What if the Worst Drought Metric is Wrong?
A collaboration between Lewis McLain & AI
Texas is about to spend $174 billion on a bet about the weather. The bet is priced against seven dry years from the Eisenhower administration. What happens if that is the wrong benchmark—in either direction?
Every five years, Texas produces one of the most ambitious infrastructure documents in America: the State Water Plan. The Draft 2027 State Water Plan, released for public comment in April 2026, recommends roughly 6,700 water management strategies and 3,000 projects at a capital cost of $174 billion—more than double the $80 billion price tag of the 2022 plan. Voters have already begun writing the checks: in November 2025, Proposition 4 passed, dedicating up to $1 billion a year in sales tax revenue—as much as $20 billion over two decades—to the Texas Water Fund.
Nearly every number in that plan rests on a single foundational assumption, written into the state’s planning rules: the future will be no drier than the past. Specifically, no drier than the worst drought in the roughly 130-year instrumental record—the “drought of record.”
It is worth asking a question the plan itself never asks out loud:
What if that metric is wrong?
It could be wrong in two directions. It could be too strict—an expensive over-preparation for a freak event that may never repeat, in which case Texans are overpaying by tens of billions of dollars. Or it could be too lenient—a benchmark that tree-ring scientists, the state climatologist, and one very thirsty coastal city all suggest is already obsolete.
The interesting thing is that we can put rough numbers on both possibilities. This essay tries to do that.
The Metric: Planning for the Worst Thing That Ever Happened
Texas water planning law requires the state’s 16 regional planning groups to calculate how much water every city, utility, irrigator, and factory would have “during drought of record conditions”—and then to find strategies to close any gap.
The planning rules in 31 Texas Administrative Code Chapter 357 require regions to evaluate existing supplies that are “legally and physically available” for use during the drought of record, using the state’s official water availability models.
For most of Texas, the drought of record is the monster of 1950–1957—seven consecutive years of failed rains that, as the Texas Water Development Board puts it, remains the worst statewide drought in Texas history in both duration and intensity. It is the drought that killed farms, emptied small towns, and prompted the creation of the modern water-planning system in the first place.
The mechanics matter here.
The drought of record is not merely a historical reference. It is the input to a specific calculation called firm yield: the maximum amount of water a reservoir could have delivered every year, without fail, through a repeat of that drought.
Firm yield is what a water supply is “worth” in the plan. Change the benchmark drought, and every reservoir in Texas is suddenly worth a different amount of water—on paper, overnight, without a drop of rain falling or failing to fall.
As Matt Nelson, the Texas Water Development Board’s deputy executive administrator of planning, described the design philosophy to the Texas Tribune, the plan contemplates supplying enough water during the worst drought that has ever occurred in each region of the state, without restricting water use to the public—even if every region experienced its worst drought simultaneously.
Under that standard, the Draft 2027 State Water Plan finds Texas short 3.6 million acre-feet per year in 2030, growing to 5.8 million acre-feet per year by 2080 if nothing is built.
The 2022 plan described the equivalent 2070 shortfall as more water than Lake Livingston, Amistad Reservoir, and Lake Travis can hold combined when full.
The stakes of doing nothing are stated plainly. The Texas Water Development Board estimates that a severe drought without new supplies could leave about one in four Texans without enough water, with economic losses of as much as $177 billion a year within 50 years.
The drought of record, therefore, is not a trivia answer.
It is the denominator of the entire $174 billion program.
Direction One: What if the Standard Is Too Strict?
Here is the version of the question a skeptical taxpayer might ask:
The 1950s drought was, by definition, an outlier—the single worst stretch in the recorded history of the state. Why size a permanent, multi-hundred-billion-dollar infrastructure program to the outlier?
What would happen if Texas planned to the second-worst drought instead? Or the third?
For most of the state, the rankings are approximately these:
- The drought of 1950–1957.
