Excessive Rainfall over Vermont

Among the more remarkable events to occur in New England during the last few weeks are the incredible amounts of rainfall centered over Vermont between July 9 and 10.

This event had a clear signature that was evident several days in advance and was overall handled well by both models and forecasters in the areas that ultimately experienced the highest precipitation totals and remarkable flooding, which resulted from all that rain falling on soils already saturated from weeks of well-above average rainfall across nearly the entirety of the state.

The very wet antecedent conditions combined with excessive rainfall produced one of the more remarkable flood events in recent memory in Vermont, equalling and exceeding in some locations the flooding due to the remnants of Hurricane Irene in 2011.

Let’s set the stage.

This extraordinary rainfall event in Vermont occurred after the state had already received roughly 150-250% of their average rainfall over the previous two weeks. Vermont averages roughly four inches of rainfall per month in the summer months, so all the dark blue to yellow dots on the map below correspond to locations which had already received more than their average precipitation for the whole month of July within the two weeks prior to the event. Much of this rainfall could be attributed to a persistent trough to the west and a blocking high pressure to the east, allowing warm, moist air to be drawn northward from the Gulf of Mexico and the tropical Atlantic Ocean into the area and be tapped by disturbances passing through the region.

Map of precipitation totals in the two weeks prior to the event. Locations with dark blue dots had already received much of their entire July rainfall in the preceding two weeks (>=200% of average). You can also see that New Hampshire also had a rather wet spell, though it was spared the worst of the rainfall between July 9 and 10.

Now, let’s get to the actual event itself.

The favorable atmospheric setup for this particular rainfall event was already being noted nearly a week before with warm, humid air being drawn into into the region with sufficient upper-level dynamics to wring much of this moisture out, while the low pressure system only slowly moved eastward. The atmospheric setup of the previous week became more concerning in the lead-up to the event. The amount of moisture involved and the minimal movement of the system would act to keep a relatively narrow corridor under near continuous moderate-to-heavy rainfall for an extended period of time before the best forcing for ascent would finally move eastward out of the affected area.

Let’s look a little more deeply into each of those ingredients.

The warm, humid air being drawn into the low pressure system was exceptionally moist as compared to climatological averages for this region. Atmospheric scientists and meteorologists typically measure the amount of moisture present in the atmosphere above any particular location as “precipitable water,” which is the depth of water that would result if all the water vapor above a location were condensed in one fell swoop and fell as rain. The greater the number, the more moisture is present, and the greater the probability for heavy precipitation to occur, provided there are methods by which that air mass can be lifted, it’s water vapor condensed, and eventually precipitated.

The below figure is from a weather balloon launched at Upton, NY at 00Z on July 10 (8:00 p.m. on July 9 EDT) showing the extremely moist air streaming into the Vermont area during the event. The two lines on the figure show the temperature of the air (right line) and the dewpoint (left line). The close proximity of those lines from the surface to nearly the top of the troposphere (where the lines both tend to jag to the right) show that the air moving into Vermont during the event was nearly saturated between the surface and the tropopause, and in this particular case, if all the moisture was condensed, would have contained a 48.57mm (1.91 inches) of water.

00Z Sounding from Upton, NY showing the very moist atmosphere streaming into Vermont at the time of the event, with precipitable water values of about 48.57 mm (red box), or about 1.91 inches of water.

Sources of Lift

So, we have a deep layer of very warm, moist air, and the next most important ingredient is how that moisture can be lifted, condensed, and precipitated as rain. During the summer, the jet stream is usually weaker, and does not play as large a role in influencing the strengthening of storms. But during this event, as shown in the below map, there were two healthy jet-streaks that were involved in uplifting the air. The first is located across the central Mid-Atlantic, oriented predominantly west-to-east, whereas the other, stronger one begins over upstate New York and heads northeast over Quebec. This dual jet-streak setup promotes upper-level divergence over the left-front exit region of the first jet streak, and the right-front entrance region of the second, which when overlaid over each other, encompass most of New England, due to how the atmosphere responds to accelerations in atmospheric flows. The upper divergence noted at upper-levels of the atmosphere promotes lower-level convergence and upward motion throughout the atmospheric column.

Lower-Level Dynamics

For most of the event, the surface map did not look all that exceptional meteorologically. As shown below, a surface low pressure system very slowly meanders along the southern New England coastline for much of the event. Even though the surface front is stationary, the upper-level dynamics are forcing the low-level flow from the the warm to the cool side of the stationary front, with the shallow frontal slope promoting uplift well to its north in a surface trough that extended north of the low’s center (dashed tan line emanating from the low).

This surface trough prompted convergence and uplift to occur along a preferred axis, oriented roughly south to north, and also roughly aligned, and in the middle of the western and eastern borders of Vermont. The other item to note here is the extremely slow motion of the system, moving only from southern central New Jersey to the southern coast of Massachusetts over a 24-hour period. Not only this, but flow around the low and into the surface trough maintains roughly over Vermont for much of this period, promoting the training of precipitation over the same locations for much of the period. The slow-moving nature of the system was due to a massive blocking high pressure system off the Atlantic Coast of Canada that ‘blocked’ the progress of the low pressure system out of the area.

High-resolution surface maps between 00Z July 10 to 00Z on July 11 showing the low pressure center and the frontal systems associated with the low throughout the course of the event. Note the extremely slow progress of the low throughout this period, acting to keep the plume of moisture and heavy precipitation directed into Vermont.

