Everything You Need To Know About Fall Frost In Cincinnati

Low and high temperatures have been above average the last few days in the Queen City, but the latest computer guidance suggests cool, Canadian air will arrive by next weekend. While there is no imminent threat for frost with this next shipment of cool air, the likelihood of frost will rapidly increase over the next month.

When forecasting frost, a meteorologist often looks for a very light or calm wind, a mostly clear to clear sky, and temperatures dropping into into or below the upper 30s early in the morning. Frost can form with air temperatures dropping into the mid and upper 30s; temperatures to or below 32° are not needed for frost. Why? The answer to this has to do with how temperatures are measured.

The piece of equipment used to measure weather conditions at most airports in this country – called an Automated Surface Observing System or ASOS – measures the temperature 2 meters above the ground.Here is a picture of an ASOS and where the temperature sensor is located:


Because relatively cold air sinks and warm air rises, temperatures below this sensor are always colder than temperatures at the sensor. For example, the sensor may measure at air temperature of 36°, but the temperature at the ground may be 32° or lower. Patchy frost can form at the ground when the temperature at the sensor drops to 38°.

The first frost of the fall in Cincinnati almost always occurs in October; while the exact temperature where frost occurs can vary, using a temperature of 36° or 38° yields roughly the same dates on average:


The first frost has occurred as early as mid September and as late as late November.

Frost does not necessarily mean the end of the growing season, but frost can easily kill plants – especially if they are sensitive to cold. A freeze or hard freeze signals the end of the growing season for all seasonal vegetation. On average, the first freeze or hard freeze of the fall in Cincinnati occurs in late October or early November:


Note that a freeze has occurred as early as late September, and a hard freeze has occurred as early as early October.

The averages and the range of dates can be helpful, but there are more than two ways to measure first fall frost dates. Statistically, the most common ways to measure the “center” of a set of data are involve taking the mean, median, and mode. Simply put, the “mean” is the average, the “median” is the “center” value in the chronological list, and the “mode” is the most common value occurring in the list. For example, if we assume the first fall frost occurs when the temperature (measured 2 meters above the ground) drops to 38°, here are the mean, median, and mode dates for the first fall frost in Cincinnati:


If we assume the first fall frost occurs when the temperature drops to 36°, here are the mean, median, and mode dates for the first fall frost in Cincinnati:


These averages are based on data from 1871 to 2013. What is the mean, median, and mode date for our first freeze?


The mean, median, and mode dates for the first hard freeze in Cincinnati are:


Historically and statistically, if you assume the first fall frost occurs when the temperature drops to 38°, there’s a 75% chance we get our first frost by October 15th. The date slides about one week later if you use 36° as a temperature:


At this point, the cold blast coming in the wake of Friday’s cold front does not look to bring widespread frost to the Tri-State. Longer-range computer guidance suggests temperatures will warm back near or above average by the middle part of next week. Beyond the first full week of October, guidance suggests waves of cold air will gradually push southeast from southern Canada. While the first of these series of cold blasts may not bring frost, reinforcing shots of cold, Canadian air in October suggests our first fall frost will come very close the historical average date.

Posted in Weather | Leave a comment

What Happened To Tuesday Night’s Forecast?

As you probably noticed, the chance for showers and storms in the Tri-State Tuesday night was overdone. Even on-air at 5pm Tuesday, I mentioned a threat for severe weather – especially west of Cincinnati – and scattered showers and storms. I played the threat down on-air tonight, but the radar tonight suggested I didn’t play it down enough.

At 2pm Tuesday, the Storm Prediction Center had the western half of the Tri-State and most of Indiana under a slight risk for severe storms:


As a broadcast meteorologist, you need to present this threat, even if you don’t fully agree with the Storm Prediction Center. I’ll agree with the Storm Prediction Center that there was adequate low-level moisture, warmth, and instability available for thunderstorms to form and become strong or severe.

The missing piece – or at least the piece that was most in question – was the upper-level support. Tuesday morning’s NAM model showed the upper-level disturbance (yellow/orange/red) approaching Cincinnati (the black dot) at 8pm Tuesday:


Ahead of a disturbance like this, lift in the atmosphere increases, and the chance for storms – including severe storms – increases. Tuesday morning’s GFS model showed a disturbance of similar strength approaching the Tri-State at 8pm Tuesday:


Two models showing a disturbance moving through the Ohio Valley raises confidence about the coverage and intensity of showers and storms. The GFS model was farther south with the disturbance, and the NAM model had the disturbance more spread out.