- The drought of 2011, the driest twelve consecutive months ever measured statewide, within the broader 2010–2015 dry period.
- The drought that ended in 1918, which held second place for nearly a century until 2011 surpassed it.
The Texas Water Development Board has never published this counterfactual, so the following figures are the author’s estimates, built from the plan’s own arithmetic.
The benchmark drought mainly governs surface water—the firm yields of the state’s nearly 200 major reservoirs and the reliability of river diversions. Together, those sources account for roughly 7 million acre-feet per year of drought-rated supply.
Groundwater availability, which provides about half of the state’s supply, is governed primarily by aquifer depletion, desired future conditions, permits, and regulatory limits. It would change much less under a different drought benchmark.
Shorter, sharper droughts such as 2011 or 1917–1918 place less stress on multi-year reservoir storage than seven consecutive dry years. Firm yields calculated against those shorter events would therefore be meaningfully higher.
We also have a real-world calibration for how much one “rank step” between droughts may be worth.
When the 2008–2016 drought surpassed the 1950s drought on the lower Colorado River, the Lower Colorado River Authority reduced the firm yield of the Highland Lakes from approximately 600,000 acre-feet per year to about 500,000 acre-feet per year.
One step in the drought rankings changed the yield of a single reservoir system by roughly 17 percent.
Apply that scale statewide, and the estimates look something like this:
| Benchmark | Reduction in Identified Need | Approximate Gallons | Estimated Capital Avoided |
| Second-worst drought, 2011 | 0.7–1.5 million acre-feet per year, or 12–26% of the 5.8 million acre-foot gap | 230–490 billion gallons per year | $16–$34 billion, plausibly as much as $45 billion |
| Third-worst drought, 1917–1918 | 1.0–2.2 million acre-feet per year, or 17–38% of the gap | 330–720 billion gallons per year | $23–$50 billion, possibly more |
The dollar figures scale from the draft plan’s own average of roughly $23,000 in capital cost for each acre-foot per year of new supply: $174 billion buying approximately 7.6 million acre-feet per year of strategies by 2080.
The upper ends reflect the likelihood that the projects eliminated by a relaxed standard would be the most expensive marginal projects—new reservoirs, long-haul pipelines, and seawater desalination facilities.
The honest answer to the skeptic, therefore, is yes.
The drought-of-record standard is plausibly “costing” Texas somewhere between $20 billion and $50 billion in infrastructure that a second- or third-worst benchmark might deem unnecessary. It may also be responsible for roughly one-quarter to one-third of the identified water gap—a few hundred billion gallons a year.
But notice what that framing conceals.
Those billions would not truly be saved. They would be wagered.
A plan sized to the second-worst drought delivers exactly its promised reliability until the day a 1950s-scale drought returns—an event that has already happened once in the instrumental record and, as the evidence suggests, many times before it.
At that point, the unbuilt gap reappears not as a line item in a state plan but as empty intake pipes.
The relevant number would stop being $30 billion in avoided capital and become as much as $177 billion per year in economic damages.
The drought-of-record standard is, in effect, an insurance policy. Arguing that it is too strict is arguing that Texas should self-insure against a catastrophe it has already experienced.
That might still be a defensible argument—if the evidence suggested that the 1950s represented a freak upper boundary.
The evidence points in the opposite direction.
Direction Two: What if the Standard Is Too Lenient?
The Tree Rings Say the Record Is Too Short
The instrumental rainfall record in Texas extends only to the 1890s.
To see further into the past, scientists read tree rings—and what they find is uncomfortable.
In a landmark 2011 study published in the Texas Water Journal, Malcolm Cleaveland, Thomas Votteler, David Stahle, Richard Casteel, and Jay Banner reconstructed drought conditions across Texas back to the year 1500 using baldcypress, Douglas-fir, and post oak chronologies.
Their conclusion was direct: several extended droughts of the past were longer or more intense than the 1950s drought. Severe, decadal-scale droughts had occurred in Texas at least once a century since the 1500s.
