Mammatus clouds: what they are and why they form

Mammatus clouds are rounded, pouch-like lobes that hang from the underside of a cloud, most often from a thunderstorm anvil. They are a supplementary feature — a shape a cloud can take, not a cloud type of their own — and there is still no agreed explanation of how they form.

The pillar guide on how to read weather radar covers what the radar measures. This page is about a feature the radar almost never resolves, and about why the claim most often made for mammatus — that a tornado is coming — is not what the National Weather Service says.

The two definitions

The World Meteorological Organization’s International Cloud Atlas uses the Latin form. It defines mamma as “hanging protuberances, like udders, on the under surface of a cloud,” and classifies it as a supplementary feature rather than a genus or species. The Atlas records mamma occurring with cirrus, cirrocumulus, altocumulus, altostratus, stratocumulus and cumulonimbus.

The National Weather Service glossary uses the English plural and adds the operational caveat: mammatus clouds are “rounded, smooth, sack-like protrusions hanging from the underside of a cloud (usually a thunderstorm anvil),” and they “often accompany severe thunderstorms, but do not produce severe weather; they may accompany non-severe storms as well.”

Both describe a surface, not an object. Like the anvil and the overshooting top in the guide to cumulonimbus clouds, mammatus is a property of a parent cloud.

Where they sit relative to the storm

Schultz and colleagues, in their 2006 review in the Journal of the Atmospheric Sciences, note that cumulonimbus anvil mammatus have been observed on both the upshear and downshear sides of the outflow anvil. The anvil is not the storm’s inflow base, and the NWS glossary describes anvils spreading hundreds of miles downwind and sometimes upwind. A sky full of mammatus can be a long way from anything the storm is doing at the ground.

It is also a high-altitude feature. The same review reports anvil heights in the mammatus literature from 1.4 km to 8 km, and documents one radar case with the anvil at 5.5 to 7.0 km above ground and the lobes below a cloud base of 5.0 to 5.5 km. They have also been photographed on cirrus, on aircraft contrails and on volcanic ash clouds, where no thunderstorm is involved at all.

Size and lifetime

The measured numbers, from the 2006 review’s summary of the observational literature, which rests on a handful of opportunistic studies.

Ten proposed mechanisms, none settled

The 2006 review lists ten hypothesized formation mechanisms: anvil subsidence, subcloud evaporation and sublimation, melting, hydrometeor fallout, cloud-base detrainment instability, radiative effects, gravity waves, Kelvin–Helmholtz instability, Rayleigh–Taylor instability, and Rayleigh–Bénard-like convection.

The most commonly cited is subcloud evaporation and sublimation: hydrometeors fall from the cloud base into unsaturated air and cool it as they evaporate or sublime, driving saturated descent in lobes.

The review is blunt about the evidence: “Theories describing mammatus remain untested, as adequate measurements for validation do not exist because of the small distance scales and short time scales of mammatus.” Some mechanisms are mutually exclusive — gravity waves and Kelvin–Helmholtz instability both need stable stratification, while several observations point to unstable layers below cloud base — and the authors conclude that the observed variety of mammatus may mean more than one mechanism is at work.

The tornado claim

The persistent version is that mammatus are a warning sign, or that a tornado follows them. The NWS glossary settles it in one clause: mammatus often accompany severe thunderstorms, but do not produce severe weather, and may accompany non-severe storms as well.

The 2006 review is less generous still. It explains the thin research literature this way: “Because they are not directly related to significant weather events on the ground and they do not apparently hold insights into forecasting severe convective storms, mammatus generally have been viewed as no more than a curiosity in the atmosphere.”

The geometry says the same thing. Mammatus mark the underside of an anvil kilometers up; a tornado comes from the base of the updraft, near the ground and well away horizontally — the region covered in the wall cloud guide. A display tells you a storm with a well-developed anvil is or was nearby, which is why they fill so many photographs from severe weather days. The NWS Weather Spotter’s Field Guide (June 2011) does not list mammatus among the things spotters are asked to report.

What radar shows, and what it does not

Radar does not detect cloud; it detects precipitation-sized particles. That argument, and the beam-height argument under it, is set out in the wall cloud guide and applies here inverted. A wall cloud is invisible partly because it hangs too low to be sampled at range. Mammatus hang from a cloud base often 5 km or more above ground, so the lowest tilts pass far beneath them.

The arithmetic: under standard refraction the center of the 0.5° beam is about 1.5 km above the radar at 100 km range. Putting the beam center on a 6 km anvil base at that range means subtracting the roughly 0.6 km the beam gains from earth curvature and refraction and raising the remaining 5.4 km over 100 km, an elevation angle near 3 degrees — a tilt the WSR-88D scans. Resolution is the harder limit: a beam just under 1 degree wide spans about 1.6 km at 100 km, the width of one average lobe and three times the 0.5 km it hangs down.

There is also very little signal. In one case the 2006 review documents in detail, NSSL’s Cimarron dual-polarization radar was pointed vertically as a decaying mesoscale convective complex crossed central Oklahoma from 1128 to 1134 UTC on 4 August 1992. The perturbations beneath the anvil showed descent in echo above 7 dBZ and ascent in echo below 7 dBZ, the descent ranging from 1 to 4.5 m per second. For scale, the NWS glossary defines the echo top as the height of the highest tilt where reflectivity above 18 dBZ can be detected. A mammatus field lives well below that.

Where mammatus have been resolved, it has been by research radars scanning in the vertical. In a second case in the same review, a University of Oklahoma Doppler on Wheels took a range-height scan at 0104 UTC on 15 February 2000 in northern Utah showing lobes of higher reflectivity about 1 to 2 km across beneath a forward-sloping cloud base, with wind-speed minima at the bottoms of the largest lobes. The authors add the limitation that matters here: it was not obvious that the mammatus could be identified in the same radar’s plan-position scans. A WSR-88D volume scan is a stack of plan-position cones.

What radar does show of the parent storm is the anvil’s precipitation and its echo tops: the outline of what is falling inside the storm, not of the towers and anvil you can see. Stepping up through tilts over the anvil shows the ice hydrometeors the lobes are made of, averaged into sample volumes far larger than a lobe. It does not show the lobes. For hydrometeors that fall from a cloud base into dry air and evaporate, see virga; the 2006 review concludes the relationship between the two, if any, is unclear.

Next: the shelf cloud, and the fine line radar really does show beneath it.

WeatherStory Radar renders every elevation of a Level II volume as 3D storm structure, so the anvil above a mammatus display can be looked at from the side rather than only from above.