Noetic Logo
All posts
Guides·

The Permit Every Site Plan Ends At Is Being Rewritten

"Revised the definition of Commencment of Construction in Part I.B. to clarify and describe construction support activities consistent with the 2022 United States Envionmental Protection Agency (EPA) National Pollutant Discharge Elimination System (NPDES) CGP."

That is item A in the fact sheet TCEQ (the Texas Commission on Environmental Quality) wrote for the 2023 renewal of TXR150000, the construction general permit, issued under the federal Clean Water Act's stormwater-permitting program, that covers essentially every private construction site in Texas. Both misspellings are TCEQ's. Item B modified the dewatering definition "for clarification and for consistency with the 2022 EPA NPDES CGP." Item M added a requirement to evaluate dewatering controls during dewatering, again "to be consistent with the 2022 EPA NPDES CGP."

Then TCEQ reached for the federal permit three more times in its Executive Director's Response to Comments, using it to fend off commenters. "Response: Consistent with EPA's 2022 CGP, Part IV.F. of this permit includes…" "Response: This section of the CGP is consistent with EPA's 2022 CGP." At least six invocations across two documents, one state agency naming where it got the words. Which is why the federal draft EPA opened for comment this month is the earliest chance a Texas engineer has to shape the next Texas permit.

EPA published the draft 2027 Construction General Permit in the Federal Register on August 3, 2026. Comments close September 17. TXR150000 expires March 5, 2028, and TCEQ opened stakeholder work on its replacement in spring 2026.

The federal comment window sits inside the Texas drafting window. That is the whole argument. The provisions worth reading first are the ones TCEQ copies by choice rather than the ones federal law already dictates, and TCEQ's fact sheet tells you which are which.

Who the federal permit actually binds

EPA is the direct NPDES permitting authority in three states, Massachusetts, New Hampshire, and New Mexico, none of which has an approved state program. EPA's authorization table adds the District of Columbia, most territories (the U.S. Virgin Islands is delegated and is the exception), Indian country nationwide, and certain federal operators.

Texas appears on that table twice, for Indian country and for federal operators. Three federally recognized tribes sit inside the state. The federal CGP does touch ground in Texas directly, on a narrow slice of it, alongside the TCEQ permit that governs everything else.

Everywhere else the federal permit works indirectly, and EPA says so about itself. Its CGP frequently-asked-questions page answers whether state permits have to match the federal one: "Although states may issue permit requirements that differ from EPA's, the Agency recognizes that many state-issued permits follow EPA's permit and content."

Floor, permission, habit

Three separate mechanisms get collapsed into "states follow EPA," and keeping them apart is the difference between reading the draft usefully and reading it as somebody else's paperwork.

The first is a genuine federal floor. 40 CFR 123.25(a) requires state programs to have legal authority implementing a list of NPDES provisions, though "States need not implement provisions identical to the above listed provisions" as long as they establish requirements "at least as stringent as the corresponding listed provisions." That constraint lands on program structure, leaving permit wording alone. The technical floor for construction sites lives in the effluent limitation guidelines at 40 CFR Part 450: Section 450.21(a) lists eight non-numeric requirements binding on every construction stormwater permittee in the country, including minimizing exposed soil, preserving topsoil unless infeasible, and providing "natural buffers around waters of the United States… unless infeasible."

Read Section 450.21(b) twice. Stabilization, covering or seeding exposed soil so it stops washing off the site, "must, at a minimum, be initiated immediately" once earth-disturbing work permanently stops, or temporarily stops for more than 14 calendar days. That 14-day trigger is federal law. The number of days allowed to complete stabilization is "a period of time determined by the permitting authority." Every completion deadline you have ever worked to is a policy choice by EPA or by TCEQ.

The second mechanism is express permission to go past the floor. 40 CFR 123.1(i)(1) lets a delegated state adopt requirements "more stringent or more extensive" than the federal ones.

The third is habit, and it's the only one this article is really about. Copying a pre-vetted federal template is cheaper to draft and easier to defend than writing from scratch. TCEQ marks the boundary itself: fact sheet item L says it "Moved Part III.G. regarding Erosion and Sediment Controls to new Part IV to clarify that these requirements are federal effluent limitations," and elsewhere the same fact sheet ties exactly those requirements to five sections of the federal effluent guidelines in 40 CFR Part 450. Items A, B, and M carry no such citation; they cite EPA's permit instead. That's the tell, the same layering logic that stacks five rulebooks on a single parcel.

