severe Thunderstorm Warning | 25.3% | 29.2% | 26.6% | 28.1% | 29.2% | 31.5% | 34.7% | 33.0% | 41.2% | 44.6% | 49.5% |
| Flash Flood Warning | 4.8% | 3.4% | 3.1% | 3.9% | 3.2% | 4.1% | 4.1% | 3.4% | 6.6% | 7.4% | 6.8% |
Similar to the table with forecasted severe hail probability, the table with forecasted significant severe hail probability largely shows a noisy relationship between forecasted probabilities and warning probabilities. However, a steadier and more discernible pattern emerges when the forecasted significant severe hail probability reaches or exceeds 70%. This is likely due to the fact that only the most volatile severe weather setups will prompt significant hail probability forecasts over 60%, and such volatile setups usually prompt numerous warnings. Overall, this table still suggests that the accuracy of hail probability forecasts can be significantly improved.
| Forecasted probability of at least 6 combined severe wind and severe hail events | 5% | 10% | 20% | 30% | 40% | 50% | 60% | 70% | 80% | 90% | 95% |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Tornado Warning | -- | -- | 8.6% | 6.9% | 4.0% | 2.4% | 2.1% | 2.2% | 2.6% | 3.7% | 5.7% |
| Severe Thunderstorm Warning | -- | -- | 2.0% | 5.4% | 16.5% | 20.7% | 26.2% | 24.8% | 28.7% | 33.5% | 37.6% |
| Flash Flood Warning | -- | -- | 15.8% | 11.0% | 5.6% | 3.8% | 3.2% | 3.2% | 3.3% | 4.1% | 4.8% |
Since severe thunderstorm warnings primarily encompass hail and wind, the aggregated hail/wind probability forecast should correlate most strongly with severe thunderstorm warning probabilities. This expectation is confirmed by the above table, which generally shows a steady increase in severe thunderstorm warning probabilities as the aggregated hail/wind probability forecast increases. There is a notable fluctuation when the aggregated hail/wind probability forecast is 70% where the severe thunderstorm warning probability drops. This suggests that the 70% probability is being issued when lower probabilities are warranted.
Since tornado warnings can be issued without severe hail and without severe wind, a weaker correlation is expected. However, a seemingly counterintuitive result was obtained when comparing tornado warning probabilities to the aggregated hail/wind probability forecasts. Generally speaking, the probability of a tornado warning actually decreases as the aggregated hail/wind probability forecast increases (r = -0.51). This result was likely skewed by a very specific yet respectably common circumstance: tornado watches for tropical cyclones. The lapse rates within a tropical cyclone tend to be very shallow, which heavily limits the potential for hail and convective downdrafts (wind). Accordingly, tropical cyclones will prompt tornado watches that are almost exclusively for a tornado threat, which causes the aggregated wind/hail probability forecast to tend lower for tropical cyclones. This effect is compounded by the fact that tropical cyclones often maintain very long-lived tornado potential (prompting longer lasting tornado watches), and numerous tornado warnings are often issued near a tropical cyclone. This conclusion is further supported by the fact that flash flood warning coverage also peaks when the aggregated hail/wind probability is below 40%.
Within the 2007-2025 timeframe, there has never been a watch with an aggregated wind/hail probability forecast of 10% or lower. This is due to the fact that watches are intended to highlight areas where conditions appear favorable for hazardous weather within the timeframe of a watch. Accordingly, issuing a watch for a single severe storm or a plethora of weak storms goes against the intended purpose of a watch. As such, if coverage and/or intensity are too low, a truthful aggregated wind/hail probability forecast would be less than 20%, which usually does not justify a watch.
Part VI: Flash flood warning probability as a function of WPC categorical risk
The Weather Prediction Center (WPC) issues a variety of outlooks for hazards not covered by the SPC. Since flash flooding is the deadliest precipitation-related hazard in the United States, another item of interest is the WPC's excessive rainfall outlook (ERO). The ERO forecasts the probability that flash flood guidance (FFG) will be exceeded within 25 miles of any given location. The FFG represents a best guess of how much liquid precipitation would be needed within a 1-hour, 3-hour, or 6-hour timeframe to result in flash flooding. Put another way, if the rainfall ever exceeds an FFG value at any given location, a natural expectation would be for flash flooding to occur (thereby prompting a flash flood warning).
