Introduction
During the evening of Saturday July 22, 2017 and the early morning hours of Sunday July 23, 2017 thunderstorms trained roughly along the Ohio River over southern Ohio and northern Kentucky resulting in nearly 8" of rain in approximately a 12 hour period.
Flash flooding claimed one life and there was extensive property and infrastructure damage in the area mostly from Maysville, KY to the northwest into Bracken Co.
The Synoptic Setting
Satellite
During the afternoon of Saturday July 22, 2017 GOES-16 showed a very active convective environment across southern Indiana, southern Ohio and northern Kentucky.
The GOES 16 Mesoscale Domain Sector (MCD) 1-minute satellite loop below showed multiple outflow boundaries and multiple lines of small convective cells acting to help initiate convection in the region.
 |
| GOES 16 Mesoscale Domain Sector Saturday afternoon July 22, 2017 Courtesy: CIRA, Colorado State University. |
Surface
Surface analysis charts (below) showed a regional-scale convergence zone (solid black line) likely caused by three factors.
- After two days of thunderstorms along and south of the front there were multiple outflow boundary remnants in the region.
- Pre-frontal convergence was occurring south of the quasi-stationary front.
- The synoptic air flow and terrain were interacting to increase the convergence.
The front, drawn as a cold front on the 00z surface, chart quickly became quasi-stationary as several weak, small lows rippled northeastward along the boundary.
 |
Surface analysis 00z 23Jul2017. Note I used convergence (-divergence) to place boundaries and fronts - not traditional pressure, wind and dew point data. A well defined convergence zone is shown as a black solid line. See below for the animated gif version, every two hours from 00z to 12z . Plotted with Digital Atmosphere 2015.
|
Radar
An animated gif radar loop from KILN starting at 00z 23 July 2017 and continuing through 12z July 2017 (8 PM EDT 22nd - 7 AM EDT 23rd) shows thunderstorms training to the southeast and increasing in intensity and coverage as interaction with the surface convergence zone took place.
Winds
From the 925 hPa and 700 hPa 00z streamlines it is obvious that the waves of rain were being steered by the winds at or around 700 hPa.
In fact northwesterly flow dominated from above 900 hPa upwards past 450 hPa on the 00z KILN sounding steering the convective storms to the southeast.
 |
| Plotted and analyzed with Digital Atmosphere 2015 |
 |
| Plotted with RAOB |
500 hPa
 |
500 hPa analysis charts (height in geopotential meters, vorticity 10^-5 sec^-1) show a fast-moving shortwave trof moving out of Alberta, heading southeast and weakening across the northern Great Lakes. In response 500 hPa heights fell across southern Ohio and northern Kentucky. At KILN the heights fell from 5880 m at 00z to 5830 m at 12z 23 JUL 2017. At KBNA the corresponding heights were 5920m and 5890m. The 500 hPa temperature at KILN fell during the same period from -4°C to -8°C. Further investigation will likely reveal the cooling aloft aided in maintaining the heavy rain clusters. The 5880m geopotential height contour highlighted in red to emphasize the height falls. Charts: Courtesy U.S. Taxpayers available at: https://nomads.ncdc.noaa.gov/ncep/NCEP
|
Rainfall Totals
The highest radar estimated rainfall total (7.96" new algorithm, 6.0" old algorithm) was southwest of Ripley, OH and 5.4 miles northwest of Maysville, KY (as mapped by RadarScope) over rural northwest Mason Co. KY.
Radar totals compared favorably with amounts reported to NWS ILN by observers.
Even though surface observations were non-existent near where radar indicated the highest rainfall total, at the red "X", approximately 5.4 miles northwest of Maysville, a water rescue and flash flooding reported near Minerva (red circle) increases my confidence that the radar estimates were reliable. In addition a localized radar rainfall maximum of 6.80", is indicated 6.0 miles south-southwest of Maysville (yellow circle) near where NWS ILN observers reported 7.42" of rain. Courtesy NWS ILN.
Flooding was extensive and damage to property and infrastructure was extensive in this hilly area with deep, steep-sided creek valleys flowing into the Ohio River. One person was killed when his mobile home was washed into a tree and broke apart.
Convergence at the Surface
Convergence is what happens on an interstate highway as you approach an accident that is blocking part of the roadway. Divergence is what happens when you pass the accident. The same thing happens with flowing air for a variety of reasons.
When air converges at the surface, the air that piles up has only one way to go and that is up.
Convergence/Divergence from Change of Wind Speed
 |
| Convergence and divergence in the atmosphere or on the interstate. By: Steve Horstmeyer |
On approaching the wreck, the number of vehicles in a given distance increases - they converge. Once the accident is in your rear-view mirror the vehicles spread out again - they diverge. In the atmosphere is wind speeds decrease down stream convergence will take place unless there is some compensating motion.
Divergence and the flip side of the coin convergence is defined as the percent change in area per unit time. In three dimensions it is the percent change in volume per unit time.