What the draft actually changes

EPA's fact sheet accompanying the draft describes the package as small: "the changes EPA is proposing for the 2027 CGP are narrow in scope, targeted at specific issues, and are expected to streamline requirements." It sorts the changes into seven categories: one is nothing but open questions, one is the water-quality-standards rewrite covered in its own section further down, and the remaining five categories, several bundling more than one provision, resolve into eleven adopted items.

On those eleven, the count is six loosening items against five tightening ones. Close to even. Anyone selling this as a deregulatory sweep hasn't counted.

The six that loosen:

  • Perimeter sediment controls at Part 2.2.4 go back to the 2017 wording, covering only perimeter areas with "the potential to receive stormwater from disturbed areas."
  • The arid, semi-arid, and drought-area stabilization exception at Part 2.2.15.b loses its "seasonally dry period" gate and applies year-round.
  • A new frozen-ground and snowfall exception at Part 2.2.15.b.v–vi lets a site "temporarily suspend the need to stabilize the site while the conditions persist."
  • The Notice of Termination at Part 8.2.2.a drops "before" photos and keeps only post-stabilization ones.
  • Part 2.2.9 drops the "native" qualifier from topsoil preservation.
  • Part 2.3.6 stops requiring EPA-specific spill documentation that duplicates what 40 CFR parts 110, 117, and 302 already require.

EPA's stated reason for the perimeter change is money: "to help operators more effectively direct resources towards permit compliance and eliminate expenditures on extraneous stormwater controls."

The tightening items read differently. Almost all of them are EPA fixing EPA. The same Part 2.2.4 that shrinks perimeter-control coverage also deletes a 2022 exemption from the upgradient-of-buffer requirement, which EPA now calls "an error and inconsistent with permit requirements that predated the 2022 CGP." Part 2.2.13.f requires stabilizing a sediment basin before first directing stormwater into it, after EPA inspections "found instances where significant downstream sediment discharges had occurred as a result of operators immediately starting to use the basin after excavation and prior to sufficient stabilization." Geotextiles stop counting as permanent non-vegetative stabilization at Part 2.2.15.c.ii because they degrade in sunlight, and a new Part 2.2 objective statement sets minimizing off-site sediment as the design standard, one EPA can hold against an undersized control.

The best of them is Part 2.2.1: the requirement to minimize exposed soil "was not included in the 2022 CGP," so EPA is adding "the exact language from 40 CFR 450.21(a)(3)" back in. The current federal permit falls short of its own governing effluent limitation guideline. That is the floor being restored, not raised.

Two things don't change, whatever you have read: inspection frequency (Parts 4.2 through 4.4 carry the existing weekly, bi-weekly, and dewatering-day requirements forward untouched) and the 50 NTU dewatering benchmark (NTU measures how turbid, or cloudy, the water is; a lower number means clearer water).

EPA is also asking whether to cut turbidity-monitoring frequency during long dewatering jobs, "whether to reduce the frequency of turbidity monitoring from daily to weekly for sites that will be discharging dewatering water continuously for an extended period of time." That's a request for comment, not draft text. Asking. Electronic submission of the SWPPP (Stormwater Pollution Prevention Plan) alongside the NOI (Notice of Intent) is also still a question, not a requirement.

What these provisions look like when EPA writes the ticket

These are EPA's own enforcement actions, not TCEQ's, but they price exactly the categories of language TCEQ has already copied into TXR150000 once and may copy again. On EPA settlement worksheets, perimeter controls and stabilization deadlines are priced line items.

KM2 Contractors paid $1,020 for the Fagerlie Subdivision in Lake Stevens, Washington, 12.55 acres disturbed, inspected February 13, 2024, discharging to the impaired Pilchuck River: $300 for three sections of silt fence needing maintenance, $600 for one exposed, unworked stockpile near a diverted drainage channel, $120 for a site map missing sampling-station and BMP locations, BMP being the standard term for a stormwater control such as a silt fence or basin.