It is hard to verify exactly where these disturbances were Tuesday evening (given they are 18,000 feet above the ground), but the 9pm run of the RAP model (which updates every hour) is our best hope for verification. Here’s where it placed upper-level disturbances at 10pm Tuesday night:


Clearly, the RAP has the strongest disturbance northwest of Cincinnati and has it more compact than the NAM model. Showers and storms developed well to the northwest of Cincinnati on the nose of the upper-level disturbance Tuesday afternoon, and outflow from the early evening storms supported new storm development through late evening (as 11pm Tuesday radar shows):


Models clearly missed the mark, and as a result, the forecast could have been better. Just like in the winter where the exact track has a big impact on snowfall totals, the track of this disturbance had a big impact on the placement and strength of showers and storms.


Posted in Uncategorized | Leave a comment

My Take On Upcoming SPC Severe Weather Risk Changes

For years (really decades), the Storm Prediction Center has issued severe weather risks for the contiguous United States using “slight,” “moderate,” and “high” risk categories. Areas of the country that are most likely to see severe weather are typically placed under a “slight” risk for severe storms at least several times a year, a “moderate” risk up to a couple of times per year, and a “high” risk once every year or two when a major severe weather outbreak is expected.

But all of this is about to change. The 3-categories currently used to classify severe weather will soon expand to 5-categories.

In this interview, Greg Carbin with the Storm Prediction Center says “in the modern era, with the Internet, anybody can look at these [severe threat] graphics. So it’s our responsibility to convey that risk information in a way that’s a little easier to understand for the lay person as opposed to the expert. [...] What we are hoping to do with these categories is convey that risk with meaningful words, colors, and numbers.”

What does this change look like? Here’s a breakdown of what the severe weather risk categories are now and what they will be beginning October 22nd:


In simple terms, the “moderate” and “high” risk categories really won’t be changing this fall, and the current “slight” risk category will be broken down into 3 different categories (marginal, slight, and enhanced). Technically, the “marginal” risk will be a new category just under the current “slight” risk category.

Higher-end risks will still be higher-end risks. Late in the morning of March 2nd, 2012, the Ohio Valley was covered with a “moderate” to “high” risk for severe weather:


Had this same severe weather outlook been issued using the 5-category severe weather outlook coming this fall, the risk for severe weather in the heart of Ohio Valley would have been the same:


The risk for severe weather would have been labeled differently for the northern Indiana, northern Ohio, and parts of the Tennessee Valley.

The changes from SPC were really made to break down lower-end severe weather threats more. Last Tuesday morning, a “slight” risk for severe storms was issued for much of New England:


Had this same outlook been issued using 5 severe weather categories, a “marginal” risk for severe storms would have surrounded the “slight” risk from central New England through the Carolinas:


If you’re confused with all of the changes, you’re not alone. If you’re confused by the current outlook categories, I understand that, too. Here’s the probability table (based on a severe weather report occurring within 25 miles of a point) used by SPC forecasters to draw the severe weather threat for the current day:

If you think that’s messy, here’s the probability table SPC forecasters will use to issue severe weather outlooks beginning October 22nd:


There are different tables for days 2 and 3 of the forecast. Making these forecasts is a challenge given model uncertainties and forecast time constraints. Creating these outlooks is not an easy job.

Why are the outlooks changing? I’m not entirely sure, and I don’t think the public does either.

I’m not so sure a change is needed here. More importantly, I’m not so sure the general public is familiar with and/or understands the current outlook categories and knows what the outlook categories mean. Changing what people don’t know by heart will likely come with a sense of confusion and questions about why there was a change.

Ultimately, this change should benefit the public, but I don’t think it will. Introducing new categories does not necessarily mean better understanding. As the tables above show, there is a lot that goes into placing a part of the country under a “slight,” “moderate,” or “high” risk. The public doesn’t understand the math behind each of these risks, but the subjective reasoning behind these risks isn’t common knowledge either. What does a “slight” risk really mean for my family? Honestly, meteorologists may disagree on what it takes for the Storm Prediction Center to issue a “slight” risk for severe weather in a given part of the country. If the lines are blurred or slightly blurred in the meteorological community, how will the public understand?

