Missouri and the New Madrid Seismic Zone: Updated Risk Assessment

On the night of December 16, 1811, the ground beneath the small settlement of New Madrid in the Missouri Territory shook so violently that river boatmen 300 miles downstream were thrown from their bunks. The Mississippi River churned and foamed. Sand boils erupted across the floodplain by the thousands. Trees snapped. Banks caved into the river. The first event — estimated today at approximately M7.5–7.7 — was followed by a M7.0 aftershock within hours. Then, on January 23, 1812, a second mainshock of comparable size struck. Then, on February 7, 1812, the third and possibly largest event — estimated at M7.7–8.0 — generated ground motion so severe that eyewitnesses reported the Mississippi River running backward in some reaches as the land subsided upstream of the rupture and rose downstream. Church bells rang in Boston, more than 1,500 kilometers from the epicenter. New lakes were created from dry land as the terrain collapsed. Sand and sediment erupted through the soil surface across 15,000 square miles of the Central Mississippi Valley in what remains the most spectacular paleoliquefaction field in the geological record of North America.

The 1811–1812 New Madrid sequence is the largest earthquake sequence in the recorded history of the eastern United States — and it occurred in a region that today contains the cities of St. Louis, Missouri; Memphis, Tennessee; and more than a dozen smaller cities directly atop the seismic zone's fault system, with a combined metropolitan population approaching 5 million people. The broader area potentially affected by a repeat of the 1811–1812 sequence encompasses approximately 12 million people across Missouri, Illinois, Tennessee, Arkansas, Kentucky, and Mississippi — a region where earthquake preparedness lags decades behind California despite hosting what may be the most consequential seismic hazard in the interior of the continental United States.

What makes the current scientific situation unusual — and what makes this post more than a simple recounting of historical hazard — is that the New Madrid Seismic Zone is the subject of one of the most consequential and least-resolved debates in American earthquake science. A series of GPS-based geodetic studies, published beginning in the early 2000s and refined since, has found little to no detectable elastic strain accumulation on the NMSZ faults — the opposite of what a reloading seismic zone accumulating strain toward another 1811–1812 sequence should show. This geodetic finding, if taken at face value, would suggest that the NMSZ is either aseismically releasing its strain, is not currently accumulating strain toward a new great earthquake, or is operating on a fundamentally different mechanical model than plate boundary faults. The implications for hazard assessment are profound — and have not been resolved.

The Reelfoot Rift: The Geological Host

The New Madrid Seismic Zone sits within the Reelfoot rift — a failed continental rift that formed approximately 500 million years ago during the Late Proterozoic and Cambrian periods, when what is now the central United States nearly rifted apart but did not complete the process. The rift is a subsurface feature, buried beneath hundreds of meters of younger sedimentary rocks in the Mississippi Embayment — a broad southward-facing reentrant in the basement topography where the Cretaceous seas advanced far inland, depositing the thick sequence of soft sediments that now underlie the Mississippi River floodplain.

The Reelfoot rift is approximately 300 kilometers long and 70–100 kilometers wide, oriented roughly northeast-southwest and crossing the modern Mississippi Valley near its juncture with the Ohio and Missouri Rivers. The rift's primary structural elements — the Cottonwood Grove fault, the Reelfoot fault, and the New Madrid North fault — form a curving, interlocking fault system that in 1811–1812 produced the three major mainshocks in rapid succession as stress transferred from one fault to the next through Coulomb failure.

The Sedimentary Cover: Why the Mississippi Valley Amplifies

The geological setting of the NMSZ includes a factor that would make a repeat of 1811–1812 dramatically worse for modern populations than the original event: the soft, water-saturated sediments of the Mississippi Embayment that blanket the entire region. The Mississippi River floodplain is underlain by Holocene and Pleistocene alluvial sands and silts to depths of 20–50 meters, overlying Cretaceous and Tertiary marine sediments that in turn overlie the crystalline basement at depths of 800–1,500 meters beneath the deepest part of the embayment.

This thick, soft sedimentary column serves as a profound amplifier of seismic energy. Studies have shown that the Mississippi Embayment sediments can amplify ground motion by factors of 5–15 at the dominant periods most damaging to buildings, and that the high water table and loose alluvial sands produce some of the highest liquefaction susceptibility anywhere in the United States. The 1811–1812 sequence produced the most extensive liquefaction field in recorded North American history — sand boils over thousands of square kilometers. A modern repeat would liquefy the foundations of buildings, roads, bridges, levees, pipelines, and utility infrastructure across an enormous area, with consequences extending far beyond the direct shaking damage.

