Case Study · Hurricane · 2017
August 2017. Hurricane Harvey made landfall near Rockport as a Category 4, then stalled. For four days it sat over Southeast Texas and poured. 60.58 inches. The most rainfall from a tropical cyclone in U.S. history. Most flooded homes were outside FEMA's 100-year floodplain — areas the official maps had designated as low risk. Harvey didn't find a gap in the flood mapping system. Harvey proved the gap had always been there.
Houston, Texas · August 25–31, 2017
Hurricane Harvey made landfall near Rockport, Texas on August 25, 2017 as a Category 4 hurricane with 130 mph sustained winds. Rockport was badly damaged by the wind. But the wind was not Harvey's story. Harvey's story began when it weakened to a tropical storm and stalled. A blocking high-pressure system prevented Harvey from tracking out of the region in the normal fashion. The storm sat over Southeast Texas for four days, pulling moisture from the warm Gulf of Mexico and dumping it on the same area repeatedly. In that window, the storm produced rainfall totals that the National Hurricane Center's tropical cyclone report characterized as "catastrophic." At Nederland, Texas, Harvey dropped 60.58 inches — the highest rainfall total ever recorded from a tropical cyclone in the continental United States, with a return period that World Weather Attribution's analysis estimated at more than 9,000 years for the three-day rainfall peak at Baytown.
The scale of the flooding was extraordinary. FEMA reported approximately 30,000 water rescues across Texas. More than 200,000 homes were damaged or destroyed. More than 30,000 people were displaced. 68 direct deaths were attributed to the storm — predominantly by drowning, reflecting Harvey's character as a flood event, not a wind event. But the number that became Harvey's most enduring policy legacy was not the rainfall total or the death count. It was a fraction: the proportion of flooded homes that were located outside FEMA's designated 100-year floodplain. According to the American Flood Coalition's five-year retrospective, most homes affected by flooding from Harvey were outside the 100-year floodplain. Those homeowners had received no mandatory flood insurance requirement. They had received no official designation indicating they were in a flood hazard area. The flood maps had told them, in effect, that they were safe.
Aug 25, 2017
Landfall
68
Direct Deaths
$125B
Damage (2017 USD)
60.58"
Peak Rainfall
Category 4
At Landfall
The American Flood Coalition's analysis found that the problem had a documented cause: Texas was using rainfall data from as far back as 1961 to map its floodplains. In 2018 — the year after Harvey — NOAA released updated statewide rainfall estimates known as Atlas 14, which revealed that Texas was already receiving significantly more rainfall than those 1961-era baselines indicated. The floodplain maps, in other words, had been mapping risk against a historical baseline that was already out of date before Harvey arrived. Harvey didn't create the mismatch between the maps and reality. Harvey revealed it. At $125 billion in damage — second-costliest hurricane in U.S. history — it was an expensive disclosure.
The Science
Think of a hurricane as a rain-producing machine that moves across the landscape. A fast-moving storm drops heavy rain for hours over any given location before the system tracks away. Harvey didn't track away. A blocking high-pressure ridge prevented the storm from moving in its expected direction, and Harvey essentially parked over Southeast Texas. The 24.5 trillion gallons of water that Axios estimated fell on Texas and Louisiana didn't come from an unusually intense moisture source — it came from a machine that didn't leave. The same Gulf moisture that feeds every Gulf Coast storm fed Harvey, repeatedly, over the same geography, for four days. The total rainfall was not a function of Harvey's intensity. It was a function of its immobility.
FEMA's flood maps designate the "100-year floodplain" as the area with a 1% annual chance of flooding — a statistical concept based on historical rainfall and runoff data. When those maps are built on old rainfall records, they systematically understate risk in areas where rainfall patterns have changed. The American Flood Coalition's analysis of Harvey found that Texas was using rainfall data from as far back as 1961 to set these flood boundaries. NOAA's Atlas 14 update in 2018 — released the year after Harvey — found that Texas was already receiving more rain than those older baselines indicated, before accounting for any future change. The 100-year floodplain designation, in much of greater Houston, was based on a 1961-era precipitation baseline that the actual climate had already exceeded. Harvey didn't flood those homes by producing unprecedented rainfall. It flooded them by producing rainfall that was entirely consistent with the updated baseline those maps hadn't yet incorporated.