I don’t know of a meteorologist that knows the probability tables above like the back of his or her hand. I also don’t know of a meteorologist that had a complaint about the current SPC severe weather outlooks. There was no outcry to make a change from the meteorological community (at least not one that I knew of).

So why was there a change? The devil is in the details.

Over the years, there have been a lot of severe weather events that have occurred outside or barely inside of “slight” risk areas. A great example of this is the Evansville/Newburgh/Henderson F3 tornado of November 6, 2005. That area was in a 2% tornado risk area (not high enough alone to warrant a SPC “slight” risk) at 6:59pm on the evening of November 5, 2005. At 2am on November 6, 2005, the F3 tornado began near Henderson, Kentucky and continued through Evansville and Newburgh, Indiana. The tornado killed 25 and injured dozens. Outside of Tornado Warnings issued by NWS Paducah, there was really no suggestion from SPC that a deadly tornado would happen that night. This was a big “oops” moment from the Storm Prediction Center. If the severe hail and wind threat were non-existent, this 2% tornado area would have been in a “marginal” risk for severe storms given the new outlook categories coming this fall, and perhaps some would have paid attention to this tornado risk. Placing areas in a “marginal” severe risk may cause more to take the risk for severe weather seriously, but it may also lead to more “false alarms.” How many “marginal” SPC risks with no damage will it take before people ignore them and/or higher severe weather threats?

The names of the categories also bother me. A “marginal” risk downplays the severe weather risk when the potential is there – be it small – for something significant to happen. Also, how is an “enhanced” risk higher than a “slight” risk for severe storms? More importantly, what are the differences in the risk, and do the category names clearly suggest a difference in the severe weather threat? Did the SPC ask the opinion of social scientists to ensure their category naming convention would resonate with the public?

Discussing the severe weather risk categories with the people of Cincinnati on Fountain Square last week made me realize that people want information as simplified as possible when it comes to storms. Many I spoke with didn’t understand the need for SPC to change the categories. Many didn’t understand what “slight,” “moderate,” and “high” risks for severe weather really meant. Many don’t want to try and understand 5 different severe weather categories. Many just want to know how bad the weather is going to be on any given day or how it will impact their daily routine. These conversations reminded me that simple is better when it comes to discussing weather.

When I think of SPC expanding the severe weather threat categories to five, I think of the Homeland Security Advisory System, which was discontinued in 2011:


People never knew what the categories meant, and Secretary of Homeland Security Janet Napolitano said the scale provided “little practical information” when she phased out the scale in 2011. Let’s hope the SPC’s 5-category scale finds more success the DHS’ scale which was uninformative, nondescript, and unhelpful.

Posted in Reflections, Weather | 1 Comment

The Problem With Sunday’s Highland County Tornado

On Tuesday afternoon, the National Weather Service in Wilmington, Ohio confirmed an EF-0 tornado in central Highland County that lasted approximately 4 minutes (from 8:39 to 8:43pm) Sunday evening. According to the damage survey, the tornado produced wind speeds up to 85mph and created a swath of damage 2.5 miles long. The tornado length is the 9th highest on record for any EF-0 Tri-State tornado since 1950.

At 1:45pm ET on Sunday, the entire Tri-State was put under a Tornado Watch:


In addition to SPC’s slight to moderate risk for severe storms issued days in advance, this was an early indication that tornadoes would be possible Sunday afternoon and evening. Many – including me – weren’t buying into the need for a watch Sunday afternoon. Afternoon clouds only allowed spotty showers to form. Showers and storms finally started to move into the Tri-State and intensify slightly after some late day sunshine. There was one Severe Thunderstorm Warning issued for a part of the Tri-State Sunday night: A Severe Thunderstorm Warning for Fayette, Union, and Butler County issued at 6:26pm…


This warning did not verify. Showers and storms were spaced out for much of the evening and were not particularly strong. There was a report of a tree down on a house in Independence, Kentucky around 8:30pm, and that was the only report of damage for quite a while. Reports of damage came in late from Highland County; the first report of damage from Highland County came about 45 minutes after the damage had occurred:


This is not an uncommon report to have after a severe storm. It was a bit surprising to see the report given that no Tornado or Severe Thunderstorm Warning was issued for Highland County and radar data suggested there would be some areas of strong but sub-severe winds.

What exactly did radar data show? Here is the reflectivity scan (showing shower and storm intensity) from the NWS Wilmington, Ohio radar at 8:38pm Sunday night:


The tornado was confirmed southwest of Highland County, and it is clear that showers and storms in the area were intense. These showers and storms had good inflow, but there was no pronounced hook echo. What did the Doppler part of Doppler radar show?