Similar to the SPC's convective outlooks, the forecasted probability of rainfall exceeding FFG corresponds to a categorical risk level. Currently, the WPC uses the following risk categories: no risk, marginal risk, slight risk, moderate risk, high risk. If the risk is higher, the probability of experiencing a flash flood warning should be higher, and this supposition is confirmed in the below table:
| Warning Type | No Risk | Marginal Risk | Slight Risk | Moderate Risk | High Risk |
|---|---|---|---|---|---|
| Flash Flood Warning | 0.012% | 0.44% | 3.4% | 19.3% | 47.2% |
It should be noted that, in 2022, the WPC significantly raised the threshold for a moderate risk (neighborhood probability changed from 20% to 40%) and a high risk (neighborhood probability changed from 50% to 70%). However, when examining the coverage of flash flood warnings by year, no significant change was evident between pre-2022 risks and post-2022 risks. In other words, anyone within a high risk before 2022 was just as likely to see a flash flood warning as someone within a high risk after 2022, and the same also holds true for a moderate risk. Since the coverage of flash flood warnings better fits the old threshold values (19.3% ≈ 20%, 47.2% ≈ 50%), this finding suggests that the aforementioned threshold change probably should be reverted.
Comparing these results to the flash flood warning probability as a function of SPC risk (see Part II and Part III), SPC slight risk days have flash flood warning probabilities (2.5%) comparable to WPC slight risk days (3.4%), especially those that involve a 5% tornado risk (4.7%). This finding would suggest that WPC should, on average, be issuing slight risks for excessive rainfall in and around areas where the SPC has a slight risk of severe thunderstorms. Additionally, the probability of experiencing a flash flood warning on an SPC high risk day (27.0%) is slightly greater than the probability of experiencing a flash flood warning on a WPC moderate risk day (19.3%). This suggests that WPC should, on average, issue at least a moderate risk of excessive rainfall in and around areas where the SPC has a high risk of severe thunderstorms.
Part VII: Warning probabilities for other hazards as a function of watch existence
Although the main focus of this research has been directed towards the "short-fused" warnings (tornado warnings, severe thunderstorm warnings, and flash flood warnings), another item of interest might be the rate at which longer lead time watches are converted to warnings. After all, if a watch for a highly impactful hazard (e.g. a hurricane) is highly likely to become a warning, it would be worthwhile to treat the watch as if it were a warning and prepare accordingly.
The following longer lead time hazards were examined:
- Coastal Flood Watch: issued when conditions appear favorable for life-threatening flooding along a major coastline.
- Extreme Cold Watch: issued when life-threatening cold temperatures or wind chills are expected.
- Extreme Heat Watch: issued when life-threatening hot temperatures are expected (formerly known as "excessive heat watch").
- Fire Weather Watch: issued when conditions appear favorable for "critical fire weather" conditions.
- (Flash) Flood Watch: issued at 24+ hour lead times when conditions are expected to favor flash flooding.
- Freeze Watch: issued when conditions appear favorable for the first freeze of the cold season or a freeze in the warm season.
- High Wind Watch: issued when conditions appear favorable for damaging non-thunderstorm winds.
- Hurricane Watch: issued when hurricane conditions appear possible at a particular location.
- Storm Surge Watch: issued when life-threatening storm surge flooding (usually from a tropical cyclone) is expected.
- Tropical Storm Watch: issued when tropical storm conditions appear possible at a particular location.
- Winter Storm Watch: issued when conditions appear favorable for a life-threatening winter precipitation event.
Tropical weather watches and winter weather watches can branch off into multiple different warnings depending on what appears most likely. For instance, a tropical storm watch can be issued initially, but it can later be upgraded to a tropical storm warning or a hurricane warning. Similarly, a winter storm watch can be issued initially, but it can later be upgraded to a winter storm warning, an ice storm warning, or a blizzard warning. Accordingly, these results will be presented in a more detailed table. The results for all other alerts are presented below:
| Watch Type | Warning Probability |
|---|---|
| Coastal Flood | 43.2% |
| Extreme Cold | 50.9% |
| Extreme Heat | 55.2% |
| Fire Weather | 67.4% |
| (Flash) Flood | 10.1% |
| Freeze | 73.3% |
| High Wind | 52.1% |
| Storm Surge | 79.0% |
To read this table, find the hazard that a watch is currently in effect for. The probability that a warning for the same hazard will be issued is provided in the second column. For example, if a coastal flood watch is in effect for your location, the probability of you seeing a coastal flood warning is 43.2%.
One result that really stands out is the low watch-to-warning conversion rate for (flash) flood watches. The watch-to-warning conversion rates for all other hazards is about or above 50%, so the result for (flash) flood watches is significantly lower. One could definitely make the argument that (flash) flood watches are being issued too frequently based on this result. Even though flash flooding is admittedly a localized hazard, the WPC can issue risk areas that have warning coverages much greater than 10%, so it is possible to increase the watch-to-warning conversion rate for flash flooding. Otherwise, if about 10% of a (flash) flood watch area becomes a warning, that means 90% of the watch area will not even see a warning. That potentially is a problematically large false alarm rate for the deadliest precipitation-related hazard in the United States.