The Paleoseismic Record: Evidence for Prehistoric Sequences

The most important scientific evidence for understanding the NMSZ's long-term behavior comes not from the 215-year instrumental and historical record since 1811 but from the paleoliquefaction record extending thousands of years into the past. Systematic mapping and radiocarbon dating of prehistoric sand blow features — the preserved remnants of ancient liquefaction episodes — has documented at least two major earthquake sequences preceding 1811–1812, both of comparable size to the historical sequence.

Paleoliquefaction studies by Tuttle, Schweig, and others identified evidence for a prehistoric sequence at approximately 900–1000 CE — roughly 800–900 years before 1811 — and an earlier sequence at approximately 300–600 CE. Both prehistoric sequences produced sand blow fields of similar size and spatial distribution to the 1811–1812 features, implying comparable earthquake magnitudes (M7.5–8.0) on the same fault system. The recurrence interval derived from these data is approximately 500–800 years between major sequences — with the caveat that only three sequences in approximately 1,500 years of record provide a very limited statistical sample.

At 215 years since the last great sequence, the NMSZ is well within the range of observed interseismic intervals — the minimum observed interval is roughly 400–500 years, meaning that by the paleoseismic record alone, the next great sequence is not statistically overdue. But the GPS debate described below complicates this interpretation significantly.

The GPS Controversy: Is the Zone Reloading?

The most consequential scientific development in NMSZ research in the past two decades is a series of GPS geodetic studies that have found remarkably low rates of surface deformation across the NMSZ fault system — substantially lower than would be expected if the fault were accumulating elastic strain at the rate required to produce M7.5+ earthquakes on the paleoseismic recurrence interval of 500–800 years.

The Strain Rate Problem

On a fault accumulating strain at rate V with shear modulus μ and seismogenic area A, the seismic moment per unit time is μ × V × A. For the NMSZ to produce the 1811–1812 sequence (total seismic moment approximately 3 × 10²⁰ N·m) on a 500-year recurrence, it would need to accumulate strain at a rate that should be detectable by GPS as approximately 2–4 mm/yr of surface velocity gradient across the fault zone — comparable to what GPS networks routinely measure across active faults in the western US.

Multiple GPS studies, using dense networks of monuments across the Central Mississippi Valley, have found no detectable surface velocity gradient attributable to locking of the NMSZ faults — velocity gradients consistent with zero strain accumulation to the measurement uncertainty of approximately 0.3–0.5 mm/yr. This finding, reported by Newman, Holt, and others beginning in 1999 and confirmed in subsequent studies with improved GPS networks, created an apparent contradiction: a fault system with clear paleoseismic evidence for M7.5–8.0 earthquakes on a 500–800 year cycle is showing no detectable elastic strain accumulation at the surface in 25+ years of GPS monitoring.

The Post-Seismic Relaxation Model

The most widely discussed resolution to the GPS-paleoseismic discrepancy is the post-seismic relaxation model, developed primarily by Seth Stein of Northwestern University and colleagues. This model proposes that the apparent lack of current strain accumulation reflects a genuine physical state: the 1811–1812 earthquakes released so much stored elastic energy that the surrounding crust is now in a post-seismic relaxation period, during which viscoelastic flow in the lower crust and upper mantle is redistributing the stress released by the great earthquakes. In this model, elastic strain would not begin to re-accumulate significantly until the post-seismic relaxation is complete — potentially hundreds of years from now.

If the post-seismic relaxation model is correct, the next great NMSZ sequence is likely centuries away, the GPS data accurately reflects the current non-accumulation state, and the USGS hazard maps — which are based partly on the paleoseismic recurrence — may overestimate the current probability of great earthquakes at the NMSZ. This is essentially the "reassuring" interpretation, and it is the one that Stein and colleagues have advocated most prominently in the scientific literature and in public policy discussions.

The Persistent Hazard Model

The alternative — championed by Roy Van Arsdale of the University of Memphis, Mike Tuttle, and others — maintains that the GPS data does not resolve the hazard because the strain accumulation mechanism at the NMSZ may not produce surface-detectable deformation on the timescale of GPS monitoring. Several mechanisms could explain this: deep creep on the fault below the seismogenic zone that partially accommodates loading without surface deformation, aseismic slip on the fault surface at rates below GPS detection, or strain accumulation distributed over such a wide area that the localized GPS gradient at the fault zone itself is below detection limits.

This model points to the fact that the background microseismicity of the NMSZ — 200+ M2+ events per year, far above the background rate of the surrounding stable continental interior — is inconsistent with a fault zone that is genuinely dormant and not accumulating stress. A dormant fault zone in a stable continental region would not produce this elevated microseismicity rate. Something is maintaining the critically stressed state of the NMSZ fault system, even if that something does not manifest as detectable surface deformation on a GPS timescale.

The Current USGS Hazard Assessment

The USGS National Seismic Hazard Map for the Central United States treats the NMSZ as one of the dominant contributors to earthquake hazard in the region. The hazard map is based on a model that incorporates both the paleoseismic recurrence data and the background seismicity rate, with the GPS strain-rate findings entering as an epistemic uncertainty rather than as a definitive update to the hazard model.