Storm relative velocity data from NWS Wilmington’s radar showed strong winds moving toward from the radar (green) south of Hillsboro around 8:40pm. The red pixels on the southwestern flank of this storm showed winds on average moving away from the radar in Wilmington. These red and green colors are not close together or bright, suggesting little or no rotation. Some storms in the Tri-State had stronger rotation Sunday night, and they did not produce tornadoes. What made this storm a troublemaker?

The real problem here is the radar data. This graphic from NOAA shows why this storm likely didn’t receive a warning:


Where the tornado began, NWS Wilmington’s radar beam was scanning about 1,100 feet above radar level (the radar is about 100 feet off of the ground in Wilmington). Despite being one county away from the radar site, the beam was likely too high to see the tornadic circulation or any parent circulation. Even with a recent upgrade – called SAILS – to the radar, the upgrade does not allow the radar to scan closer to the ground. The upgrade allows low-level scans to come from the radar during times of active or severe weather, but the upgrade does not give meteorologists the ability to see all tornadoes, including if they are far from the radar site.

The lowest scan from nearly all of the NWS radars covering this country is 0.5° above the ground. Why? Honestly, it’s fear of people getting blasted with radiation from these radars. The sun gives off a lot more radiation every day, but people aren’t constantly lathering up on sunscreen every time they go outside.

For years, the FAA has operated a radar in Kenton County, continuously scanning at 0.1° above the ground. To my knowledge, I’ve seen no complaint about this radar emitting radiation even closer to the ground than the NWS’ radar. Why should we be so considered with an NWS radar scanning at 0.1° instead of 0.5°? This fear of radiation and the bureaucracy surrounding it may actually be putting lives at risk. Lower-level scans will likely improve lead times on Tornado Warnings and Severe Thunderstorm Warnings. Lower-level scans will allow us to track hazardous weather with more accuracy. Why would we not want this?

Despite all of the improvements made to radars over the years (from more frequent updates to higher resolution to even more radars), the lowest-level scan is not getting any lower; this needs to change. The benefits of lower-level scanning outweigh the consequences; an upgrade that involves lower-level scan angles will allow us see tornadoes like the one that hit Highland County Sunday night with ease.


Posted in Uncategorized | 1 Comment

Why Tornado Warnings Should Be Issued For Every Tornado

One of the tenets of meteorology is debate. Computer forecast models are consistently at odds with each other about the timing of intensity of weather systems. There are disagreements between meteorologists about the differences between what a mostly sunny, partly cloudy, and partly sunny day looks like. Some weather-related topics, like climate change, are politically charged and constantly challenged.

Of all of the debates I’ve heard, the one that surprises me the most involves when and how the National Weather Service should issue Tornado Warnings. A Tornado Warning is issued when spotters see a tornado, funnel cloud, or rotating wall cloud or when weather radar suggests (or in some cases, confirms) rotation in a thunderstorm is strong enough to produce a tornado. Based on the limitations of technology and the density of the spotter network, many Tornado Warnings do not verify. Radar is a tool designed to track rotation, but radar does not always match what a spotter in the field sees. Some spotter reports are unreliable or misleading, occasionally prompting warnings that did not need to be issued. Even with these considerations, however, the threat of a tornado should not be ignored for any reason. Whether weak or strong, all tornadoes are dangerous.

While there are certain situations and environments which will undoubtedly support and create tornadic thunderstorms, most tornadoes form in far less supportive environments. Most weak tornadoes last on the order of minutes, and larger, upper-level circulations in a tornadic thunderstorm usually don’t last much longer. Many of these weak tornadoes form from thunderstorms in a larger complex of storms. Meteorologists often call these MCSs or QLCSs (mesoscale convective systems or quasi-linear convective systems, respectively). Areas of rotation in a complex of thunderstorms can be hard to see due to the number of storms and given that most tornadoes in a QLCS are short-lived (on the order of minutes).