That being said, the probability of experiencing a flash flood warning within a (flash) flood watch is still higher than the probability of experiencing a flash flood warning within a tornado watch or a severe thunderstorm watch. This does suggest that (flash) flood watches demonstrate forecast skill to some extent, but one could definitely argue that the watch-to-warning conversion rate for flash flooding should be higher than 10%.
| Warning Type | Probability Given A Winter Storm Watch |
|---|---|
| Winter Storm Warning | 47.0% |
| Blizzard Warning | 8.3% |
| Ice Storm Warning | 2.3% |
| Any Warning | 54.5% |
When examining winter precipitation specifically, about 54.5% of all winter storm watches become a precipitation-related warning of some form. This is on par with the watch-to-warning conversion rates for most other long lead time hazards.
Blizzard warnings and ice storm warnings are relatively rare because they require a specific combination of ingredients, which normally occurs in a small area. Blizzards require a combination of heavy snow (or blowing snow), sustained winds of 35+ mph, visibility under 1/4 mile, and for these conditions to last at least 3 contiguous hours. Even within the most intense winter cyclones, a combination of high winds and heavy snow normally occurs within a narrow corridor.
Ice storms require persistent heavy freezing rain leading to ice accumulations greater than 0.25". This normally occurs in a narrow corridor just ahead of a warm front, and the surface air temperature ahead of said warm front must also fall within a specific range (usually between 27 °F and 31 °F). Since the conditions required for an ice storm are so precise, it makes sense why ice storm warnings, on a national scale, get issued less frequently than blizzard warnings.
| Warning Type | Probability Given A Tropical Storm Watch | Probability Given A Hurricane Watch |
|---|---|---|
| Tropical Storm Warning | 80.9% | 74.5% |
| Hurricane Warning | 9.2% | 63.0% |
In the above table, tropical storm watches are converted to tropical storm warnings at a very high rate (over 80%). Similarly, tropical storm watches do not often become hurricane warnings (probability less than 10%). With this result in mind, persons under a tropical storm watch should treat the tropical storm watch as if it were a tropical storm warning. Overall, this indicates that the accuracy of tropical storm watches is high (not necessarily being issued too frequently or too rarely).
A slightly more concerning result becomes evident when looking at hurricane watches. The above table shows that a hurricane watch is more likely to result in a tropical storm warning than a hurricane warning. Ideally, hurricane watches should be converted to hurricane warnings at a higher rate, so this suggests that hurricane watches are being issued when a tropical storm watch is more appropriate.
Conclusions
This analysis has revealed a discernible pattern between warning probabilities and longer-lead time products (watches and outlooks). This information can help provide some perspective on what exactly is expected to happen within a watch or an SPC risk area. How to present this information to the general public in the most effective way possible is a potential avenue of future work for social scientists. However, NOAA offices, emergency management agencies, venue managers, and politicians could theoretically benefit from having the information provided in this analysis.
Aside from the potential messaging improvements, this analysis also establishes that there is an appreciable amount of forecast skill within watches and outlooks. This is especially impressive when one considers that individual NWS offices have different philosophies when it comes to issuing warnings, so the existence of such a clear pattern (Pearson correlation coefficients generally 0.90-0.99) despite these potential inconsistencies is a testament to the quality of these watches and outlooks. In other words, if these watches and outlooks were unskilled forecasts, the warning probabilities would show a more chaotic and unexplainable pattern.
Appendix A: Tornado warning probability as a function of severe thunderstorm warning existence
In most cases, the first warning issued for a severe storm is a severe thunderstorm warning, which can later be "upgraded" to a tornado warning. Naturally, this prompts some questions about how frequently an upgrade occurs, what the probability of an upgrade is at any given location within the severe thunderstorm warning, and whether these values correlate with other discernible details.
First, the unconditional averages will be presented:
| Upgrade Probability | Average Overlap | Total Probability |
|---|---|---|
| 6.0% | 36.0% | 2.2% |
Nationally, about 6% of all severe thunderstorm warnings have at least part of their polygon upgraded to a tornado warning while the severe thunderstorm warning is still in effect. When an upgrade occurs, the tornado warning polygon covers, on average, about 36% of the severe thunderstorm warning's area. Multiplying these two probabilities together yields the overall probability of an upgrade occurring at any given point within a severe thunderstorm warning, which is about 2.2%. In other words, if you're under a severe thunderstorm warning, there is a 2.2% chance (on average) that you will experience a tornado warning while the severe thunderstorm warning is in effect.
