What do Thermals Look Like?

Wayne Angevine

Copyright (c) 1995 Wayne M. Angevine
May be freely redistributed on Internet as llong as this message is included.

This is an article that I posted to this list and rec.models.rc last June. I also highly recommend the article by Roland Stull in the last proceedings of the Madison Soaring Symposia. See the classified ad in RCSD for how to order that volume.

Editor's note (5 Nov 2005): a newer version of this article is available at Wayne's web site


Model sailplane and free flight fliers are interested in the structure of thermals, which provide the energy for their flying. Here is my attempt to describe thermals. I'm an atmospheric physicist working in the boundary layer. This is not a scientific article, but my views based on extensive reading and observations.

The Boundary Layer

The short answer to the question is that thermals are columns of rising air. A longer answer requires what may seem like a digression into boundary layer physics. The boundary layer is the layer of air near the earth's surface that is affected by the surface on scales of an hour or so. The sort of boundary
layers we're interested in are convective boundary layers, which occur in the daytime over land in weak to moderate wind conditions. There are other sorts, but they don't produce thermals as such. I'll also assume relatively flat and uniform terrain, and at most fair-weather cumulus clouds.

Boundary layer physics is a subfield of atmospheric physics or meteorology, but the scales (and therefore the forces) of interest are different. It is easy to become confused if one tries to apply basic large-scale or storm-scale meteorological concepts to the boundary layer.

A convective boundary layer is a few hundred meters to 3 km thick, depending on the amount of incoming solar energy, the amount of moisture in the ground, the larger-scale weather (high or low pressure), the wind speed, and other factors. Call the boundary layer height zi. The bottom of the boundary layer is a *surface layer* about 0.1*zi thick, say 100-200 m. The surface layer is heated by contact with the surface. The top of the boundary layer is a temperature inversion (hence zi, inversion height).

So to first order, thermals are columns of warm and therefore buoyant air that rise from the surface layer to the inversion. The spacing between thermals is about 1.5*zi, say 1-2 km. The thermals themselves are somewhat less than half that, say 500-1000 m in diameter. Most thermals span the boundary layer vertically. There is, of course, a distribution of sizes. Between thermals are broad areas of sink. The sink is weaker than the lift because it covers a larger area. The opposite is true at the top of the boundary layer, but we rarely fly that high.

There are, as always, complications. Sometimes we fly in the surface layer and sometimes in the lower part of the boundary layer. Rising air in the surface layer (the lowest 100-200 m) is in the form of small plumes, themselves a few tens of meters in diameter. These plumes converge near the top of the surface layer to form thermals. The surface layer to boundary layer transition is not sharp, so we often find ourselves flying in either well-organized thermals or disorganized plumes, or some of both.


Thermals evolve over time, are influenced by terrain, and are shaped by and move with the wind. Boundary layer thermals form and dissipate with time scales of 10-30 minutes, surface layer plumes faster. This can lead to the apparent phenomenon of "bubbles" or detached thermals or plumes. Plumes and thermals respond to irregularities in the surface (different amounts of vegetation, houses, and so on) by forming more often in some places than others. Dark ground (if it's not wet!) and sheet-metal roofs are well- known thermal concentrators. If the wind is light, thermals may stay attached to the hot spot. If not, thermals may form repeatedly over the hot spot and drift downwind. Thermals drift with the average wind over their height, so they may travel at a higher speed and in a somewhat different direction than the surface wind. Thermals also tilt if the wind is stronger at higher altitude, th usual case.

Thermals are not uniform, nor do they have sharp edges. The edges interact with the surrounding air, so thermals have a warm, usually fairly smooth core surrounded by turbulent edges. The air around the edges may be in the form of blobs and may be either rising or sinking. This leads to the common idea that thermals are toroidal (donut-shaped). It's probably more accurate to think of thermals as vertical cylinders. Roland Stull (see reference at end) writes, "...the best model might be the 'wurst' model...",
that is, that thermals look like vertical sausages. Air detrained from the thermal edges is cooled, and cannot be recirculated into the thermal except at the ground. Vortex rings of the size of thermals are not observed. Stull also writes, "Real thermals are not perfect columns of rising air, but twist and meander horizontally and bifurcate and merge as they rise."

The strength of thermals is controlled by the amount of sunlight and the surface conditions. If the surface is wet or moisture is being emitted by healthy plants, a larger fraction of the incoming heat from the sun will be used to evaporate water than to heat the air. Water vapor does contribute to buoyancy, but less than heat does. These factors probably account for most of the difference between soaring conditions in the western and eastern U.S.

Variations on the theme

So far I've described the situation in the middle of a day with light wind and high pressure. I wish all contest days were like that! If the wind is stronger, turbulence driven by wind shear (the difference between the winds at one height and another) may interfere with the formation of thermals and the lift will be light and spotty. If the barometric pressure is low, there will likely not be an inversion to define the boundary layer top. This will tend to produce larger thermals that are farther apart, at least until the rain starts!

Do thermals rotate?

They do, but not predictably. Even dust devils don't have a preferred direction of rotation (see Stull, p.449). Thermals are too small and too short-lived to be affected by the earth's rotation (Coriolis force) or by the equator/pole thermal gradient. Their rotation is determined by local terrain. Rotational velocity in the core of a typical thermal is small compared to the vertical velocity.


Those who are interested in following up the topic further can consult the following references. An Introduction to Boundary Layer Meteorology
Roland Stull (Kluwer)

should be in any good University library. The chapter on convective boundary layers is quite readable. A recent paper on imaging of the boundary layer is

Calculations of Area-Averaged Vertical Profiles of the Horizontal Wind Velocity from Volume-Imaging Lidar Data, in the Journal of Geophysical Research, vol. 97, pp.18,395-18,407, 1992.
Schols and Eloranta