Consider the scenario we had on Halloween night of 2013. Here’s a radar snapshot late in the evening on October 31, 2013:


While it is very clear in this imagery that lines of showers and storms look strong and well-defined, using radar imagery some multiple radars is not very helpful for detecting rotation in thunderstorms. Using radial velocity data from a single radar site will be far more helpful for assessing how winds are moving relative to the radar. A snapshot of radial velocity data from the Terminal Doppler Weather Radar near the Dayton International Airport at 10:58pm on October 31, 2013 shows several areas where winds were moving towards and away from the radar in close proximity (circled):


There was adequate support for severe storms and tornadoes (especially weak ones) that night. While instability was not strong, the jet stream, upper-level flow, and upper-level support was. Knowing that that this entire area was in an area where severe storms are possible, which areas of rotation circled in the image require a Tornado Warning? Some couplets (zones of rotation) are stronger than others, but you’d have a lot of false alarms if you issued on every couplet.

One tornado confirmed that night in the Ohio Valley occurred near Vandalia, Ohio. Even with a radial velocity scan produced by the radar every minute, the rotation the vicinity of the tornado is suddenly strong then suddenly weak in less than 5 minutes:


Even with high-resolution radar data, it is difficult to warn this community that a tornado is coming. It takes time for the National Weather Service to issue a Tornado Warning. It takes time for the media to break into programming to explain why a Tornado Warning was issued, show which communities are affected, and track the storm. It takes time for people to react and take cover. In this case, by the time all of this happened, the tornado had already dissipated.

While the tornado confirmed near Vandalia, Ohio on the evening of October 31, 2013 did not kill anyone, it injured 8 people. Unfortunately, many Ohio Valley tornadoes have killed people.

Historically, most tornadoes in the Tri-State since 1950 have been weak, receiving an F0, F1, EF0, or EF1 rating. For the sake of simplicity, I’ll classify “Tri-State tornadoes” as tornadoes since 1950 where any part of the tornado path is in the Tri-State. I’ll also count injuries, deaths, and damage caused by the entire tornado in my calculations even if part or most of these totals occurred outside of the Tri-State; odds are these “boosted” totals will be from stronger, longer-track tornadoes. Most tornadoes that have occurred in the Tri-State, however, began and ended in the Tri-State, so I will allow for this approximation.

The graph below shows that stronger tornadoes in the Tri-State have occurred less often than weaker tornadoes:


This is no great surprise; stronger tornadoes almost always require strong shear, instability, lift, and moisture. But do Tri-State tornadoes with a higher rating kill more people? Historical records suggest “yes,” but to a point:


It is important to note that weak tornadoes (tornadoes with an F0, F1, EF0, or EF1 rating) have only killed one person in the Tri-State since 1950, while strong tornadoes (with an F2+ or EF2+ rating) account for roughly 99% of all Tri-State tornado deaths.

I won’t go into great detail about it here, but I believe the spikes in F2/EF2 and F4/EF4 fatalities are more about what, when, and where the tornadoes hit and less about the strength of the tornado.  The time of day, the time of year, population density in the path of the storm, and other factors likely contribute to the “spikes.” The F-scale and EF-scale are two different rating scales, and lumping and EF- and F-scale rated tornadoes into bins may also affect how the graph looks. The point I am highlighting is that stronger tornadoes tend to be killer tornadoes.

Injuries are more common than deaths with tornadoes, and – locally – more injuries have occurred with stronger tornadoes than with weaker ones:


There has only been one Tri-State tornado given an F5 or EF5 rating since 1950: the Boone County/Sayler Park tornado on April 3, 1974; this is likely the reason for a large drop in the injury count from F4/EF4 to F5/EF5 tornadoes.

So why issue Tornado Warnings for weaker tornadoes if they kill and injure fewer than F2/EF2+ rated tornadoes? If this were the case, fewer Tornado Warnings issued would lead to a lower false alarm rate, and fewer people would ignore Tornado Warnings. Why not just worry about the big tornadoes and ignore the small ones?

There are two big reasons. Here is the first:


The Mission of the National Weather Service is to protect “life and property.” While protecting lives is of utmost importance, the Mission Statement also includes the words “and property.” The warnings that come from the National Weather Service and the tracking and alerting that broadcast meteorologists do is all in an effort to protect you and what you own. Regardless of whether they work for the NWS, in the media, academia, or the the private sector, meteorologists – as a whole – are committed to the NWS’ mission.