These values might seem a bit low, but National Weather Service offices usually stop issuing severe thunderstorm warnings on a storm with a tornado warning as long as the tornado threat is deemed to be high. If NWS offices issued severe thunderstorm warnings along with tornado warnings for the same storm, the resulting upgrade probability and overlap values would be higher. Once the tornado threat eventually decreases, the tornado warning will be replaced with a severe thunderstorm warning (unless the storm also weakens below severe limits). In other words, not every tornado warning will be preceded by a severe thunderstorm warning, and a severe thunderstorm warning does not always follow a tornado warning.
During the summer time, it is more likely for disorganized storms to prompt severe thunderstorm warnings. Such disorganized storms are much less likely to prompt tornado warnings, so a natural expectation would be for more severe thunderstorm warnings to be upgraded to tornado warnings in other months. This expectation is confirmed in the below table:
| Month | Upgrade Probability | Average Overlap | Total Probability | Ratio To Baseline |
|---|---|---|---|---|
| January | 17.6% | 35.8% | 6.3% | 2.90 |
| February | 12.8% | 35.2% | 4.5% | 2.07 |
| March | 11.3% | 38.5% | 4.3% | 1.99 |
| April | 11.0% | 36.6% | 4.0% | 1.85 |
| May | 7.9% | 36.1% | 2.9% | 1.31 |
| June | 4.2% | 34.1% | 1.4% | 0.66 |
| July | 2.8% | 33.4% | 1.0% | 0.44 |
| August | 2.4% | 35.7% | 0.9% | 0.40 |
| September | 3.3% | 38.2% | 1.3% | 0.57 |
| October | 8.3% | 38.8% | 3.2% | 1.49 |
| November | 14.5% | 36.5% | 5.3% | 2.43 |
| December | 16.1% | 36.6% | 5.9% | 2.71 |
Based on the above table, any given severe thunderstorm warning in a winter month is considerably more likely to be upgraded to a tornado warning, and such an upgrade is most likely to occur in January. In fact, any given location within a severe thunderstorm warning in January is almost 3x more likely to experience a tornado warning (indicated by the rightmost column) when compared to the unconditional average value of 2.2%.
Another natural expectation would be for more intense storms to prompt more tornado warnings. Intensity can be roughly gauged by the expected magnitude of two hazards: straight-line wind gusts and hail. Although storms that produce widespread wind damage are more likely to be linear in nature (which is usually less favorable for tornadoes), more intense thunderstorm lines often produce more tornadoes along the leading edge of the line. With that in mind, a correlation should exist between a severe thunderstorm warning's wind gust and the probability of being upgraded to a tornado warning. Sure enough, that expectation is confirmed below:
| Warning Wind Gust | Upgrade Probability | Average Overlap | Total Probability | Ratio To Baseline |
|---|---|---|---|---|
| < 58 mph | 4.6% | 45.5% | 2.1% | 0.96 |
| 60 mph | 5.3% | 37.0% | 2.0% | 0.91 |
| 70 mph | 11.8% | 31.8% | 3.7% | 1.72 |
| 80 mph | 17.6% | 27.1% | 4.8% | 2.19 |
| 90 mph | 27.3% | 25.2% | 6.9% | 3.16 |
| 100 mph | 30.0% | 25.8% | 7.7% | 3.55 |
There are a few interesting trends evident in this table that are worth discussing. One is that severe thunderstorm warnings for 80+ mph winds get upgraded to a tornado warning at a very high rate. However, the areal overlap between tornado warning and severe thunderstorm warning generally decreases as the expected wind speed increases. This is likely due to the fact that severe thunderstorm warnings for thunderstorm lines tend to be larger, so a subsequent tornado warning is more likely to overlap with at least a portion of the severe thunderstorm warning polygon. At the same time, tornado warning polygons are usually small, so issuing a small polygon within a larger polygon (as is often the case for severe thunderstorm warnings for 80+ mph winds) will also result in less areal overlap. Even so, the net result of these two competing effects is a positive correlation (r = 0.84) between severe thunderstorm warning wind speed and the probability of experiencing a tornado warning within the severe thunderstorm warning.
