Backbuilding is a pattern in which new thunderstorm cells keep forming on the upwind side of an existing storm or storm complex, usually the west or southwest side in the United States, so the rain area seems to stay in place or even creep backward while each individual cell moves away.
How a storm complex can stand still
A storm complex moves by two processes at once. Individual cells are carried along by the mean wind through the cloud layer, while new cells form wherever moisture, instability and lift come together. That second process is called propagation.
When new cells form on the upwind flank fast enough to cancel the downwind drift of the older ones, the system barely moves. In a 1998 Storm Prediction Center (SPC) paper, Stephen Corfidi calls that case quasi-stationary and reserves back-building for systems where propagation outruns advection by the mean wind.
The lift for that repeated development can come from storm-generated outflow boundaries, or from a low-level jet converging along the upwind flank of the system, which keeps low-level convergence focused on the back side.
How it differs from training
Backbuilding and training storms can show up together, but they describe different parts of the problem. Training is the path: rain areas pass one after another over the same ground. Backbuilding is the supply: new cells keep regenerating at the upwind end, so the line keeps getting refilled.
Research by Russ Schumacher and Richard Johnson describes backbuilding systems as ones where cells repeatedly form upstream of their predecessors and pass over a particular area, producing large local rainfall totals. That combination can turn a few storms into a multi-hour flood event.
How forecasters anticipate it
The SPC mesoanalysis page carries an upwind propagation vector from a method published by Corfidi and colleagues in 1996. It estimates system motion as the mean wind through the cloud-bearing layer plus a propagation term assumed equal and opposite to the 850 mb low-level jet.
When those two pieces nearly cancel, the estimated motion is slow, a signal that a system could stall. Weather Prediction Center training material notes the original vector method usually works best for back-building systems, where new cells on the rearward flank slow the system down.
What the vectors do not tell you
The vectors are a first guess based on two winds. They do not show the outflow boundary, front or terrain feature where cells will actually keep forming, so radar trends and surface analysis decide the short-term call.
Corfidi also notes that quasi-stationary systems typically sit within axes of the highest deep-layer equivalent potential temperature (theta-e) or precipitable water. Both quasi-stationary and back-building systems are frequently tied to excessive rainfall and flash flooding, so moisture and rain rate complete the picture.