The USGS model projects that the 2% probability of exceedance in 50 years (approximately 2,500-year return period) peak ground acceleration at St. Louis and Memphis — the design standard for the most critical structures — is approximately 0.3–0.5g. This is comparable to the hazard levels in some parts of the San Francisco Bay Area and considerably higher than most of the eastern United States outside the NMSZ and Charleston zones. The 10% probability of exceedance in 50 years (500-year return period) PGA — used for standard building code design — is approximately 0.15–0.25g at St. Louis and Memphis, reflecting the fact that the most common design-level earthquakes in this region are moderate (M5–6) events rather than M7.5+ sequences.

What a Modern Repeat Would Do: The Consequence Scenario

FEMA, in collaboration with the USGS and the Central United States Earthquake Consortium (CUSEC), has conducted detailed consequence analyses for a modern M7.7 NMSZ earthquake. The most widely cited scenario — a M7.7 on the Cottonwood Grove fault, roughly equivalent to the December 16, 1811 mainshock — produces results that are sobering for a region that has invested comparatively little in seismic preparedness relative to its hazard level.

Casualties and Displacement

A M7.7 NMSZ earthquake occurring during a winter evening — when populations are predominantly in residential buildings — is projected to kill approximately 3,500–4,000 people and injure 80,000–120,000, primarily from structural collapses of unreinforced masonry buildings throughout Memphis and the smaller cities of the Central Mississippi Valley. The daytime scenario reduces the casualty estimate to approximately 2,000–2,500 deaths as residential occupancy is lower. These estimates are substantially higher than the comparable southern San Andreas scenario in large part because the building stock in the NMSZ region — particularly in older cities like Memphis and St. Louis — contains a high proportion of unreinforced masonry buildings that receive no formal seismic assessment or retrofit requirements under current law.

The Infrastructure Crisis: Bridges, Pipelines, and Levees

The most consequential infrastructure vulnerability in the NMSZ scenario is probably not the building stock — as severe as those consequences are — but the transportation and utility network built on and across the highly liquefiable sediments of the Mississippi River valley. The Mississippi and Ohio River crossings that constitute the lifeline infrastructure of the Central United States — the highway and rail bridges at St. Louis, Memphis, and Cairo — are almost entirely unreinforced or inadequately reinforced against the design earthquake, and many sit on the same liquefiable alluvial sediments that produced the most dramatic geologic features of the 1811–1812 sequence.

The Levee System

The Mississippi River floodplain is protected from routine flooding by an extensive system of earthen levees — the Mississippi River and Tributaries Project, built and maintained by the US Army Corps of Engineers, constitutes one of the most extensive flood control infrastructure systems in the world. These levees are built on the same liquefiable alluvial sediments that will fail catastrophically in a major NMSZ earthquake. Post-earthquake failure of the levee system — in a scenario where the earthquake occurs during spring flood season when river levels are high — could inundate the Mississippi floodplain across hundreds of miles, adding a major flood disaster to the earthquake consequences. The Army Corps of Engineers has evaluated the seismic vulnerability of the levee system and identified significant exposure, but systematic seismic hardening of the levee infrastructure has not been funded at the scale the assessment recommends.

The Preparedness Gap: Central US vs. California

Perhaps the most striking aspect of NMSZ hazard is the enormous gap between the scientific characterization of the risk and the public and institutional awareness of it. California has experienced multiple damaging earthquakes in the modern era — 1989 Loma Prieta, 1994 Northridge — that have driven successive rounds of building code improvement, retrofit programs, emergency management investment, and public awareness campaigns. The Central Mississippi Valley has experienced no damaging earthquake since 1895, when a M5.6 caused limited damage in eastern Missouri. In 130 years, the region has built a modern urban fabric — millions of buildings, thousands of bridges and miles of pipelines, major hospitals and schools — with almost no consideration of seismic design beyond whatever local code provisions were in force at the time of construction.

Tennessee enacted its first modern statewide seismic building code in 1990 — 57 years after California's post-Long Beach code. Missouri's seismic provisions are applied inconsistently across the state. Neither state has a mandatory URM retrofit law for existing buildings. Arkansas, Mississippi, and Kentucky — all within the NMSZ hazard zone — have even less developed seismic regulatory frameworks. The hospital systems of Memphis, St. Louis, and the smaller regional medical centers have not been systematically assessed for seismic performance against the design earthquake, and the mass casualty capacity of the regional trauma system is far below what the scenario would demand.