Some of the strongest tornadoes that have ever occurred in the Tri-State caused thousands if not millions of dollars in damage:


Even weak tornadoes can cause hundreds of thousands of dollars in damage. An F1 tornado in Dearborn County in the early morning hours of April 9, 1999 caused an estimated $250,000 (in USD at the time) worth of damage. Should we – the weather community – inform viewers when there’s an imminent threat of a tornado, regardless of whether it will injure people, kill people, or cause damage? Absolutely. People deserve the right to know what is coming for their them. Should the National Weather Service not issue a Flash Flood Warning if it will only cause homes to be damaged but not kill anyone? Should a broadcast meteorologist only cover a winter storm if it has the potential to be life threatening? Should meteorologist in the private sector only create a product or service that prevents injuries but doesn’t work to prevent deaths? The answer to all of these question is a resounding “NO.”

The second – and just as important – point is that discerning weak from strong tornadoes isn’t easily done in real-time. Despite incredible improvements in technology in the last 50 years, there will always be limitations to what a radar and spotter network can give a meteorologist. Radar doesn’t scan at the ground, and there will always be cases where a radar sees strong circulation but there is no tornado. Spotters are important for being the “ground truth” in the field, but spotters are not everywhere. Spotters can report a tornado and/or describe it, and radar can – in some cases – confirm a damaging tornado in progress. This, unfortunately, is where radar and spotters reach their maximum effectiveness.

Spotters and radar can’t rate a tornado. The EF-scale is based off of damage. In order for a tornado to get an EF rating, a National Weather Service survey team must survey the damage. Books and binders worth of documentation are often brought to the scene damage site so that the National Weather Service can compare what they see to a specific set of guidelines and give the tornado a rating. These surveys can take hours or even days.

As time goes on, we will learn more about how tornadoes form, how they dissipate, their environments, how to track them, and how to detect them with more accuracy. We will not, however, gain the ability to rate tornadoes on the EF scale in real-time. In other words, trying to rate a tornado as it cuts through a community is not worth our time. If we can’t definitively predict the rating of a tornado in real-time, why should we attempt to gauge which tornadoes will kill or injure people and which ones won’t? This is a dangerous game with no winners.

Tornado Warnings were created to warn those in the path that a tornado is imminent. Whether a tornado is radar indicated or confirmed by a spotter in the field, the threat for a tornado is real when a Tornado Warning is in effect. Some tornadoes will cause damage; others will kill and injure people. A meteorologist’s job is to warn, prepare, and educate. Daring to guess which storms will play nice and which ones won’t is best left to those who create the weather instead of forecasting it.

Posted in Reflections, Weather | Leave a comment

A Personal Reflection Of The April 9, 1999 Tornadoes

It was the loudest thunderstorm I’ve ever heard in my life.

There was a cadence of thunder. Lightning resembled a strobe light. The lightning and thunder was so intense that you couldn’t sleep through it if you tried. It didn’t last a minute; it lasted 10 minutes. It wasn’t constant thunder and lightning; it was loud, bright, and constant. Based on the thunder and lightning alone, you knew something was wrong. And there was.

I woke up the next morning really not remembering what had happened hours ago. Sun was coming through the window, and the storms had moved out by 7am when I woke up. Sycamore Schools had been called off, and I remember hearing it on my alarm radio. Family members called asking if we were okay. There were tree branches down in my area, but there was nothing suspicious going on outside. I remember wondering who had moved our gas grill to the other side of the deck that morning; no human moved it.

It was clear once the TV was on that there was extensive damage on the other side of Blue Ash. It was likely a tornado based on the severity of the damage, but it was not confirmed at that point.

There 5 tornadoes in the Tri-State in the early morning hours of April 9, 1999. The map below shows 4 of them; an F1 tornado near Addyston is not shown:


A pair of thunderstorms were out to make trouble that night. One storm created two tornadoes in southeastern Indiana. Another caused damage in northeastern Hamilton County and southern Warren County. While the southern storm started strong, the northern storm would win out and cause the most damage that morning:


The first tornado of the night was an F3 in Ripley County, touching down near the Big Oaks Refuge and dissipating before it moved in Dearborn County. The storm relative velocity product showed strong inbound and outbound motion (in green/blue and red, respectively) in southern Ripley County just before 4am on April 9, 1999; the storm-relative velocity product is essentially the raw radar velocity product with the motion of the storm subtracted out.


While this tornado was significant and killed 3 people, a much larger, powerful tornado would develop less than one hour later from a separate thunderstorm.