Another feature worth highlighting is the result for "sub severe" winds (gusts under 58 mph). Severe thunderstorm warnings for sub severe wind gusts are upgraded to tornado warnings at a rate roughly equivalent to the national average. However, when an upgrade does occur, the subsequent tornado warning covers, on average, almost half of the severe thunderstorm warning polygon. So, a valid interpretation of this particular result would be "this probably will stay a severe thunderstorm warning, but, if it is upgraded to a tornado warning, there is an almost 50% chance you're going to be included in the tornado warning".
Looking at the same table, but for severe thunderstorm warning hail sizes:
| Warning Hail Size | Upgrade Probability | Average Overlap | Total Probability | Ratio To Baseline |
|---|---|---|---|---|
| < 0.75" | 5.7% | 37.0% | 2.1% | 0.98 |
| 0.75" | 4.6% | 35.1% | 1.6% | 0.74 |
| 0.88" | 4.7% | 36.7% | 1.7% | 0.79 |
| 1.00" | 4.9% | 37.1% | 1.8% | 0.83 |
| 1.25" | 5.4% | 36.3% | 2.0% | 0.90 |
| 1.50" | 7.3% | 36.2% | 2.7% | 1.22 |
| 1.75" | 10.5% | 38.2% | 4.0% | 1.84 |
| 2.00" | 12.0% | 34.9% | 4.2% | 1.92 |
| 2.50" | 15.0% | 34.0% | 5.1% | 2.35 |
| 2.75" | 17.1% | 35.8% | 6.1% | 2.82 |
| 3.00" | 19.6% | 33.4% | 6.6% | 3.01 |
| 4.00" | 26.7% | 33.8% | 9.0% | 4.15 |
| 4.50" | 23.1% | 42.1% | 9.7% | 4.47 |
One immediately evident pattern is that larger hail sizes are correlated with a greater likelihood of a severe thunderstorm warning being upgraded to a tornado warning (r = 0.99). This makes sense, because very large hail is almost exclusively produced by supercells, which is also the type of storm most likely to produce a tornado. However, there are a few anomalies in this table worth addressing.
When the hail size is less than 0.75", there is a relatively high likelihood of a severe thunderstorm warning being upgraded to a tornado warning. This result is likely attributable to the fact that intense thunderstorm lines often produce little in the way of hail, so this is mostly reflecting a tendency for thunderstorm lines to prompt more tornado warnings than storms that are considered "marginally severe".
Another fluctuation occurs when the severe thunderstorm warning hail size is 4.50" (grapefruit size). The decrease in upgrade probability could be due to small sample size (on average, about 2 severe thunderstorm warnings with this hail size are issued per year nationally), but it could also be related to the fact that most extreme hail events occur during the early summer when tornadoes and tornado warnings are also slightly less common. Despite the lower upgrade probability, the combined probability is still the highest out of all the hail sizes.
Finally, another attribute of severe thunderstorm warnings worth examining is the existence or non-existence of the "Tornado...Possible" tag. National Weather Service forecasters can include this tag in a severe thunderstorm warning if a storm has some potential for producing a tornado, but the potential is not high enough to justify a tornado warning. Since the inclusion of this tag is a binary outcome (it either exists or it doesn't) and the tag did not become universally standardized until ~2015, it is only sensible to examine what the effect the tag's inclusion has on the previously examined metrics.
| Upgrade Probability | Average Overlap | Total Probability | Ratio To Baseline | |
|---|---|---|---|---|
| Tornado...Possible Included | 25.5% | 35.5% | 9.1% | 4.17 |
It is conceivable that the "Tornado...Possible" tag can be a precursor to a formal tornado warning. Storms that are still maturing can exhibit tornado potential in their infancy, but the storm's maximum tornado potential may not be realized until some time later. In other words, the "Tornado...Possible" tag can act as an intermediate step between a storm first becoming severe and a storm prompting its first formal tornado warning. Sure enough, the "Tornado...Possible" tag has a significant effect on the probability of a severe thunderstorm warning being upgraded to a tornado warning (by at least a factor of 4). Since a probability of 9.1% is roughly equivalent to the probability of a tornado warning given a tornado watch (7.0%), it would be sensible to treat a severe thunderstorm warning with a "Tornado...Possible" tag as if it were a localized tornado watch.
Appendix B: Warning probability as a function of mesoscale discussion existence
Both the SPC and WPC can issue "mesoscale discussions", which generally highlight specific areas where hazardous weather is expected within the next few hours. The SPC can issue mesoscale discussions for short-term severe weather potential, winter weather potential, and (on rare occasions) categorical upgrades to the convective outlook. The WPC issues mesoscale (precipitation) discussions for short-term heavy rainfall and flash flooding potential. With this in mind, it would be reasonable to expect that severe weather mesoscale discussions may correlate with the probability of experiencing a tornado warning or a severe thunderstorm warning. It would also be reasonable to expect a correlation between mesoscale precipitation discussions and the probability of experiencing a flash flood warning.