Houston is one of the most developed large metropolitan areas in the United States without comprehensive land-use zoning. The rapid expansion of impervious surface — paved roads, parking lots, rooftops — across Harris County over recent decades has dramatically reduced the soil's capacity to absorb rainfall. In natural or semi-natural landscapes, a significant fraction of precipitation infiltrates into the ground or is absorbed by vegetation. In a heavily developed urban landscape, the same rainfall runs off almost immediately into the drainage system and nearby bayous. Harvey's rainfall totals would have produced flooding in any landscape. In metropolitan Houston, the urban hardscape accelerated runoff and overwhelmed drainage systems that were already operating at capacity from previous storm events.
Timeline
01
August 25, 2017: Harvey makes landfall near Rockport, TX as a Category 4 hurricane with 130 mph winds. Rockport is badly damaged. The NHC tropical cyclone report documents catastrophic wind damage along the initial landfall corridor. Harvey then weakens to a tropical storm and slows. A blocking high-pressure system prevents it from tracking out of the region.
02
August 26–29: Harvey stalls over Southeast Texas and dumps rainfall over the same geography for four consecutive days. Houston receives catastrophic totals — streets, highways, and bayous overwhelmed simultaneously. I-10, I-45, and US-59 are submerged. Record water levels on Buffalo Bayou, Clear Creek, Dickinson Bayou, and Cypress Creek. FEMA's 30,000 water rescues begin; informal volunteer rescues by the tens of thousands add to the count.
03
August 31: Cedar Bayou station logs 51.89 inches. Total accumulation at Nederland, TX reaches 60.58 inches — shattering the U.S. record for tropical cyclone rainfall. 24.5 trillion gallons of water have fallen on Texas and Louisiana combined. Most flooded homes are outside the FEMA 100-year floodplain. 200,000+ homes are damaged or destroyed. 68 direct deaths, predominantly drowning. $125B in damage.
04
2018 and beyond: NOAA releases Atlas 14 — updated statewide rainfall estimates showing Texas already receives more rain than 1961-era flood maps reflected. Harris County begins $2.5B flood bond program for infrastructure investment. The American Flood Coalition documents that most Harvey-flooded homes were outside 100-year floodplain. National conversation about floodplain map accuracy, flood insurance reform, and development regulation in flood-prone areas is elevated to legislative priority.
Human Decisions
What worked
FEMA's 30,000 official water rescues represent only part of what occurred. Thousands of private citizens arrived from Louisiana and elsewhere with their own boats — an informal volunteer rescue network that the media dubbed the "Cajun Navy" — and conducted rescues that formal emergency services could not reach. The PMC hospital preparedness analysis of Harvey documents that formal emergency response organizations were overwhelmed and required volunteer assistance. The community-level response capacity demonstrated in Harvey became a case study in how distributed, informal rescue networks can function as essential emergency infrastructure when professional services are saturated.
In 2018, Harris County voters approved a $2.5 billion flood infrastructure bond — the largest flood control bond in U.S. history to that point. The American Flood Coalition's retrospective documents that the county began significant bayou widening, detention pond construction, and home buyout programs in the most flood-prone areas. This represents a genuine structural investment in the hydrological infrastructure of the region, motivated directly by Harvey's documentation of the gap between mapped flood risk and actual flood exposure.
What failed
The American Flood Coalition's analysis of Harvey is specific: Texas was using rainfall data from as far back as 1961 to designate floodplains. NOAA's Atlas 14 update, released the year after Harvey, found that Texas was already receiving more rainfall than those older estimates indicated. The 100-year floodplain designation for millions of Houston-area properties was based on a baseline that the actual historical record had already exceeded. Homeowners who bought in good faith outside the designated floodplain had made a risk assessment based on maps that were systematically understating their exposure.
Outside the FEMA-designated Special Flood Hazard Area, flood insurance is not required as a condition of a federally backed mortgage. Most homeowners outside those zones do not carry flood insurance voluntarily. Harvey flooded an enormous number of homes outside those zones and exposed a population of homeowners who had been told — accurately, by the existing mapping system — that they were not in a high-risk flood area, but who sustained catastrophic losses and had no insurance to cover them. The National Flood Insurance Program's exposure-vs.-mapped-risk gap, which Harvey made impossible to ignore, has driven ongoing federal flood insurance reform discussions since 2017.
The ScienceDirect social vulnerability research on Harvey found that flooding from the storm was disproportionately extensive in Black, Hispanic, disabled, and economically disadvantaged neighborhoods. In many cases, these communities had developed in areas that were affordable precisely because they sat closer to bayous and drainage corridors — areas that were designated low-risk on maps but that Harvey demonstrated had significant actual flood exposure. The intersection of inadequate flood maps, development in topographically vulnerable areas, and populations without resources to self-insure or self-evacuate produced a pattern of disparate impact that is consistent across major flood events nationally.