LGI Homes – Washington paid $4,420 for Earlington (Skyway) Village in Seattle after a November 18, 2024 inspection: silt fences "not secured to posts, completely trenched in, and turned uphill," an eroding area left unworked "greatly in excess of" the two days allowed, a concrete washout "nearly filled to the brim and with less than 12 inches of freeboard," the buffer space left before it overflows. Ten priced line items, none of them exotic, stacking into one number the way constraints compound across disciplines. LGI is publicly traded.

Granite Three paid $3,000 on 45.5 acres in Washougal, inspected January 14, 2025. Nine hundred dollars of that was 15 separate site inspection reports missing the required schedule for remedial actions, at $60 each.

Coverage itself is enforceable. TPWR Developer, CBG Building Company, and Bowman Consulting DC settled on May 8, 2024 for failing to get NPDES coverage at The Parks at Walter Reed in Washington, D.C.: $27,000 plus a minimum $40,000 Rock Creek Park restoration project built around the Hay's Spring amphipod, an endangered species found only in that park.

For scale, Beazer Homes paid $925,000 in December 2010 across 362 sites in 21 states. That is a fifteen-year-old number, and nothing recent approaches it.

The biggest change in the package came from the Supreme Court

Set outside that count entirely is the single biggest change in the package. City and County of San Francisco v. EPA, 604 U.S. 334, decided March 4, 2025 by a 5–4 vote with Justice Alito writing. San Francisco's sewer system, which combines stormwater and sewage in the same pipes, discharges to the Pacific through the Oceanside plant. EPA's 2019 permit renewal barred any discharge contributing to a violation of an applicable water quality standard without saying what the city had to do.

EPA quotes the result against itself in the 2027 fact sheet. The Clean Water Act "does not authorize EPA to include 'end-result' provisions in NPDES permits." Those are provisions that "do not spell out what a permittee must do or refrain from doing but instead make a permittee responsible for the quality of the water in the body of water into which the permittee discharges pollutants."

So "Discharges must be controlled as necessary to meet applicable water quality standards" comes out of Part 3. In its place, Part 3.1 prohibits discharges containing observable floating, settled, or suspended solids, scum, sheen, or debris; discoloration, odor, film, or sheen indicating a chemical release; and visible construction debris, replacing the permit's older, vaguer standard of simple "cloudiness." EPA had already written these indicator conditions into an April 2025 modification limited to Lands of Exclusive Federal Jurisdiction, and the draft extends them to every permittee, on the theory that they "reflect a subset of common physical indicators that States use in their water quality standards."

A site inspector will feel the downstream edits. At Part 4.6.2.c, if one of those conditions shows up in the receiving water but not in the discharge, no corrective action is required. Part 5.1.3 loses "not meeting applicable water quality standards" as a corrective-action trigger and points to Part 3.1 instead. The same change runs through a dozen numbered parts, from Part 1.1 to Part 7.2.

Calling this a loosening gets it wrong. EPA is complying with a case it lost. A permittee can no longer be cited because a receiving water fails ambient standards; someone has to point at a named condition in the discharge. That cuts both ways, as Winstead's read of the draft notes: an observable sheen or discoloration now raises a causation question about whether it came from the site or from upstream. Local rules underneath don't move when the federal trigger narrows, and Austin's impervious cover limits never depended on the federal "end-result" standard the Court just struck down.

Texas sits on one end of a real spectrum

Oregon was asked to do what Texas does, and refused on the record. Commenters on the 1200-C permit, effective December 15, 2025 and running to December 14, 2030, asked DEQ to revise conditions to match EPA's federal CGP. DEQ declined, calling the request "an objection based solely on the cost of compliance" and asserting the same 40 CFR 123.1(i)(1) authority to exceed the federal permit. Oregon also layers a fixed distance trigger for work near waters listed as impaired for turbidity or sedimentation on top of the federal buffer language.

California never joined the federal cycle at all. Its current permit, Order WQ 2022-0057-DWQ, adopted September 8, 2022 and effective September 1, 2023, runs numeric turbidity and pH thresholds California has kept since 2010, a framework EPA has never adopted nationally. There is no federal numeric turbidity limit for those numbers to sit above, and the 2027 draft sets no general turbidity benchmark for ordinary discharges. Californians call their permit the "2022 CGP" too, a coincidence of nicknames rather than a relationship to EPA's.

