GFS model rainfall forecast for Harvey. (

The expected extreme rainfall across the Texas and western Louisiana this week is hard to imagine, with some models predicting locally more than 40 inches and widespread amounts in the 15- to 30-inch range.

The combination of the Gulf of Mexico’s rich supply of moisture, Harvey’s ability to draw in this moisture and the lack of atmospheric steering currents make such amounts possible.

The figure below illustrates the key processes involved in pulling off this sodden hat trick.

(Tom Rabenhorst, UMBC)

It all starts with saturated air with very high dew points, entering the storm in the lower layers of the atmosphere. The green shades in the diagram indicate dew point values in the low to mid-70s, which are common over the very warm gulf water.

The warm water and rapid flow of air over the water (i.e. the hurricane winds) surface contribute to enormous “fluxes” of water, moving from ocean to air. Then, as shown by the red arrows, all of this vapor-laden air converges (squeezes together) into the region of low pressure that marks the center of the storm. A tropical storm gathers or sequesters water vapor from an enormous area, much larger than the actual rainy footprint of the storm itself.

Another key factor is that these late-summer gulf storms, on average, tend to move very slowly or, in the case of Harvey, not at all. This is because they are disconnected from the river of fast-flowing wind in the middle and upper troposphere called the jet stream.

In the diagram, the westerly jet stream (purple arrow) is shown displaced north of the system, coursing along its northern periphery. The jet may, in fact, be displaced as far north as Canada. The storm, then, receives no push; if it is not embedded in a fast flow, it will meander, or simply sit and spin, for days at a time. This means all that rain simply piles up over the same locations. Note that the total distance covered by our prototypical storm (dashed black arrow), spanning Day 1 to Day 5, covers little ground.

Yet another factor has to do with the alignment of rain bands parallel to the in-flowing stream of low-level air. Look at the northernmost rain band in the diagram. It is oriented east-west and along the red arrow, which represents in-flowing moist air. The rain band is actually composed of a series of small, convective cells, perhaps six to 10. These discrete cells each deliver torrential rain at a rate of two to three inches per hour. It may only take 10 to 15 minutes for each band to pass over, but if several cells pass over the same location in succession (location “A”) — all steered in the same direction — then the rain total in that one spot adds up very rapidly.

This process is called “cell training,” and the cumulative effect of numerous cells is shown in the inset figure (upper right). The so-called “rain train” really turns on a prodigious rate of freshwater delivery.

The latest expected rain totals for Harvey, for a seven-day period, are shown in the figure below. This is the best guidance offered by the National Oceanic and Atmospheric Administration’s Weather Prediction Center (WPC) based on a blend of computer guidance and forecasting “rules of thumb.” The totals along the southeast Texas coast are so high (20 inches or more, in purple) that the color scale chosen must use nonlinear divisions to convey the rain amounts.

National Weather Service rainfall forecast through Thursday. (

It’s my belief that the greatest contributor to the extreme rain totals will come from the storm’s very, very slow motion. If this was a progressive system, say, moving at 20-25 mph, then these totals would be halved.

Part of the storm system may meander back over the very warm gulf, which would help sustain the system at tropical storm force (or a wind close to this threshold), days after landfall. I also wonder about a phenomenon that could help sustain Harvey while over land. It’s termed the “brown ocean effect,” and it hypothesizes the following: When soils become saturated and there is standing water over a large inland area (from the flooding rains), a storm can be sustained over land, because of the large release of latent heat from the ground surface (in the form of evaporating water). The land, in effect, mimics the energy supply of the ocean.

Hurricane Harvey hit the Texas coast on Aug. 25 with extreme rain and 130 mph winds, and residents are bracing for more bad weather. (Victoria Walker,Whitney Leaming,Kurt Kuykendall/The Washington Post)

It’s important to understand that many people assume the storm surge and high winds lead to the most deaths from tropical weather systems. But over the United States, more people typically die from freshwater flooding — that is, the flooding induced by heavy rains.

In a hurricane, the intensity and coverage of rain generally increases inward toward the storm’s center. Historically, though, there is a poor correlation between overall storm intensity (as measured by the maximum sustained wind) and total water volume delivered by the storm. The Saffir-Simpson intensity scale is a wind intensity scale, e.g. Category 1 through 5, and this says absolutely nothing about how much rain will fall.

Some of the worst flooding has come from tropical storms, not hurricanes, and the post-tropical remnants of hurricanes. A recent example is the extreme rainfall unleashed by Tropical Storm Allison during 2001 along the Gulf Coast, with 41 inches falling over coastal Texas. Tropical Storm Claudette delivered 42 inches of rain in 24 hours to a small region in Texas in 1979. Thus, there is precedent for tropical rainfall, in extremis, over Texas.