They concluded that the continued use of the 1950s drought as a worst-case planning scenario was questionable.
The deeper paleoclimate record is worse.
Tree-ring networks across North America document a 16th-century megadrought described by researchers as the most severe and sustained drought event of the last 500 years—and perhaps the last 1,000 years when looking back to A.D. 1200.
That drought persisted for roughly 40 years in some locations. It extended from Mexico across the American Southwest and into Texas. It coincided with catastrophic epidemics and the failure of early European colonies.
The drought of the 1950s lasted seven years.
The paleoclimate record contains droughts four and five times that long.
In statistical terms, the “drought of record” is the worst draw from a sample of only about 130 years. The full distribution, visible in tree rings, contains substantially worse outcomes.
Planning to the sample maximum and calling it the worst case is not necessarily conservatism.
It may simply be an artifact of a record that is too short.
One River Basin Has Already Lived It
This is not merely hypothetical.
On the lower Colorado River, the drought that began in 2008 surpassed the 1950s event and became the new drought of record for the basin. Six of the ten lowest inflow years into the Highland Lakes occurred after 2008.
The Lower Colorado River Authority responded by reducing the firm yield of the Highland Lakes by about 100,000 acre-feet per year—from approximately 600,000 to 500,000 acre-feet.
Every downstream plan based on the older number had, in hindsight, been planning against the second-worst drought without knowing it.
Texas planning rules now explicitly recognize this possibility. When updated hydrological models identify a new regional drought of record, future plans must absorb it.
The benchmark is therefore not a fixed constant of nature.
It is a low-water mark that keeps getting broken.
The 2011 drought displaced 1917–1918 as the second-worst statewide event. The 2008–2016 drought surpassed the 1950s in the Colorado River basin.
Each new record reveals that the previous “worst case” was not the worst.
Corpus Christi Is the Case Study Nobody Wanted
Then there is Corpus Christi.
During 2025 and 2026, the city came closer to an outright water-supply failure than almost any large American city in recent memory.
Inside Climate News summarized the problem bluntly: unrealistic assessments of the region’s reservoir system helped turn drought conditions into an emergency.
The region’s own 2026 water plan acknowledged that a repeat of the drought of record would already produce an immediate shortage for the City of Corpus Christi, the area’s major water provider.
That shortage could grow if a new drought of record occurred.
That is exactly what began to happen.
The city’s water officials watched reality outrun the model. The region was rapidly approaching drought-of-record conditions. The reservoirs were declining faster than prior calculations had predicted, during a period of intense heat and prolonged dryness.
Robert Mace, executive director of the Meadows Center for Water and the Environment at Texas State University, drew the broader lesson.
Beginning in the 1990s, something appeared to change in the weather of South Texas. Droughts became hotter and longer. In addition, the tree-ring record showed periods drier than anything in the instrumental record.
His conclusion was simple:
Communities should plan for droughts worse than the drought of record.
The Physics Have Changed Beneath the Benchmark
Even a perfect repeat of the rainfall totals from the 1950s would no longer produce the same amount of usable water.
That is one of the most important weaknesses in the current metric.
It may hold rainfall constant while failing to hold temperature constant.
Higher temperatures dry soils more quickly. As soils become hotter and drier, a smaller share of rainfall reaches rivers, streams, and reservoirs as runoff.
Hotter and longer summers also remove more water from reservoir surfaces through evaporation.
The state climatologist’s Extreme Weather in Texas report, produced with Texas 2036, projects that by 2036 Texas will average about 3 degrees Fahrenheit warmer than the late-20th-century norm. The state may experience approximately twice as many 100-degree days as the 2000–2018 average, along with roughly 7 percent more summer evaporation.
On broad, shallow Texas reservoirs, evaporation is not a rounding error.
It is one of the largest uses—or losses—of stored water in the state.
Yet the state’s planning models remain heavily backward-looking.