The low pressure system also provided some low-level convergence which also promoted upward vertical motion, and also provided a focus for inflow of warm, moist air over Vermont. That combined with a sharpening negatively-tilted mid-level shortwave moving in from the west added extra oomph to the upward motion in the mid-levels of the atmosphere. Finally, the consistent SSE flow into Vermont helped to direct the flow at somewhat of an angle to the Green Mountains. Because the low-level flow had sufficient instability the extra nudge upwards from the mountains led to some extra precipitation on the eastern, or windward side of the mountains. Roughly, you can think of each of these processes as promoting uplift, focusing that uplift over Vermont, and wringing out the moisture in a narrow corridor, leading to the heavy rainfall measured during the event.

And finally, not only were there multiple sources of lift available to wring as much water out of the deep layer of moist air flowing over Vermont, but the rainfall process itself was very efficient, adding to precipitation totals. When warm clouds (clouds which are above freezing) are sufficiently deep within a warm environment, raindrops develop by larger drops colliding and combining with smaller cloud drops (the collision-coalesence model) versus the Bergeron process, which involves ice crystals growing at the expense of liquid droplets in ‘cold’ clouds which are below freezing.

And what meteorologists call ‘efficient’ in precipitation terms is that they effectively convert precipitable water into rainfall. There are three things to note about this: These processes typically occur in the tropics where the freezing line is higher as the air is usually quite warm. Secondly, with the freezing line being so high, there was less ice in much of the clouds, and so there was less charge separation, and as a result there was not much lightning occurring despite the substantial rain rates. And finally, because warm rain processes were dominant, the droplet size distribution of the rain that was falling caused precipitation estimates based on the radar returns to actually underestimate the amount of precipitation that fell, and made it appear (on radar) that it was not raining as hard as it was. The inconsistency of the relationship between radar returns and the actual amount of rain that falls (the Z-R relationship) is one of the reasons that actually measuring rainfall in precipitation cans is so important!

The Twitter graphic showing the corridor of heavy precipitation that fell over the two day period of July 9 to 10 within the National Weather Service’s Burlington office area of responsibility. Not the north-south axis of heavy rainfall (>7.00 inches) and how it located predominantly along and to the east of the Green Mountains, indicating the role of terrain in uplifting low-level moisture.

All of the above ingredients combined to lead to excessive rainfall over nearly the whole state of Vermont and produced the totals seen in the above map. You can view an animation of the radar images taken every hour throughout the event from 18Z on July 9th through 06Z on July 11th (2pm July 9th EDT→ 1am on July 11th EDT), showing the incredible stream of precipitation directed over Vermont for much of the period.

Let’s look into how to place the amount of rainfall experienced in Vermont during this extraordinary event into context, and in this analysis we’re going to use Montpelier, VT as a test case. Montpelier has averaged about 4.5 inches of precipitation during the month of July between 1990 and 2020. It received 5.28’’ of precipitation on July 10, which was a record for any date in July, and among the greatest single-day precipitation totals ever measured at the station in its history.

According to the Atlas 14 point precipitation frequency estimates, which estimates the frequency at which certain amounts of precipitation could be expected, this amount of precipitation is expected to occur at this station only once every 100 years within a 24 hour period, making it quite an exceptional event!

As a side note, that website predicts precipitation recurrence intervals for precipitation at many different locations within the United States showing the expected benchmark amounts of precipitation anywhere between 5 minutes (e.g. 0.30 in/5 min about 1 time a year) and 60 days (21.3 in/60 days about 1 time every 1000 years) and that which would be expected each year versus once every 1000 years. It is important to note three items: 1) Since all these values are statistically determined based on past rainfall amounts determined at the station, there is a substantial amount of uncertainty in the amounts of precipitation that would be expected at the longer return intervals, as these extreme events may not have been sampled during the station’s established period of record. 2) If a station has already experienced a ‘1-in-100 year event,’ it does not mean that it will take 100 more years to experience the next one, only that that particular event has a 1% probability of occurring in any particular year. 3) A ‘1-in-100 year’ rainfall event is not the same thing as a ‘1-in-100 year’ flood event because of the importance of ground conditions before the rainfall event, and the fact that streamflow is an integrative property of all points upstream of a location, not only a property of the rainfall experienced at one particular location.

Finally, let’s look into the flooding situation, and the amount and types of flooding information that can be quickly found online in the event of heavy rainfall and flooding in your area.

There are several locations to obtain streamflow information online, but here we will be focusing on two locations: 1) real-time streamflow information from the United States Geological Survey (USGS), and 2) predicted stream gauge heights from regional River Forecast Centers. The first source links to all the streamflow gauges which are currently reporting information, and allows you zoom in by state, and finally by location, either by clicking on the different color dots on the map, or clicking on ‘statewide streamflow real-time table’ link on the right, which can be organized by river basin or by county. Once you click on a station, it will show the graph for the last seven days, 30 days, or year of either the height of the river flow past the station gage, or the total volume of water passing by the station, or discharge, which both can be useful in flooding situations to see if the water is still rising, if it is, how quickly.

The second site consists of fewer stations, but provides stream height predictions three days out into the future. The markers on the map are colored according to the type (if any) of flooding expected to occur within that three-day span. Once you click on the station, the river height over the last 3 days will be loaded along with the expected river levels over the next three days. These forecasts can be quite useful for larger rivers which tend to have delayed crests, where their peak flow is delayed somewhat behind when the peak precipitation rates occur. At even fewer stations in a few locations, there are ‘inundation maps’ which allow you to determine which areas might flood if the river reaches a given depth.

For example, in the case of Montpelier, the Winooski River crested at height of 21.35 feet on July 11, the second highest crest at that particular location on record. If we go to the inundation map for this station and select the nearest river depth, in this case 21.8 feet, you can see which areas of Montpelier were likely underwater during the worst of the flooding, and how deep the water was at any particular point along the river when it crested.

Karl Philippoff, Weather Observer – Research and IT