Writing numbers that specific draws a challenge. The California Building Industry Association, the Building Industry Legal Defense Foundation, and the California Business Properties Association sued over Order No. 2009-0009-DWQ, and on December 27, 2011 the Superior Court issued a judgment and writ of mandamus. The State Water Board's own amendment order records what it won and what it lost: the court "ruled in favor of the State Water Board on almost all of the issues the petitioners raised."

What it lost was narrow. The court struck the numeric effluent limitations for pH (6.0 to 9.0) and turbidity (500 NTU) that "applied only to Risk Level 3 and LUP Type 3 construction sites," plus two ancillary requirements tied to those limits, because the Board "did not have sufficient BMP performance data to support" them. Two other numeric categories were never touched: the separate limits set for active treatment systems, which had performance data behind them and survived intact, and the numeric action levels at other risk levels, a lower tier that flags a response rather than setting an enforceable limit, which were already in the 2009 order and were never challenged at all. California kept almost everything except the two numbers it couldn't back with data.

Washington shows that the spectrum runs provision by provision. Its permit, reissued November 19, 2025 and effective January 1, 2026, keeps turbidity oversight broader and tighter than the federal permit, which sets no general turbidity benchmark for ordinary discharges at all. The same permit narrowed its jurisdictional language to exclude "any federal operator or Lands of Exclusive Federal Jurisdiction," a change the Washington Stormwater Center says is "aligning the state permit with recent EPA terminology updates." Diverge on substance, converge on definitions.

Maryland is the other flavor of convergence, and the least comfortable one. Its prior general permit, 14-GP, expired December 31, 2019 with no replacement ready, EPA intervened in May 2020, and the replacement, 20-CP, didn't take effect until April 1, 2023. Catching up, Maryland picked up construction-and-development (C&D) rule elements its prior permit had never implemented. Thin programs converge harder than thick ones, late and under pressure.

What predicts which end a state lands on is capacity. California, Oregon, and Washington have mature, well-staffed water-quality programs and the appetite to defend original language. Texas and Maryland reach for EPA's wording wherever the effluent guidelines leave room.

The two timelines overlap right now

The 2022 federal permit expires February 16, 2027. EPA's original 30-day comment period would have closed September 2, 2026; it was extended to September 17, and the docket is EPA-HQ-OW-2025-0760.

On the Texas side, TXR150000 took effect March 5, 2023. TCEQ held a stakeholder meeting on June 9, 2026 on preliminary proposed changes for the 2028 renewal and took informal comments through June 23 under the subject line "2028 CGP Renewal." That window has already closed. The one still open, the only one left where a Texas engineer's comment can still reach the language TCEQ drafts from, is EPA's, through September 17. EPA's 2022 CGP took effect February 17, 2022; TCEQ's took effect just over twelve months later. The fact sheet shows what TCEQ did with those twelve months.

The comment file is thinner than the stakes deserve. NAHB (the National Association of Home Builders) raises no objection to the observable-indicator shift, the year-round arid-area exception, or the winter-weather flexibility. It also raises two points this piece hasn't otherwise covered: it presses for "minimize" rather than "prevent" on visible dewatering turbidity, and says it has previously raised concerns about a 5-acre stabilization threshold and may recommend further changes before the permit is final. The SBA Office of Advocacy posted a notice on August 4 telling small construction firms to read the notice and comment. No major environmental group had filed a public position on the draft as of mid-August.

The one thing to do before September 17

Sort every provision the way TCEQ's own fact sheet sorts them. If it traces to 40 CFR 450.21, it is a floor, and a comment will not move it much. If it is wording EPA chose, the kind TCEQ copied and defended at least six times in 2023 already, a comment has purchase. Start with Part 2.2.4 on perimeter controls, Part 2.2.15 on stabilization timing, Part 3.1 on the new observable conditions, and Part 4.6.2 on what an inspector has to see before corrective action attaches.

If nobody in Texas comments, the next TXR150000 still gets written. It just gets written from a template that homebuilder associations shaped and Texas site engineers did not. The federal docket, EPA-HQ-OW-2025-0760, closes September 17, 2026, and the Texas permit that will govern from 2028 to 2033 is on a drafting table right now.