The School and Hospital Problem

Two categories of building require particular attention in the NMSZ preparedness context: schools and hospitals, for the same reasons they demand attention in every high-hazard region — concentrated occupancy of vulnerable populations, expectation of continued function after a disaster, and the life-safety catastrophe that results when they fail. Unlike California, which has mandatory seismic evaluation and retrofit requirements for schools (Field Act) and hospitals (SB 1953), the states of the NMSZ region have no equivalent mandatory programs. Schools and hospitals in Memphis, St. Louis, and throughout the region were built to the building codes in force at their time of construction — often no seismic provisions at all — and have not been systematically assessed for their performance in the design earthquake.

Memphis, Tennessee has taken more initiative than most NMSZ cities in school seismic assessment — a series of evaluations beginning in the 2000s identified a significant fraction of Memphis-Shelby County Schools as having high seismic vulnerability, and the district has undertaken a phased renovation and replacement program, though it is not yet complete. Similar programs are absent or embryonic in most other NMSZ cities, representing a significant and unaddressed life-safety exposure that could be the single highest-concentration source of casualties in a major daytime earthquake.

The Science-Policy Tension: What the Debate Means for Regulation

The GPS strain-rate controversy has had direct policy implications that make it more than an academic scientific debate. In the mid-2000s, the USGS and other agencies proposed extending seismic design provisions to additional counties in the NMSZ region — specifically, applying higher design ground motions to buildings in counties where the hazard maps showed elevated risk from the NMSZ. This proposal met significant opposition from building industry groups and from some scientists (including Stein) who argued that the hazard was overstated in light of the GPS data and that the additional construction costs were not justified.

The debate resulted in a policy outcome that left many seismologists uncomfortable: building code provisions were not uniformly extended to all at-risk areas, partly in response to scientific uncertainty that, at the time and in context, functioned as a argument for less caution rather than more. The GPS data — which shows absence of detectable strain accumulation but does not prove absence of hazard — was used to support a regulatory outcome that reduced preparedness investment in a region with unambiguous paleoseismic evidence for M7.5+ great earthquakes.

The current scientific consensus — to the extent one exists — is that the uncertainty itself warrants caution. The paleoseismic record is clear: M7.5–8.0 earthquakes have occurred on the NMSZ at least twice in the past 1,500 years, and possibly more frequently. The GPS data provides insufficient basis for concluding that the hazard has disappeared — only that we cannot currently detect the strain accumulation that a simple elastic rebound model would predict. In a region with 12 million people and a building stock built largely to no seismic standards, the appropriate response to uncertainty is investment in resilience, not deferral of it.

Comparing the NMSZ to Other American Earthquake Hazards

In the context of the other major earthquake scenarios covered in this series, the NMSZ occupies a distinctive position. It is not the most probable scenario — the Wasatch Fault's 57% probability in 50 years and the San Andreas's ~60% probability for a M6.75+ event both exceed the NMSZ's roughly 7–10% probability of a M7.5+ event in 50 years. But the scale of potential consequences — $300 billion in losses, 3,500+ deaths, loss of major Mississippi River crossings, and widespread liquefaction across the heartland of American agricultural and industrial infrastructure — rivals or exceeds the San Andreas scenario in economic impact despite the lower probability.

The most critical distinction between the NMSZ and other high-hazard zones is the preparedness gap. California has spent 90 years building seismic resilience into its infrastructure since the 1933 Long Beach earthquake. The NMSZ region has spent 215 years since 1811 not having another great earthquake — a circumstance that has effectively precluded the periodic forcing function of disaster that drives preparedness investment in active seismic regions. The result is a hazard profile where the gap between current resilience and what the scenario demands is the largest of any major American seismic zone — a gap that neither the scientific uncertainty about GPS strain rates nor the relatively low annual probability of a great earthquake justifies leaving unaddressed.

Conclusion

The New Madrid Seismic Zone presents earthquake science with one of its most challenging problems: how to communicate and act on a hazard that is unambiguous in the geological record, significant in the probabilistic hazard assessment, genuinely uncertain in the geodetic data, and chronically under-resourced in preparedness. The 1811–1812 sequence happened. The paleoseismic record shows it happened before. The GPS data does not prove it will not happen again — it shows only that we cannot currently detect the strain accumulation we expected to find, which could mean many things, most of them not reassuring on the timescale of infrastructure planning.

For the 12 million people living in the NMSZ hazard zone — in the Mississippi River cities of St. Louis and Memphis, in the small towns of the bootheel of Missouri and the Arkansas delta, in the communities of western Kentucky and western Tennessee — the relevant question is not whether the GPS debate has been resolved. It is whether the next great earthquake will find their schools structurally sound, their bridges capable of carrying emergency vehicles, their levees intact, and their communities with the institutional knowledge and material reserves to survive weeks of isolation from the infrastructure that normally sustains them. The paleoseismic record says that question will eventually be answered. The current preparedness record suggests the answer is not yet what it should be.