Severe weather mesoscale discussions can be further broken down into different categories. Some mesoscale discussions are issued when severe thunderstorm activity may prompt the issuance of a convective watch. Any such discussions also include a statement on the likelihood of a watch being issued. Other discussions are issued for ongoing watches, either as an update of what is expected or to highlight a particularly problematic storm or storm cluster.
Since 2007, all severe weather mesoscale discussions for potential watches include a qualitative statement about the likelihood of a watch being issued. Possible statements include: "watch unlikely", "watch possible", "watch likely", "watch needed soon", "severe thunderstorm watch likely", "tornado watch likely". Since watches are more likely to be issued for higher-end severe weather events, a reasonable expectation would be for higher watch issuance probabilities to positively correlate with higher warning issuance probabilities. It would also be reasonable to expect higher tornado warning probabilities if the phrase "tornado watch likely" is used.
The results for the qualitative likelihoods are given below:
| Phrase | Tornado Warning Probability | Severe Thunderstorm Warning Probability |
|---|---|---|
| Watch Unlikely | 0.26% | 4.6% |
| Watch Possible | 0.33% | 5.1% |
| Watch Likely | 0.42% | 5.5% |
| Watch Needed Soon | 0.34% | 5.0% |
| Severe Thunderstorm Watch Likely | 0.10% | 5.5% |
| Tornado Watch Likely | 0.97% | 3.8% |
One important thing to note right away is that these probabilities are lower than what is presented in the table for warning probability as a function of watch existence. One reason why these probabilities are lower is because severe weather mesoscale discussions do not last as long as watches (usually 1-2 hours), so fewer warnings will be issued within a shorter timeframe. Another reason for these low values is because SPC forecasters ideally try to issue a mesoscale discussion about 1 hour before they issue a watch, and their goal is to issue the watch before any warnings are issued. So, severe weather mesoscale discussions for potential watches are, to some extent, intended to exclude warnings.
Even with all that said, there is still a discernible pattern here. Generally, as the watch probability increases, the probability of a tornado warning and severe thunderstorm warning increases. However, there is a slight fluctuation when the phrase "watch needed soon" is used, and this could be attributed to small sample size (this phrase is rarely used at all).
When looking at "severe thunderstorm watch likely" and "tornado watch likely" specifically, the pattern matches what is expected; tornado warning probability is higher when "tornado watch likely" is used. Although, one possibly surprising result is that "severe thunderstorm watch likely" has a tornado warning probability lower than "watch unlikely".
Even though these numbers are low, they are still significantly higher than the probability of experiencing a warning with no watch in effect (0.038% for tornado warnings and 0.43% for severe thunderstorm warnings). So, even if the SPC does not plan on issuing a watch, the existence of a mesoscale discussion increases the probability of experiencing a warning by about a factor of 5-10. Therefore, it might be worthwhile to tell the general public to pay a little extra attention to the weather if a mesoscale discussion is issued.
Since 2012, the SPC has also included a quantitative probability for a watch issuance in each mesoscale discussion for a potential watch. As of 2025, 6 different probabilities were used operationally: 5% (unlikely), 20% (unlikely), 40% (possible), 60% (possible), 80% (likely), and 95% (likely). Examining the quantitative results:
| Watch Probability | Tornado Warning Probability | Severe Thunderstorm Warning Probability |
|---|---|---|
| 5% | 0.22% | 4.2% |
| 20% | 0.29% | 5.4% |
| 40% | 0.32% | 5.7% |
| 60% | 0.36% | 6.0% |
| 80% | 0.45% | 5.8% |
| 95% | 0.66% | 6.5% |
Overall, the probability of a watch being issued correlates most strongly with the probability of experiencing a tornado warning. This result matches expectations, however, severe thunderstorm warning probability exhibits a less consistent (albeit still positively correlated) pattern. It would seem that once the watch probability reaches 60%, the probability of experiencing a severe thunderstorm warning asymptotes to about 6%.
The other category of severe weather mesoscale discussions is for watches already in effect. Any such mesoscale discussions are likely to be issued while storms are ongoing within the watch(es), so it would reasonable expect the probabilities for tornado warnings and severe thunderstorms to be higher than in the above tables.
| Watch Type | Tornado Warning Probability | Severe Thunderstorm Warning Probability |
|---|---|---|
| Severe Thunderstorm Watch | 0.45% | 12.2% |
| Tornado Watch | 3.7% | 11.7% |
As expected, these probabilities are higher, but they are still significantly lower than the table presenting warning probabilities as a function of watch existence. Again, it is important to note that severe weather mesoscale discussions typically are in effect for 1-2 hours (contrasted to watches, which usually last 3-9 hours). Relatively few warnings will be issued in such a short timeframe, so this could explain why these probabilities are about half of the values presented in Part I.