Surface water is among the water supplies most vulnerable to climate change. But incorporating explicit climate projections into the official planning framework has proven politically and technically difficult.
The institutional response is that the five-year planning cycle continually updates the models as conditions change.
That defense has merit—but it also carries a dark implication.
A benchmark that updates only after a catastrophic drought breaks the old record is a system that learns by failing.
Lake Travis had to fall toward historic lows before the Highland Lakes were repriced.
Corpus Christi had to approach a day-zero emergency before its yield assumptions were publicly described as unrealistic.
The metric may be self-correcting, but it is self-correcting in the way a smoke alarm that sounds only after the fire has spread is self-correcting.
Meanwhile, demand is not waiting politely.
Texas data centers already consume an estimated 25 billion gallons of water annually. Some projections suggest consumption could rise dramatically over the next several years.
Whether that growth is fully reflected in long-term regional demand forecasts remains uncertain.
The Nonlinearity Trap
Why “25 Percent Worse” Does Not Mean 25 Percent Less Water
Suppose the tree rings and climatologists are right and Texas eventually faces a drought meaningfully worse than the 1950s—perhaps one with inflows 25 percent lower.
The natural intuition is that available supplies would decline by about 25 percent.
That intuition is wrong.
Reservoir yield can fall faster than inflow.
Evaporation does not decline when inflows decline. During a hotter drought, evaporation may increase.
Multi-year carryover storage—the entire purpose of a major reservoir—is precisely what a longer drought exhausts.
Junior water rights on fully appropriated rivers may not decline gradually. In the most severe conditions, they may be reduced to zero.
On the Rio Grande, researchers have estimated that legal water rights may total as much as twice the volume of water physically available in the river.
As Samuel Sandoval Solís of the University of California, Davis, has observed, dams do not create water. A reservoir without sufficient rainfall and runoff can become an expensive monument to assumptions that no longer hold.
Using the plan’s own figures, a drought with inflows approximately 25 percent worse than the benchmark could plausibly reduce firm yields by 35 to 50 percent.
That would remove approximately 2.5 million to 3.5 million acre-feet per year from the state’s drought-rated surface-water supply.
That is roughly 800 billion to 1.1 trillion gallons every year.
The projected 2080 gap would grow from 5.8 million acre-feet per year to somewhere between approximately 8 million and 9.3 million acre-feet per year.
Here is the detail worth sitting with:
Because the $174 billion program is sized to the 1950s benchmark, even if every one of its approximately 3,000 projects were built on schedule, Texas could still be 2.5 million to 3.5 million acre-feet per year short during such a drought.
Closing that residual gap from the expensive end of the supply curve—using seawater desalination, long-distance pipelines, and other high-cost strategies—might cost approximately $25,000 to $40,000 for each acre-foot per year of additional capacity.
That implies another $60 billion to $140 billion in construction.
The real program cost could rise from $174 billion to somewhere between approximately $235 billion and $315 billion.
The asymmetry between the two kinds of error is now visible.
If the metric is too strict, Texas may overspend by $20 billion to $50 billion on projects that provide real, though potentially redundant, reliability.
If the metric is too lenient, Texas may face shortages measured in trillions of gallons, economic damages of as much as $177 billion per year, and a supplemental construction bill larger than the estimated cost of over-preparation.
When the consequences of being wrong are that lopsided, the rational bias runs toward a stricter benchmark—not a weaker one.
What Would a Better Metric Look Like?
None of this means the drought-of-record standard was a mistake.
For the Texas of 1957, it represented a triumph of empiricism:
Plan for the worst event ever observed.
The problem is that “ever observed” turned out to be a moving target, viewed through a narrow historical keyhole, in a climate that no longer sits still.
Hydrologists call the broken assumption stationarity—the idea that the statistical characteristics of the past will continue to describe the future.
The field’s verdict was summarized in the title of a famous 2008 Science paper:
“Stationarity Is Dead: Whither Water Management?”
