On Tuesday morning around 10 a.m., an area of very intense rainfall set up over the Washington-Baltimore region, and then parked there for the rest of the afternoon. The precise location of the heaviest rain accumulation was not anticipated in advance, as is often the case in summertime, flash flood-producing storms — an unfortunate shortcoming of even our best forecasting efforts.

By the end of the day, over 10 inches of rain fell south of Baltimore in northern Anne Arundel County, which was the hardest hit area. In D.C., over two inches of rain fell. Dozens of water rescues were made throughout Tuesday’s event, and numerous incredible flash flood images were documented on social media.

Today we look at the final rainfall totals and piece together how Tuesday’s event took shape, so we can learn some meteorological lessons from this episode.

Impressive rainfall accumulation

Tuesday’s advisory map showed what no one likes to see – a series of flash flood watches (light green) and warnings (dark red) rushed to cover a rapidly evolving storm system. The Capital Weather Gang and the National Weather Service both forecast that the day would feature periods of moderate to heavy rain, but not to the tune of five to 10 inches.

The heavy rain was extremely localized. Figure 2 is a map of radar-derived rainfall accumulation. The white blob in Anne Arundel County near Glen Burnie is eight inch rainfall totals.  Actual gauge measurements were higher, as radar tends to underestimate heavy rain falling from shallow, “tropical-like” rain clouds, such as these.

6.30 inches of rain fell at Baltimore-Washington International Airport, which broke the record for the date, previously 4.91 inches set in 1955. It also moved Tuesday into second place for the wettest day overall at the location.  The largest amount of rain the airport has seen in a calendar day is 7.62 inches on August 23, 1933.

As if that’s not enough, 10.32 inches fell near Green Haven, Md. in Anne Arundel County. Pasadena, Md. was also hard-hit, with 7.55 inches. Dundalk, Md. recorded 8.75 inches, and Pumphrey, Md. saw 7.41.

A ways west in Virginia, Tyson’s Corner received 4.3 inches, and locations near Alexandria saw 4.28 inches. If you’re interested to see how much rain your location racked up, the incredible list of rainfall totals can be found in a statement from the National Weather Service.

The conditions at the surface

Let’s first look at the large-scale, or synoptic, pattern for this event.  The surface weather map is shown in figure 3.   A very complex weather system dominated the entire eastern U.S.: a type of mid-latitude cyclone that is more common during the cool season.   This was a classic Great Lakes low in the process of occluding – that is, its cold air mass was catching up to, and lifting, the storm’s “warm sector” away from the surface.  The occluded front is shown in magenta;  the warm front (red) bisects northern Virginia.

Note the rain bands streaming into the cyclone over the Mid-Atlantic.   A cool, overcast, rain-chilled air mass was in place across northern Virginia, Maryland and Pennsylvania, north of the warm front.   Atlantic moisture streaming northward into the cyclone was forced to rise over the cool “wedge,” condensing into thick clouds and heavy rain.

But the strongest rising motion was concentrated in a narrow, fast-flowing corridor of humid air called the “warm conveyor belt,” an air current typical of intense mid-latitude cyclones. This is shown in figure 4. The image depicts air speed and direction at 5,000 feet above the ground.  Trace the orange, red and magenta colors – the region of fastest flow – inward off the Atlantic shore.  You should make out a long, sinuous corridor arcing over the Chesapeake Bay, northward over Pennsylvania, then jetting across New York.   This is the warm, moist conveyor, also known as a low level jet.  The cyclone drew in tremendous amounts of oceanic moisture via the conveyor, but much of it dumped out over the D.C.-Baltimore region.

The deluge was focused by low-level convergence – literally, low-level “streamlines” of air that became squeezed together as they were drawn into the moist conveyor from both sides.  Additional uplift occurred as the conveyor was forced to ascend the wedge of cool air pushed up against the Appalachians.

Air entering the moist conveyor was sopping wet with water vapor. This is shown in figure 5, a map of “precipitable water” (the equivalent depth of rain water contained in the atmospheric column) shaded in green. The darkest green shades portray precipitable water in excess of two inches, with up to 2.5 inches feeding into the conveyor just offshore. The superimposed radar shows a “train” of storm cells lined up from south to north within this high-moisture plume.

Another way to visualize precipitable water is to examine the satellite water vapor image, which reveals vapor concentrations in the middle and upper atmosphere.   This is dramatically portrayed in figure 6.  White-gray regions show deep humidity, while blue shades indicate saturated (cloud-filled) air.   A deep, broad channel of high water vapor entering the storm’s eastern side is unmistakable.   The small pockets of green, yellow and red reveal the cluster of rain cells that erupted within the conveyor.  The location of the approaching occluded front is dramatically portrayed by the sharp contrast between moist air (white shades) and dry air (black) across West Virginia.

The meteorological situation aloft

The rain system got a boost from the upper-atmospheric flow, depicted in figure 7.   This shows the upper level flow in blue, distorted into an energetic trough over the Great Lakes.  Downstream of the trough, over the Mid-Atlantic, lies the jet’s active region where air was rising vigorously.  Look closely, and you will also see airflow (short arrows) fanning  apart over our region.   This difluence promotes additional ascent, as air rushes upward to fill the void. Heavy rain events are often associated with difluence aloft.

To summarize, several dynamical factors focused the uplift of air, and heavy rain, over a small region within the moisture channel (conveyor belt):  (1) upglide of air over the cool wedge;  and (2) difluent airflow aloft.

Small-ccale processes

Finally, we tunnel down to the mesoscale, where energy and moisture got honed down to just a handful of counties. Why just these counties?  I don’t have a good answer, and this is the part of meteorology that remains elusive – the density of our observation network is insufficient to resolve processes that initiate the smallest, most energetic storms.

However, figure 8 provides a hint.  Over eastern North Carolina lies a small, low-level cyclonic circulation, one that we saw spinning in the satellite cloud fields.  This is termed a “meso-low” and it developed along the warm front.   This low likely sent a surge of moisture inland, toward the northwest, as indicated by the blue arrows.  This in turn enhanced the low-level moisture convergence over our region.

Essentially, moist air kept piling over us and was forced to rise.

Another clue comes from the upper-air charts (not shown), which revealed a small pocket of energy – a “shortwave” – rotating over our region, around the primary trough.   This boosted uplift of air for a few hours as the wave passed through.

The heavy rain was convective in nature, generated by small clusters of multi-cell storms, repeatedly “training,” or moving over the same locations.  Note several orange blobs in figure 8, lined up along a corridor from south to north.  These are the heavy rain cells training along.  This is a very common process by which summertime flash floods are generated, but locating the “train track” a priori is very difficult, again because of the very small scales involved.

With flash floods, it’s all about timing. The coincidence of several factors promoting accumulation of water vapor and vigorous ascent, sustained for several hours.   These factors can combine in myriad ways, are difficult to detect, and still occasionally catch us all off guard.