As for WPC mesoscale precipitation discussions, there are four different statements that can be included in the discussion text: no statement about the likelihood of flash flooding, flash flooding unlikely, flash flooding possible, and flash flooding likely. Operationally, the "flash flooding unlikely" statement is exceedingly rare (issued 3 times in the 2013-2025 timeframe), because it is normally used to mark the end of an extreme flooding event. Otherwise, the most reasonable expectation would be for flash flood warning probabilities to be higher for "flash flooding likely" than "flash flooding possible".
| Phrase | Flash Flood Warning Probability |
|---|---|
| No Probability Statement | 0.48% |
| Flash Flooding Unlikely | 0.0% |
| Flash Flooding Possible | 3.1% |
| Flash Flooding Likely | 9.4% |
Overall, this matches the expected pattern. However, there are a few finer details worth emphasizing. One is that the rarely used "flash flooding unlikely" phrase does a great job of marking the end of a flash flooding event. If this were not the case, the probability of a flash flood warning would be considerably larger than 0.0%.
Another detail (not presented in the above table) is that the flash flood warning probabilities for mesoscale precipitation discussions have been generally declining in recent years (down from about 11% to about 9% for "flash flooding likely", and from about 5% to about 3% for "flash flooding possible"). One potential explanation for this decline is that the WPC has been issuing more mesoscale discussions in recent years (up from about 800 per year to about 1100 per year), which is perhaps leading to more false alarms. It would be difficult to call this downward trend problematic for when the phrase "flash flooding possible" is used, but the downward trend for "flash flooding likely" suggests this phrase is being overused. Ideally, the flash flood warning probability for "flash flooding likely" should be higher than the watch-to-warning conversion rate for (flash) flood watches/warnings (presented in Part VII), because WPC mesoscale discussions are issued at shorter lead times and are (unlike a lot of SPC mesoscale discussions) specifically intended to coincide with the issuance of flash flood warnings.
Appendix C: Warning probability as a function of tropical cyclone warnings
Tropical cyclones can pose a wide variety of hazards to mariners and populated areas. These hazards primarily include rough waters, storm surge, high winds, heavy rainfall, and tornadoes. The National Weather Service can issue warnings to address some of these hazards specifically, such as a hazardous seas warning, storm surge warning, hurricane warning (for both land and sea), tropical storm warning (for both land and sea), a flash flood warning, a special marine warning, or a tornado warning. Since the bulk of this study has focused on short-fused warnings, the below table will display the probability of experiencing a tornado warning or a flash flood warning based on the existence of a hurricane warning or tropical storm warning:
| Warning Type | Tropical Storm Warning | Hurricane Warning | Any Warning |
|---|---|---|---|
| Tornado Warning | 9.6% | 13.0% | 10.3% |
| Flash Flood Warning | 22.6% | 35.8% | 25.3% |
There is a slightly higher probability of experiencing a tornado warning within a hurricane warning than a tropical storm warning. However, it should be emphasized that hurricane warnings tend to cover a smaller area than tropical storm warnings, so the probability value could be inflated. Still, this result would imply that locations under a hurricane warning are slightly more likely to experience a tornado warning than locations under a tropical storm warning. Additionally, the probability of a tornado warning within any tropical cyclone warning is slightly higher than the probability of experiencing a tornado warning while under a tornado watch (see Part I). With that in mind, it might be worthwhile to treat tropical storm warnings and hurricane warnings as long duration tornado watches.
The tornado warning probabilities within a tropical cyclone warning are also very similar to the tornado warning probabilities within an SPC 10% tornado probability (see Part III). Historically, the SPC has mostly issued 5% tornado probabilities for tropical cyclones, and this result suggests that 10% tornado probabilities should, in most circumstances, be issued for tropical cyclones.
Flash flood warnings are significantly more likely to occur within a hurricane warning than a tropical storm warning. While the small area of hurricane warnings could be inflating this result as well, there is reason to believe that this pattern will hold true in almost every circumstance. Since hurricanes are (by definition) stronger than tropical storms, higher condensation rates are usually occurring in the middle troposphere within a hurricane, which also favors more intense rainfall.
