Velocity Potential and Subseasonal Hurricane Forecasting

Note: This article was initially written on July 31, 2021 to discuss long-range forecasts for an August that eventually included Hurricanes Grace, Henri, and Ida, among other named storms. It was updated July 15, 2024 to be more generalized and include updated contact information.

Hurricane prediction is hard! Nonetheless, forecasters and researchers are always looking for new ways to push the boundaries and move forward. One way we have found increased success is "subseasonal" forecasting. This goes beyond what traditional numerical weather prediction models (GFS, ECMWF, UKMET etc.) can reliably simulate for specific weather events, instead focusing on whether the large-scale environment will be favorable for tropical cyclones to develop on time scales of ~2-4 weeks. Of course, a part of this is climatology... for example, we know that the Atlantic's "Main Development Region" (MDR) west of Africa usually gets more favorable in August/September (Beryl notwithstanding), with warmer water, less wind shear, and less Saharan dust/dry air.

That part is simple enough, but odds are you have seen/will see an image like the one below, courtesy of Dr. Mike Ventrice (@MJVentrice on Twitter). This particular plot came at a time when forecasters were concerned that the 2021 Atlantic Hurricane Season was about to pick up its pace. Indeed, 7 tropical cyclones formed in August 2021, including Hurricanes Grace, Henri, and Ida, and the tropical depression that eventually became Hurricane Larry that September. We are looking at a variable called "Velocity Potential", sometimes labeled as "VP200" as it is often calculated at a pressure level of 200 mb, well above the surface. At first glance, it may be confusing how the arrows and colors relate, and exactly what this term is describing. But as we'll discuss, it's meant to be a really simple way to assess past, current, and future environments where tropical cyclones may try to form!

The arrows on this plot are a good place to set the stage. Much like how we can split vectors up into "x" and "y" parts in a math/physics class, or wind into "east-west" and "north-south" parts, we can also split the wind up into the part that's rotating, and the part that isn't rotating. The non-rotating part, also called the "irrotational" wind, is what the Velocity Potential describes! It boils down to this straightforward question: "Is air converging or diverging in the upper atmosphere?" And what happens up there often implies quite a bit about what happens below! Let's break that down with a crude animation I made for an undergraduate class I taught at Florida State in 2021, shown below.

This is based on the idea that the 200 mb level, where Velocity Potential is usually calculated, is near the tropopause, a "cap" separating the troposphere where our weather mostly takes place from the much more stable stratosphere above. Air cannot punch through the surface, and except for particularly strong convection, it doesn't punch through the tropopause either. Therefore, divergent upper-level air implies rising motion below, and convergent upper-level air implies sinking motion below! The former is associated with widespread convection (more favorable for tropical cyclones), while the latter inhibits convection (less favorable for tropical cyclones). With that in mind, let's bring it back to the original VP200 plot, and introduce some annotations to help break it down based on what was happening that day in 2021.

What you'll always see in these plots is the anomaly of VP200 (how much it differs from the average value), but the idea is the same. As mentioned in the annotation on the lower right side of the plot, these cells of diverging (rising) and converging (sinking) air typically move eastward over time. There are two main types of disturbances that cause this, which are fundamental to subseasonal hurricane forecasting: The Madden-Julian Oscillation (MJO) and convectively-coupled Kelvin Waves (CCKWs), which each deserve separate articles (and entire PhD dissertations). As we moved into August 2021, that cell of rising motion moved into the Atlantic, convection became more widespread and disturbances were more easily primed to become tropical cyclones. The video below mashes this whole discussion together, using a different example from the 2022 season with the eastward movement of the cells.

A couple of caveats to note: First, the centers of these rising cells aren't necessarily the most favorable places for tropical cyclones. This is because particularly strong divergence aloft often implies strong vertical wind shear, which can disrupt disturbances from focusing thunderstorms in a particular cluster. We often look for the most active periods to take place after a cell moves through - an environment still favorable for rising motion, somewhat weaker shear, and an often-moister environment for disturbances to traverse through after convection has already been widespread for several days. Second, this subseasonal forecasting approach doesn't allow us to point to specific storm formations, tracks, intensities, etc. In other words, I cannot reasonably tell you on July 15, 2024 that a storm will form in the eastern Caribbean on August 7. But with this Velocity Potential tool, we have gained more skill in saying things like "the next 2-3 weeks should be fairly quiet" or "mid-September is likely to be busier than late August".

If you have any questions, comments, follow-up points, suggestions for future articles, and anything in between, don't hesitate to reach out! Find me on Twitter @JakeCarstens, or drop me an email at jacob.carstens@und.edu.