NOAA's current displayed timeline lists September 2026 for preliminary NOAA Atlas 15 precipitation-frequency estimates for the contiguous United States to enter peer review and feedback. That timeline lists published estimates in 2027.
The word that matters right now is preliminary.
NOAA says Atlas 14 remains the existing national standard and authoritative source today. It also says the Montana Atlas 15 pilot was for comparison and feedback, had not completed peer review, and could differ significantly from the final release. There is no responsible Atlas 15 versus Atlas 14 design comparison to make from the national data yet because the national Atlas 15 values have not been released.
But there is a design-management issue worth addressing before the numbers arrive.
The design storm is becoming versioned data. The engineering problem will not be solved by replacing one rainfall depth in a spreadsheet. A preliminary estimate, a published current-conditions estimate, a future adjustment, and a locally adopted criterion can describe the same site without carrying the same authority or serving the same purpose. If the provenance disappears between the source, model, report, and plan set, the number can look precise while its basis is no longer recoverable.
The subset I mean
This argument is narrower than all engineering and narrower than all water work. I mean United States civil, site, transportation, and drainage projects that use NOAA precipitation-frequency estimates as an input to stormwater or hydraulic design.
A precipitation-frequency estimate is not the entire design storm. Duration, temporal distribution, watershed response, areal treatment, loss assumptions, and the governing design method still matter. I use "design storm" in the title because the precipitation estimate travels into that larger constructed input, not because an atlas lookup determines the design by itself.
Some projects use a client-mandated rainfall criterion, a state or municipal table, a federal agency standard, or a project-specific study. Some jurisdictions may incorporate NOAA values directly, while others may translate them into local tables, curves, zones, or transition rules. Those are not exceptions to the provenance problem. They are the reason the provenance problem exists.
NOAA itself draws this boundary. Its Atlas 15 page says Atlas 14 is referenced in engineering design standards and floodplain regulations published by entities outside NOAA. It says users, organizations, and government agencies will have to identify the appropriate bounds for applying Atlas 15 estimates. Its Atlas 14 FAQ tells users who encounter competing precipitation-frequency resources to consult the client or governing regulatory agency about which one to use.
NOAA provides the national precipitation-frequency information. It does not silently rewrite every drainage manual, contract, permit condition, or basis-of-design document that points to it.
One atlas is turning into several distinct inputs
Atlas 15 is planned as two volumes. NOAA says Volume 1 will represent current estimates while accounting for temporal changes in historical observations. Once published, it will supersede Atlas 14. Volume 2 will provide model-based estimates projected into the future by applying adjustment factors to Volume 1.
That creates distinctions an engineering record must preserve.
A value from Atlas 14 is the current authoritative national estimate. A value from the September 2026 Atlas 15 release will be preliminary and presented for peer review, not yet the published replacement. A future value in Volume 2 will be an adjusted projection, not the same type of estimate as the Volume 1 current-conditions value. A value in a city table may be the governing local criterion even when its ancestry is a NOAA atlas.
Those labels cannot be shortened to "NOAA rainfall" without losing information.
This is also not a simple story about a newer number being larger. NOAA has not published the national Atlas 15 estimates, and its pilot materials warn that preliminary data may differ significantly from final data. The defensible position today is not to predict direction or magnitude. It is to prepare for a source system with more states than current practice may be used to recording.
Austin already showed what transition looks like
The previous atlas change left a useful public record.
When Austin revised its Drainage Criteria Manual around Atlas 14, the city did not treat the change as a universal find-and-replace operation. For phased developments whose earlier phases were permitted or constructed using pre-Atlas 14 rainfall criteria, Rule 1.2.2.H addresses the storm drain system within the current phase, applies current rainfall criteria across the drainage area feeding the detention pond or outfall, and requires an overtopping check for the detention pond using that whole-drainage-area basis.
That is the coexistence problem in plain view. Two rainfall bases can both be legitimate within one jurisdiction because approval history, project phasing, and the status of local amendments differ.
I do not know what Austin or any other jurisdiction will do when Atlas 15 is published. The point is not to forecast a local policy. The point is that the last transition produced explicit rules about which basis applied where. An engineering team that preserved only the final depth, and not the approval path that selected it, would be unable to reconstruct that decision later.
The chain of custody belongs with the value
TxDOT's current Hydraulic Design Manual offers a practical starting point. It directs designers using Atlas 14 data to report the location selected for the depth values and a brief explanation in the drainage report or plans. It also distinguishes choices such as depth versus intensity and annual maximum versus partial duration series, and notes that a municipality may state a preference for one series.
Elsewhere, the same manual says hydraulic documentation should identify data sources, evaluate reliability, record decisions and reasons, state limitations, and preserve the design values that govern the work.