The 5:12am radar scan that night from the National Weather Service in Wilmington showed the classic “hook echo” forming just west of I-71:


The radar velocity scan showed intense rotation near Blue Ash at that same time. Blue colors in the image below show strong winds moving towards the radar, and red colors show winds moving away from the radar; the tornado is very close to where these colors meet:


The storm-relative velocity scan at 5:12am below shows the rotation as well:


4 people were killed and 65 were injured as a result of the Blue Ash/Montgomery/Symmes Township tornado on April 9, 1999. More likely would have been killed or injured from this tornado had it not been for reports of a tornado and damage from trained weather spotters in Ripley and Dearborn County. This report was received by the National Weather Service at a critical, warning decision making time. The Tornado Warning issued for Hamilton County in the early morning hours of April 9, 1999 acknowledges a report of a tornado in southeastern Indiana minutes before Hamilton County was put under the warning.


These spotters saved lives that night.

There have only been 11 tornadoes in the Tri-State since 1950 to be classified as a violent tornado (given a rating of F4, F5, EF4, or EF5). The tornado that hit Blue Ash, Montgomery, and Symmes Township was one them. These communities had roughly 30 minutes of warning lead time to take cover, but this warning occurred on a night where the severe weather threat was not excessively high. Two Tornado Watch boxes were issued for the Tri-State that night, but there was no imminent threat of a tornado during the late local news. Most went to bed hours before the hours not expecting a tornado to crash into their house. The Internet was not used like it is today, and NOAA Weather Radios were not used as often. After seeing the damage firsthand, it is surprising that more weren’t killed or injured.

The event was also a game changer for how storms were covered by local TV stations. While tornado coverage was there, it revitalized the sense of urgency that storms bring. The loss of life that morning changed TV severe weather policies and how storms were tracked and covered.

With the tornadoes from April 9, 1999 included in the count, April is the most common month for tornadoes in the Tri-State:


April 9, 1999 reminds us that tornadoes can and do strike how and when they want. They don’t wait until the sun comes up, and they don’t discriminate. Nighttime tornadoes are dangerous, and they are among the deadliest types of tornadoes because they cause damage when people are most vulnerable. Lessons were learned that morning 15 years ago; my hope is that we are better prepared for the next round of storms.

Posted in Reflections, Weather | 1 Comment

A Review Of Meteorological Winter 2013-2014

Winter is far from over, but the core of the winter season – December, January, and February – was among the snowiest and coldest on record. In fact, meteorological winter 2013-2014 was the 2nd snowiest and the 18th coldest on record in Cincinnati.

To ensure that meteorologists compare apples with apples, meteorological winter is defined as December, January, and February. Astronomical winter’s start and end date varies each year and often ends and begins at a different time each year. Meteorological winter is always 3 months long, so it’s simple to compare seasons.

To measure where a season ranks compared to other years, we must know the average temperature of each day in that season. The average temperature of a day is the high and low temperature divided by two; the average temperature of a season is the average of all of the daily average temperatures in a season. When you crunch these numbers for the winter of 2013-2014, it ranks as the 18th coldest:


Meteorological winter of 2013-2014 ranks as the 2nd snowiest on record in Cincinnati; we were close to the number one spot of 1977-1978!


While those are the two most common ways to measure a winter’s might, there are other ways. Ranking as the 14th coldest, the average low temperature this winter in the Queen City was 4.4° below average, but it was nowhere near as cold as 1976-1977:


The number of nights where we dropped below 10° in meteorological winter was double the average but well short of the record set in 1976-1977:


Cincinnati dropped below 0° 7 days between December 1st and February 28th. This is over three times the average, but 10 days short of the record:


While the days were cold, records show that the number of days in December, February, and January where the high was below 32° was about average and not even close to matching the record:


One big record was set this winter: the most number of days (32) in meteorological winter with measurable snowfall. This beats the previous record set in 1977-1978 of 30 days:


One Tornado Warnings and seven Severe Thunderstorm Warnings were issued in February 2014. The Tornado Warning was the first issued in the Tri-State during February since National Weather Service Forecast Office in Wilmington records began in 1995. The tornado confirmed by the National Weather Service in Ripley County was the first February tornado in the Tri-State since February 15, 1967.

Even after the brutal cold of meteorological winter 2013-2014, nearly all records of snowfall and cold still belong to 1976-1977 or 1977-1978. Rounds of snow and ice are far from over in the Ohio Valley. Cincinnati averages 3.1″ of snowfall each March; some in the Tri-State may see more than that Sunday into Monday!

Posted in Weather | Leave a comment