The cascade lesson
The homes that flooded in Harvey had not been designated as flood-risk properties. Their owners had made rational decisions based on official flood maps. Those maps were built on 1961-era rainfall data that didn't reflect the rainfall Texas was already receiving, let alone Harvey's extreme totals. The 100-year floodplain boundary meant something specific: it was the area that historical data suggested had a 1% annual chance of flooding. When that historical data understates reality, the boundary is in the wrong place. Harvey's central lesson is not that you should expect extreme weather — it's that official risk designations are only as accurate as the data they're built on, and that the data can be systematically wrong. The question every household should ask is not "Am I in the flood zone?" but "Does the flood zone map reflect the actual flood history of this area?"
What Changed
In 2018, NOAA released Atlas 14 — updated precipitation frequency estimates for Texas that incorporated more recent historical data. The Atlas 14 values were significantly higher than prior estimates across much of the state, confirming that Texas was already receiving more rainfall than the 1961-era baseline that had informed its floodplain maps. This data revision, motivated substantially by Harvey's documentation of how wrong the old estimates were, has since driven updates to floodplain mapping methodology in multiple states. FEMA's ongoing National Flood Insurance Program reforms include an updated risk rating methodology that incorporates more current flood risk data.
In August 2018, Harris County voters approved a $2.5 billion flood control bond — at the time the largest flood bond in U.S. history. The Harris County Flood Control District has since used those funds for bayou widening, detention basin construction, and voluntary home buyout programs for repeatedly flooded properties. The American Flood Coalition's 2022 retrospective notes that these investments represent a genuine rethinking of how the county manages its hydrological infrastructure, moving from reactive repair after floods to proactive structural capacity expansion. The bond was passed with 85% voter approval — reflecting the degree to which Harvey had made flood resilience a cross-community priority.
What You Can Do Now
Harvey's lessons apply beyond Texas. Any household near a river, creek, bayou, drainage ditch, or low-lying area should work through these five actions.
FEMA's Flood Map Service Center (msc.fema.gov) shows your property's current flood zone designation. But also search your county's flood control district records and ask neighbors about past flooding. The gap Harvey exposed — between mapped flood zones and actual flood history — is not unique to Texas. In many regions, official floodplain maps haven't been updated to reflect recent flood events or updated rainfall data. If your neighbors recall flooding that the maps don't show, believe your neighbors.
Find your local flood resourcesFlood insurance is available through the National Flood Insurance Program (NFIP) for properties outside high-risk zones — and it is typically far less expensive than coverage in mapped Special Flood Hazard Areas. Harvey's most expensive lesson for individual households is that being outside the 100-year floodplain does not mean you cannot flood catastrophically. Your homeowner's insurance does not cover flood damage. For the cost of one home repair estimate, you can carry a year of flood insurance that covers the scenario Harvey proved is possible outside mapped zones.
Insurance resilience guideHarvey's flooding was not produced by its winds at landfall. It was produced by its multi-day stall. A tropical system that is forecast to slow or stall over your region presents a fundamentally different risk than a fast-moving storm. Follow NHC track and intensity forecasts closely, but also watch for forecasts of unusually slow movement over land — that is the Harvey scenario. A day's total rainfall from a stalled tropical storm can exceed a week's worth of rain from a moving hurricane.
Hurricane preparedness guideThe most common cause of flood deaths — in Harvey and in every major flood event — is drowning in vehicles. Two feet of moving water can sweep away most cars. Water depth on a flooded road is impossible to judge visually. The NOAA "Turn Around, Don't Drown" guidance is not a slogan — it is the behavioral instruction that, if followed, eliminates the most frequent flood fatality scenario. If the road ahead is flooded, find an alternate route or wait. The risk of your vehicle being swept off the road is real and immediate; the inconvenience of waiting is not.
Flood safety guide30,000 water rescues means 30,000 people who did not have a plan that got them out before the water rose to rescue-requiring levels. Knowing your area's flood evacuation routes, identifying an out-of-area destination, and having a go bag with documents, medications, phone chargers, and several days of supplies means you can make the decision to leave and execute it quickly — before roads are impassable. Harvey's flooding developed over days; most of those 30,000 rescues could have been avoided with earlier departure decisions.
72-hour go bag guideMore hurricane case studies
Katrina (2005) covers levee failure. Andrew (1992) covers building codes. Sandy (2012) covers storm surge. Maria (2017) covers long-term infrastructure collapse. Each teaches a different lesson about what hurricanes actually destroy and how.
Full hurricane case study seriesSources