That is already a chain-of-custody model. Atlas 15 makes it more important.
For this subset of projects, I would want the rainfall basis to preserve at least six things:
- Source and status. Atlas name, volume, release status, publication or retrieval date, and whether the input is authoritative, preliminary, or project-specific.
- Location. The coordinates or documented selection point, plus the reason that point represents the site or watershed.
- Frequency definition. The selected annual exceedance probability or recurrence convention, duration, depth or intensity form, and time-series basis.
- Governing criterion. The client standard, agency manual, municipal table, permit condition, or written approval that made this input applicable to the project.
- Transformation history. Every step between the retrieved source and the model input, including interpolation, depth-to-intensity conversion, temporal distribution, areal adjustment, local factor, or future adjustment.
- Decision owner. Who selected the basis, who checked it, when it was accepted, and what event would require the team to revisit it.
The last item matters because data can change while a project is still active. A new national publication does not answer whether a permitted phase, a contract already underway, or a model accepted by a reviewing agency should be reopened. That is an engineering and approval decision, not an automatic software update.
Where the provenance has to survive
Putting the source in one calculation note is not enough if every downstream artifact drops it.
The basis-of-design memorandum should name the governing rainfall source and approval basis. The drainage report should carry the selection point, frequency definition, transformations, and limitations. The hydrology model should preserve the input dataset and units in a way another engineer can trace back to the report. The calculation package should show conversions instead of presenting only the resulting coefficient or hyetograph.
The plans should carry enough of the basis to prevent a reviewer from mistaking a locally adopted value for a direct atlas lookup. Hydraulic data sheets and general notes should agree with the report. A transmittal or review response should record any agency direction that changes the basis. Revision logs should say whether rainfall inputs changed, not merely that drainage was revised.
The handoffs deserve the same discipline. Civil to landscape. Site to roadway. Hydrology to hydraulics. Engineer to outside modeler. Designer to independent checker. Consultant to owner. Applicant to reviewing agency. Existing phase to future phase.
At each handoff, a naked number invites a reasonable but dangerous assumption: that everyone means the same storm.
EPA has already removed one comfortable shortcut
There is another sign that this transition will not fit neatly inside old workflows. EPA says it stopped maintaining the SWMM Climate Adjustment Tool on March 18, 2025 and encourages users to use the upcoming NOAA Atlas 15 effort instead.
That does not make Atlas 15 values available today, and EPA's notice does not prove a one-for-one technical replacement for every SWMM-CAT use. It is a transition signal: one federal modeling workflow is already pointing practitioners away from a separate adjustment utility and toward the coming atlas framework.
The consequence is not that every SWMM model should change now. The consequence is that model documentation needs to distinguish a historical precipitation-frequency input from a climate adjustment and, later, from an Atlas 15 Volume 2 future projection. If those all collapse into a filename such as "final rainfall," no reviewer can tell which operation occurred.
The useful work can start before the release
Engineering teams do not need the preliminary values to review their process.
They can identify where rainfall inputs live today, which artifacts repeat them, and which copies lose the source. They can decide who is allowed to replace a governing input and what approval is required. They can add status fields that distinguish preliminary information used for evaluation from published information used for design. They can require the model, report, and plans to name the same basis. They can define how an independent checker verifies the chain from source retrieval through the final design value.
They can also decide what not to automate. A script can retrieve a value, populate a table, or compare two documents. It cannot decide that a new dataset overrides a permit condition, a municipal transition rule, or an engineer's accepted basis of design. That call belongs to the responsible people and reviewing authorities.
The first-pass check is simple to state: can another qualified reviewer reconstruct why this rainfall input governed this design, using the record in front of them?
If the answer depends on asking the original modeler what they remember, the chain of custody is already broken.
What this cannot tell you
This article does not compare Atlas 14 and Atlas 15 values. National Atlas 15 estimates are not yet available, and NOAA says preliminary information is subject to peer review and change. NOAA's page also contains older lower-page prose that still refers to an earlier publication schedule; this article uses the current displayed timeline that lists 2027 for published contiguous United States estimates. It does not predict when a state, city, client, or permitting authority will adopt or translate Atlas 15. It does not say every project should be redesigned when Atlas 15 is published.
The Austin and TxDOT examples are Texas examples, not a national survey of drainage practice. They show how public agencies have documented source selection and transition before. They do not establish what another jurisdiction requires.
My claim is narrower. Once NOAA publishes a current-conditions volume, a future-projections volume, and public review releases along the way, "the NOAA number" will no longer be a sufficient description of a design input. For projects that rely on these estimates, the source, status, governing rule, transformation, and approval history have to travel with the number.
The design storm is becoming versioned data. The engineering record has to become version-aware with it.