Several possible replacements are already emerging within the Texas planning system.
Safe Yield
Plan to the drought of record plus a margin—perhaps one or two years of additional demand held in reserve.
Region C, which includes the Dallas–Fort Worth area, has long incorporated a version of this approach. Some Texas regions already plan for conditions worse than the official drought of record.
Corpus Christi also nominally adopted a safe-yield approach. Its crisis demonstrates that a safety margin works only when the underlying firm-yield numbers are realistic.
Drought-Worse-Than-Record Scenarios
Since rule changes adopted after the 2011 drought, regional planning groups may examine how their plans would perform under droughts worse than the historical record.
At present, this is often treated as a narrative or sensitivity exercise.
It could instead become a sizing requirement.
The state could require each region to demonstrate which supplies would fail, when they would fail, and how much additional capacity would be required under defined worse-than-record conditions.
Paleo-Informed Benchmarks
The tree-ring reconstructions already exist.
They are peer-reviewed, date-specific, and directly relevant to the question of long-duration drought.
Instead of planning to the worst drought since approximately 1895, Texas could establish a benchmark based on the worst reconstructed drought since 1500.
Such a standard would capture the once-a-century, decadal-scale droughts that the instrumental record misses.
The Cleaveland team’s evidence has been available in the Texas Water Journal since 2011.
Climate-Adjusted Hydrology
Texas could take the historical drought of record and rerun it at projected future temperatures.
The state could model the 1950s rainfall pattern using the evaporation rates, soil-moisture conditions, and runoff relationships expected in 2036, 2050, or 2080.
This approach would preserve the political simplicity of a historical benchmark:
“We are planning for something that already happened.”
At the same time, it would correct the physical dishonesty of pretending that the same rainfall would produce the same water supply under a warmer climate.
Probabilistic Planning
The most ambitious option would be to abandon the single-scenario benchmark entirely.
Texas could instead size its infrastructure around an explicit reliability target—perhaps a 1-in-100 annual probability of system failure.
Flood planners, engineers, insurers, and some western water agencies already use probability-based approaches.
This would make the insurance-policy nature of water planning explicit.
Legislators and citizens could debate how much reliability they are willing to purchase rather than inheriting the answer from a drought that occurred during the Eisenhower administration.
The Metric Is Not a Fact. It Is a Bet.
Here is the strange truth at the bottom of the question:
The drought of record was never merely a measurement.
It is a policy choice wearing a measurement’s clothes.
It is a decision about how much catastrophe insurance Texas should purchase, expressed as a historical statistic so that no one has to defend the policy choice directly.
Asked plainly—what if the metric is wrong?—the numbers provide a reasonably clear answer.
If it is wrong because it is too strict, Texas might spend $20 billion to $50 billion on projects that later appear redundant.
That is real money. It is roughly comparable to the entire amount Proposition 4 could dedicate to water infrastructure over two decades. It deserves serious scrutiny.
But the infrastructure would still exist. It would still provide redundancy, flexibility, and reliability.
If the metric is wrong because it is too lenient, Texas could face shortages measured in trillions of gallons, economic losses measured in hundreds of billions of dollars, and the discovery that a city’s water mathematics was fiction at precisely the moment those calculations were needed most.
Zapata County, the Highland Lakes, and Corpus Christi have each provided a preview.
The tree rings, the thermometers, and a repriced river basin all suggest that the second kind of error is not a remote tail risk.
Severe, long-duration droughts have occurred in Texas at least once a century for approximately five hundred years.
And the current century is still young.
The drought of the 1950s earned its title honestly.
But “the worst drought on record” was always a statement about the length of the record—not necessarily about the limits of drought.
Texas is betting $174 billion that the record is long enough.
The safest thing that can be said about that bet is the same thing the scientists said fifteen years ago:
It is questionable—and the direction of the error, if there is one, is unlikely to be in Texas’s favor.
