The rear-flank downdraft
The sinking air that wraps around a supercell’s mesocyclone: how it forms, why its temperature seems to matter for tornadoes, and where it shows up in reflectivity and velocity.
A rear flank downdraft (RFD) is a region of dry air sinking on the back side of a supercell’s mesocyclone and wrapping around it. That is the National Weather Service glossary definition, which adds that the RFD is often visible as a clear slot wrapping around the wall cloud.
The supercell guide introduces the RFD as the air that carves the hook and notes that its buoyancy seems to matter for tornadoes. This page goes further: what drives the downdraft, what field observations found about warm and cold RFDs, how its leading edge occludes the circulation, and what the RFD looks like in radar data.
How it forms
No single mechanism accounts for every RFD, and the literature describes two broad families that usually act together.
- Thermodynamic forcing. Precipitation falling into drier air evaporates and cools it, and the falling rain and hail drag air down with them. Cooler, heavier air sinks.
- Dynamic forcing. The rotating updraft acts as an obstacle to the flow at middle and upper levels. Pressure builds on its upwind side, and the resulting downward-directed pressure-gradient force pushes air down even without cooling. Robert Davies-Jones of the National Severe Storms Laboratory summarizes the RFD as arising “from evaporative cooling or from a downward nonhydrostatic vertical pressure-gradient force.”
The classic synthesis is Lemon and Doswell (1979), in Monthly Weather Review. Drawing on radar, aircraft, visual and surface observations, they described a downdraft originating around 7–10 km above ground on the upwind side of the updraft. Air slows against the storm, is forced downward, and reaches the surface through evaporative cooling and precipitation drag. The mesocyclone is then transformed into a divided structure, with its circulation center on the boundary between downdraft and updraft. That conceptual model still anchors how the storm is drawn.
Warm RFDs, cold RFDs
Air that sinks without much rain in it warms by compression, so an RFD can reach the ground close to the temperature of the air feeding the storm, or even warmer. Air that has been chilled by evaporation arrives cold. The NWS Louisville training material on supercell structure puts it plainly: the properties of the RFD have an impact on whether a tornado forms.
The observational basis is Markowski, Straka and Rasmussen (2002), also in Monthly Weather Review. They drove instrumented vehicles (mobile mesonets) through the hook echoes of 30 supercells, 18 tornadic and 12 not, largely during the VORTEX field project. Tornado likelihood, intensity and longevity increased as the RFD’s equivalent potential temperature, virtual potential temperature and CAPE increased and its convective inhibition decreased. On average, RFD air in tornadic storms was only slightly cooler, in those terms, than the inflow.
The physical reading: rotation near the ground has to be stretched upward into the storm. If the air carrying it is too cold and dense to be lifted, it pools at the surface instead. Davies-Jones notes that simulations show a tornado will not form when the air at the ground is too negatively buoyant to be lifted.
Two cautions keep this from becoming a rule. Grzych and colleagues, sampling ten RFDs in Project ANSWERS in 2003, found results broadly consistent with the 2002 study but also two thermodynamically warm RFDs that produced no tornado: a warm RFD is not sufficient. And Lee, Finley and Karstens (2012) found the RFD is not uniform. In the 22 May 2010 Bowdle, South Dakota supercell they documented four internal surges within one mesocyclone cycle. Significant tornadoes developed and matured while the RFD was kinematically strong and only weakly negatively buoyant, and the demise of the EF4 Bowdle tornado coincided with a much cooler surge.
The clear slot and the gust front
From the ground, the RFD is often the clear slot: a narrow wedge of clearing or higher cloud base cutting in behind the wall cloud. Davies-Jones describes it wrapping around the region where the tornado develops, about 5 to 10 minutes before the tornado forms, with rapidly sinking, evaporating cloud fragments along its edges. The wall cloud guide covers how to read that lowering itself.
When the downdraft reaches the ground it spreads out, and its leading edge is the rear-flank gust front. At low levels the storm then resembles a miniature extratropical cyclone: the forward-flank outflow boundary sits in the warm-front position and the RFD gust front in the cold-front position. As the RFD advances around the circulation it occludes the mesocyclone, and the tornado typically forms near the point of occlusion. A new mesocyclone may begin forming along the bulge in the gust front to the right of the old one — the cyclic behavior the Bowdle storm showed.
How the RFD appears on radar
Radar never sees the downdraft’s temperature and rarely sees its vertical motion directly. It sees what the RFD does to precipitation and to horizontal wind.
| Product | What to look for | Caveat |
|---|---|---|
| Reflectivity | A hook or pendant; a curved leading edge of precipitation; reflectivity thinning where dry air wraps in | Rain-wrapped storms fill the hook in |
| Base velocity | Stronger flow within the RFD, and a wind shift marking its boundary | A gradual veer along the boundary can hide it |
| Storm-relative velocity | The low-level mesocyclone the RFD wraps around | Less accurate than base velocity for placing the boundary, which moves along the ground |
The NWS glossary ties the hook to the RFD directly: large rain and hail at the interface between the clear slot and the wall cloud can show up as a hook or pendant, so a hook may indicate an RFD. The hook echo guide covers that shape in detail.
The largest radar study of the RFD before strong tornadoes comes from the Storm Prediction Center. Broyles and colleagues examined WSR-88D data for 208 supercells that produced EF3 to EF5 tornadoes between 2008 and 2019. They identified an RFD surge and an RFD occlusion before every one, with the occlusion typically forming 5 to 10 minutes before the tornado. The tornado formed well inside the RFD in 201 of the 208 cases (96.6%).
Their method is a useful lesson in reading the data. Reflectivity was the better tool for locating the RFD boundary in 130 cases (62.5%), because base velocity often did not show the boundary when winds veered gradually across it. When velocity was clearer, base velocity beat storm-relative velocity, because the boundary moves along the ground and is better placed with a ground-relative view.
RFD winds and tornado damage
An RFD is a damaging-wind source in its own right. NWS Louisville notes that RFD winds have been known to exceed 70 mph, and that significant straight-line damage can occur along the RFD gust front just south of a tornado’s location. In the NWS Norman survey of the 31 May 2013 El Reno tornado, the damaging wind swath was much larger than the tornado itself, with non-tornadic downdraft winds extending at least a mile south of it.
That is why a damage swath beside a rotating storm is not automatically a tornado track. When the NWS surveys damage it weighs debris patterns, radar imagery, photos, video and eyewitness accounts. Without conclusive evidence of rotation extending from the ground into the parent storm, the damage is classified as non-tornadic.
WeatherStory Radar reads full-resolution Level II reflectivity, velocity and storm-relative velocity, every gate, from NOAA’s public archive.