Two floods, and how this platform checks yours
Two documented events with public data, then the reason a flood report gets a wider window and a longer memory than every other report here.
Two worked examples
Both events below happened on the northern Gulf coast, and they were chosen for one unglamorous reason: the radar, gauge, survey and post-storm data are all public, so every number here can be checked. Neither is offered as the typical flood, and lesson 3 has already said there is no such thing. What generalises is not the geography but the two behaviours they both show, and those are the subject of the rest of the lesson: flooding lags its rainfall and happens downstream of it. Substitute your own regime and the lag and displacement change in size, not in kind.
29 to 30 April 2014
A slow-moving cold front stalled, and the rain came in two rounds. The first, overnight into the 29th, put down 3 to 8 inches. The second, from the evening of the 29th into the early hours of the 30th, put down 10 to 15 inches in an estimated nine hours on ground that had just taken the first round.
- Night, 28 to 29 AprRound one. 3 to 8 inches across the area. Ground saturates; Flooding that begins within about six hours of whatever caused it, counted from the causative event rather than from when anyone noticed the water. The cause is usually intense rainfall, but a dam failure, a levee failure or an ice jam qualifies and can do it on a clear day. guidance falls.
- Evening, 29 AprRound two begins on saturated ground. Rainfall rates reach about 5.68 inches in one hour.
- Overnight, 29 to 30 AprPeak flash flooding. Sinkholes open, roads are cut, water rescues are under way, part of Scenic Highway collapses. One confirmed fatality. Parts of I-10 close.
- 30 Apr onwardThe rivers respond after the rain has stopped. Fish River at Silver Hill crests at a record 23.18 feet, beating 22.78 feet from July 1997.
The NWS estimated more than 20 inches at Pensacola, against a two-day record of 20.47 inches. That figure is itself estimated rather than measured: a power outage took the gauge out, and the total had to be reconstructed from radar. The worst-case six hour rainfall was placed by NOAA Atlas 14 somewhere between a 1 in 100 and a 1 in 200 annual exceedance probability.
Hurricane Sally, 15 to 16 September 2020
A category 2 hurricane that came ashore near Gulf Shores at a forward speed of about 3 miles an hour. Lesson 2 said duration dominates; this is what that looks like at its limit. Rainfall ran 15 to 30 inches across Baldwin County into the western Panhandle.
The National Hurricane CenterThe national office for tropical cyclone forecasts and warnings. report gives the extremes: "The highest reported rainfall total from the entire event was 29.99 inches at Orange Beach, Alabama", and across the state line "a maximum total of 24.88 inches was measured". Thousands of water rescues followed. Downtown Pensacola took 2 to 4 feet of combined freshwater and surge inundation, and the Shoal River near Crestview reached 16.42 feet on the 17th, above its 15 foot major The gauge height at a given river point above which flooding begins to threaten life or property there. It is a local number tied to that gauge, not a depth of water on the ground. and the second highest The highest level a river reaches in an event, and the time it reaches it. On large catchments the crest routinely arrives a day or more after the rain has stopped, and well downstream of where it fell. on record behind 21.40 feet in September 1998.
Sally is also the Heavy rainfall arriving at the same time as elevated coastal water, so the runoff has nowhere to discharge. The surge does not add water inland, it removes the drainage, which is why the result can be worse than either forecast alone. case from lesson 3, in the NHC's own words: the surge blocked the drainage the rainfall needed. And note the date on that Shoal River crest. The storm came ashore on the 16th. The river peaked on the 17th.
Lag and displacement
Both cases show the same two properties, and together they are why this hazard needs its own validation rules.
- Flooding lags its rainfall. Instruction 10-1605 notes that flooding "may continue to threaten life or property many days after the rain ends". The Fish River record crest and the Shoal River crest both arrived after the rain had stopped.
- Flooding occurs downstream. The water that floods a town fell on a catchment, and the catchment can be twenty miles upstream of the flooding and under a completely different sky.
How this platform checks a flood report
Module 5 ended on the same principle and it holds here with more force: plausibility check, never auto-reject. A flood report is corroborated when heavy rain or hydrologic signals are consistent nearby in space and time, and the definition of nearby is what changes.
The 40 km figure is borrowed from the neighbourhood the Weather Prediction CenterThe national office for precipitation and excessive-rainfall forecasting. Excessive Rainfall OutlookThe WPC product forecasting the probability that rainfall will exceed flash flood guidance within 40 km of a point. Marginal at least 5%, Slight 15%, Moderate 40%, High 70%, and High is issued on days 1 to 3 only. and flash flood guidance already work in, which makes the comparison against those products coherent. The asymmetry matters: flooding follows rain and never precedes it, so the window opens at the rainfall and extends forward only.
The products a corroboration draws on are the ones from lesson 6: radar-only and multi-sensor quantitative precipitation estimateHow much rain the radar thinks fell. Radar-only QPE is fast; multi-sensor QPE is adjusted toward the rain gauges that reported inside it and is better and later. Neither is a measurement of rainfall., rainfall rate, the QPE to flash flood guidanceThe rainfall needed over a given area and duration to bring small streams up to bankfull, given how wet the ground currently is. Issued by the river forecast centres and recalculated as conditions change. ratio, average recurrence intervalHow unusual an observed rainfall is at that place, expressed as a return period. A rare rainfall is useful context and is not the same statement as "this will flood"., Flooded Locations and Simulated HydrographsAn NSSL system that runs hydrologic models on the MRMS rainfall field over a grid of roughly one kilometre, turning a rainfall estimate into a streamflow estimate. unit streamflow, and United States Geological SurveyThe agency that operates the national stream gauge network. A rapid stage rise on a USGS gauge is the strongest single corroboration a flood report can get. stream gauges, where a rapid stage rise is the strongest single confirmation a flood report can get. Every one of them has limitations that are encoded alongside it rather than assumed away.
What the report record is biased toward
Module 5 taught that hail reports cluster around people and roads. Flood reports do the same thing and worse, because the hazard is literally defined by its interaction with infrastructure: a flooded road needs a road.
- Reports follow the road network and population. Flooding in an empty catchment produces no report and is not therefore absent.
- Location and time uncertainty are high. The observer is usually not in the water, often reports later, and describes an extent rather than a point.
- Nighttime under-reporting is severe, for exactly the reasons lesson 4 gave: nobody is out, and those who are cannot see.
The design consequence is stated plainly, because it is a decision this platform has already made: the absence of a nearby report is weak evidence, and hydrologic signals are weighted accordingly. You are not filing into a system that assumes silence means nothing happened.
A report of major river flooding arrives 30 hours after the rain ended, from a town 25 miles downstream of where the heaviest rain fell. How should it be handled?
Both case studies in this lesson had their record crests on a day when it was not raining.Both properties in the question, the lag and the displacement, are the normal behaviour of riverine flooding rather than warning signs. The Fish River crested at a record after the April 2014 rain had ended, and the Shoal River peaked the day after Sally came ashore. The corroboration to look for is upstream rainfall and a stream gauge rise, not rain over the town.