Another important point to make is how the flash flood warning probability is greater than 20%. When examining the flash flood warning probability as a function of WPC risk (see Part VI), this would imply that tropical cyclones should, on average, prompt at least a moderate risk of excessive rainfall from the WPC. Since the flash flood warning probability for hurricanes is above 35%, this implies that a significant fraction (if not a majority) of hurricanes deserve a high risk of excessive rainfall from the WPC.
Severe thunderstorm warnings are rarely issued for tropical cyclones, because tropical cyclones do not produce hail, and the shallow lapse rates significantly limit the potential for convective straight-line wind damage. To address the threat for high winds near the cyclone center, an "extreme wind warning" is used.
Appendix D: Fire warning probability as a function of SPC fire weather risk
In addition to issuing convective outlooks, the SPC also issues "fire weather" outlooks, which highlight areas where atmospheric conditions are favorable for wildfires. Within these outlooks, there are 6 different risks that the SPC can issue: no risk, elevated risk, critical risk, extremely critical risk, isolated (10% coverage) dry thunderstorms, and scattered (40% coverage) dry thunderstorms.
The elevated, critical, and extremely critical risks apply to combinations of warm temperatures, high winds, low relative humidity, and dry fuels (vegetation that will burn easily). The severity and behavior of a fire will only be realized if a fire is started at all. Theoretically, it is possible for an extremely critical risk to be issued, but for no fires to occur. Even though conditions might be favorable for extreme wildfire behavior, they do not explicitly predict the occurrence of wildfires, because a fire would need to be igniting by a non-meteorological trigger. Such triggers might include a carelessly discarded cigarette, a car accident, a campfire, a downed power line, or a structure fire.
On the other hand, the dry thunderstorm risks do explicitly predict the occurrence of wildfires, because the expectation is for a meteorological phenomenon (lightning) to trigger the fires. This usually only occurs if thunderstorms are producing very little to no rainfall, but are producing lightning (hence the name "dry thunderstorm"). The risk level is determined by the expected coverage of dry thunderstorms (10% for "elevated", 40% for "critical"). Comparatively speaking, dry thunderstorm risks are rare. Moisture is a key ingredient for thunderstorms, so it is difficult (but not impossible) for the atmosphere to support thunderstorms in moisture-deprived conditions.
If a wildfire breaks out and is impossible to contain, local emergency managers can issue a "fire warning", which is an alert relayed by the National Weather Service. A fire warning is usually intended to alert people that they should either evacuate immediately or be prepared to evacuate if the wildfire continues to burn uncontrollably. In that respect, a "fire warning" could also be considered a "short-fused warning". However, not all wildfires prompt fire warnings, and this is partly due to the fact that labeling a fire "uncontrollable" is largely a judgment call by first responders in the field.
Since the issuance of fire warnings can be an inconsistent process, it is difficult to predict the probability of a fire warning based on the SPC fire weather risk. However, it is possible to estimate what this probability should be based on the fires that were recorded and adding a buffer (radius) around each recorded fire. A radius of ~20 kilometers (~12 miles) was chosen as the buffer, because fire warning polygons tend to be about one-fourth the area of tornado warning polygons, and the 40 kilometer (25 mile) radius for tornadoes seems to match the tornado warning probability pretty well (see Part III). With all that in mind, the expected probability of being in a fire warning given an SPC fire weather risk is:
| Elevated Risk | Critical Risk | Extremely Critical Risk | |
|---|---|---|---|
| Expected Fire Warning Probability | 0.20% | 0.40% | 2.8% |
| 10% Dry Thunderstorm Risk | 40% Dry Thunderstorm Risk | |
|---|---|---|
| Expected Fire Warning Probability | 0.49% | 0.56% |
Generally, a positive correlation does exist between the fire weather risk level and the expected fire warning probability, which indicates good forecast performance. However, the values for dry thunderstorm risk are problematically similar. Even though the 40% dry thunderstorm risk does yield a slightly higher fire warning probability, this number should ideally be at least 4 times larger than the fire warning probability for 10% dry thunderstorm risks since the coverage is at least 4 times larger. Granted, there are probably scenarios when the expected dry thunderstorm coverage might be 20% or 30% (too low to justify a 40% area), and these scenarios could be inflating the expected fire warning probability for 10% dry thunderstorm risks. In which case, the SPC might consider adding more probability levels to the dry thunderstorm category (e.g. 10%, 25%, and 40%; or 10%, 20%, 30%, and 40%).
These probabilities may seem low, but it's important to remember that wildfires are a rare phenomenon at any given point (usually about 500-1000 grassland fires are reported in the United States per year). Therefore, it is worthwhile to take the critical and extremely critical risk areas seriously (especially the extremely critical risk).