Convergence along a Front/Boundary and Into a Cyclone
 |
| Drawn by: Steve Horstmeyer based on originals by Leung Wai-hung at http://www.hko.gov.hk/education/edu01met/wxphe/ele-condiv-e.htm |
Divergence Along a Ridge or Out of an Anticyclone
 |
| Drawn by: Steve Horstmeyer based on originals by Leung Wai-hung at http://www.hko.gov.hk/education/edu01met/wxphe/ele-condiv-e.htm |
Divergence/Convergence Couplets in 3D Air Motion
 |
| Drawn by: Steve Horstmeyer based on originals by Leung Wai-hung at http://www.hko.gov.hk/education/edu01met/wxphe/ele-condiv-e.htm |
By convention divergence is positive and convergence is negative. When there is convergence at the surface the only way the air can go is up.
Definition: Divergence/Convergence
 |
Arrows represent surface winds at the corners of an area, originally a square. As the square moves downwind the winds diverge and the square becomes a rectangle, larger in area than the original. Divergence has taken place. Later the rectangle is transported downwind and the winds now converge. The area of the rectangle decreases. Convergence has taken place. Note more can happen to the square, it can be rotated, lifted and contorted into new shapes. In addition we should really look at convergence and divergence in 3D and compute it for volumes of air. By: Steve Horstmeyer
Trajectories vs. Streamlines and Surface Convergence
Streamlines |
Streamlines show a snapshot of the winds, they are an instantaneous view of air flow at a single point in time. Connect the dots to form streamlines and the big picture of air flow at one point in time is visible. The wind at any point is the tangent to the streamline at that point,
Streamlines do not predict where an air parcel is going unless the weather systems are in a steady-state. That means the systems are stationary, the systems are not decaying or strengthening (i.e. the pressure gradient is not changing) and the systems are not filling or deepening (i.e the central pressure is not changing). This may seem confusing but if a cyclone, for example, grows larger and the central pressure does not change, the pressure gradient decreases.
If you traverse the map below from Illinois, Missouri and Arkansas to the northeast and east you can see how the horizontal distance between individual streamlines is decreasing. From Cincinnati eastward along the Ohio River heading eastern Kentucky and West Virginia the streamlines are noticeably closer.
I have added divergence values on the second map and on the third have shaded in the "channel" in which the air is flowing into and through the area that received flash flooding.
The "channel" is confined to the north by the front and to the south by a high pressure system.
 |
| Plotted and analyzed with Digital Atmosphere 2015 |
 |
| Plotted and analyzed with Digital Atmosphere 2015 |
 |
| Plotted and analyzed with Digital Atmosphere 2015 |
Trajectories
A trajectory is the path a parcel of air follows through time. The motion of a parcel of air is the combination of the winds of the weather system, the movement of the system and the change in strength of the system. If a system is in steady state trajectories are the same as streamlines. Trajectories are most often calculated in three dimensions.
The image below shows the trajectories of air originating in northeast Arkansas at 18z 22 July 2017, the afternoon before the flash flooding event in the vicinity of Maysville, KY.
Note how the trajectories in the NOAA ARL HYSPLIT model are concentrated along the Ohio River. The paths are concentrated there because the air is in effect "channeled" between the front to the north and the anticyclone to the south. Several of the trajectories are spurious in post-event analysis.
The video below the image uses the same data and goes into additional details.
Surface Maps
On the maps below notice that where you find high pressure the contour value is positive indicating divergence at the surface. Where there is low pressure like the front or the convergence line the value is negative indicating convergence. In addition,
areas of surface convergence are areas of rising air. The greater the convergence the faster the upward motion and the greater the chance thunderstorms will be enhanced.
Because of the three reasons mentioned in the video the surface convergence line (solid black) persisted through the evening of JUL 22 into Sunday morning JUL 23 and was a major player in the flash flooding event.
 |
| Plotted and analyzed with Digital Atmosphere 2015 |
Model Runs: NAM 4K 00z 23 JUL 2017 Run
Below are cross section diagrams roughly from Jackson, KY (southeast) to Wilmington, OH northwest. The background shading is wind speed and the contours are vertical velocity in microbars/sec. In pressure coordinates upward motion is negative because pressure decreases upwards.
Translucent shading shows selected areas of upward motion along and near the surface convergence zone highlighted only to draw attention to them. Typical of high resolution models trying to resolve features near the size of the grid the NAM 4K has trouble pinning down the location of the upward motion.
In addition by 06z it appears convective feedback is beginning to dominate the output because of unrealistic increases in both upward and downward vertical motion.
Even though the high resolution models resolved the vertical motion associated with the surface convergence zone it is unlikely that a forecaster, using that alone, would have predicted 8" of rain in 12 hours.
However using all the available information presented here and watching the system as it developed, flash flooding, given the steep terrain and expected training was a forecast that most experienced meteorologists would have made.
 |
| Cross Sections Plotted with RAOB using NAM 4K Bufkit data archived by Iowa State University. |