References
Primary Planning Documents and Data
- Texas Water Development Board, Draft 2027 State Water Plan—Water for Texas, Phase 1, April 2026.
- Texas Water Development Board, 2027 State Water Plan portal.
- Texas Water Development Board, 2022 State Water Plan—Water Supply Needs chapter.
- Texas Water Development Board, Regional and State Water Planning Rules, 31 TAC Chapter 357 reference pamphlet.
- Texas Water Development Board, Regional Water Planning Frequently Asked Questions.
- Texas Water Development Board, Drought in Texas: A Comparison of the 1950–1957 and 2010–2015 Droughts.
Paleoclimate and Drought Science
- Cleaveland, Malcolm K.; Votteler, Thomas H.; Stahle, David K.; Casteel, Richard C.; and Banner, Jay L. “Extended Chronology of Drought in South Central, Southeastern, and West Texas.” Texas Water Journal, Vol. 2, No. 1, pp. 54–96, 2011.
- Stahle, David W., et al. “Tree-Ring Data Document 16th Century Megadrought Over North America.” Eos, Vol. 81, No. 12, 2000.
- Milly, P.C.D., et al. “Stationarity Is Dead: Whither Water Management?” Science, Vol. 319, pp. 573–574, 2008.
- U.S. Geological Survey, A Historical Perspective on Precipitation, Drought Severity, and Streamflow in Texas During 1951–1956 and 2011, 2013.
- Archfield, Stacey A., and Vogel, Richard M. “Reliability of Reservoir Firm Yield Determined from the Historical Drought of Record,” 2005.
- Nielsen-Gammon, John, et al., and Texas 2036, Extreme Weather in Texas, 1900–2036.
Reporting and Analysis
- Erin Douglas, “Texas’ Plan to Provide Water for a Growing Population Virtually Ignores Climate Change,” Texas Tribune, October 31, 2022.
- Dylan Baddour, “Texas’ Refusal to Plan for Climate Change Created a Crisis in Corpus Christi,” Inside Climate News, June 25, 2026.
- Sierra Club Lone Star Chapter, “It’s Official: The Draft 2027 State Water Plan Is Out for Public Comment,” April 28, 2026.
- Community Impact, “Cost to Prevent a Texas Water Crisis Has More Than Doubled to $174B, State Says,” May 20, 2026.
- Texas Water Newsroom, “What to Know About the Texas Drought” and “Planning for Enough Water Supply During Drought.”
- Lower Colorado River Basin Coalition, “LCRA: Current Drought Worst on Record for Central Texas,” February 25, 2015.
- Texas Tribune, “What to Know About Texas’ Looming Water Crisis,” March 13, 2025.
- Texas Water Foundation, “Voters Approve Prop 4—Dedicating $20B to Water Infrastructure,” November 2025.
- Texas+Water, Texas State University, “outlook+water: 89% of Texas Is in Drought,” April 7, 2026.
- Texas Comptroller of Public Accounts, “Texas’ Water Demands Could Outpace Supply in Parts of Texas by 2070,” September 2023.
Note on the Counterfactual Estimates
The figures in the second-worst and third-worst drought benchmark table, along with the 25-percent-worse drought scenario, are the author’s estimates.
They are derived from published Texas Water Development Board totals, including the projected 5.8 million acre-foot 2080 shortage and the approximately $174 billion cost of strategies producing about 7.6 million acre-feet per year of supply.
They also incorporate the Lower Colorado River Authority’s approximately 17 percent firm-yield reduction following the repricing of the Highland Lakes and the standard nonlinear behavior of reservoir firm yield under reduced inflows, extended drought duration, and higher evaporation.
The Texas Water Development Board has not published an official statewide sensitivity analysis showing how the identified water-supply gap would change under second-worst, third-worst, paleo-drought, or climate-adjusted benchmarks.
One acre-foot equals approximately 325,851 gallons.
