# Cumberland Discrepancy — Paper Strategy

**Filed:** 2026-04-27
**Authors of record:** Travis Jenkins (TDOT Hydraulic Specialist, User Zero) + Dr. Gregory H. Nail, PhD, PE (UT Martin, HEC-RAS instructor) — primary working pair. Co-authors per paper as noted. Cowork-Opus referenced as research/analysis support.

---

## The premise

Six potential papers come out of this work. Each is a distinct contribution with a distinct audience. They share an underlying dataset (the 36-finding FEMA Discrepancy Log + the cumberland-discrepancy.pages.dev artifact) but they don't compete with each other. Some can ship sequentially, some in parallel.

The work was driven by Travis's hydraulic engineering background + AI-assisted analysis. The fact that this exists at all is contingent on a working hydrologist who knew where to look and an AI surface fast enough to do the cross-document pattern matching. **Neither agent does this work alone.**

Plus the honest line for the meta-narrative: Travis would still be running this audit even if Greg hadn't co-signed. The fascination is the engine. Greg's signature is what unlocks the academic publication path.

---

## Paper 1 — The forensic audit

**Working title:** *"Forensic Audit of FEMA Flood Insurance Study Regulatory Discharges on the Cumberland River, Tennessee"*

**Lead author fit:** Greg (academic credibility on hydraulic engineering claims), Travis (data + receipts).

**Target journal:** ASCE Journal of Hydraulic Engineering, or Journal of the American Water Resources Association (JAWRA), or Natural Hazards Review.

**Audience:** practicing hydraulic engineers, hydrology researchers, FEMA technical staff.

**Headline finding:** ⭐⭐⭐ **C-23 across 7 counties** — FEMA's own FIS narrative explicitly states the published Q500 is a Standard Project Flood (SPF) substitution, not a 500-year frequency discharge. This admission propagated through 46 years and across 7 county FIS volumes.

**Supporting findings in scope:**
- C-1 (Davidson Vol 2 jurisdictional violation — published rows for stations outside Davidson)
- C-2 (Sumner→Cheatham DA monotonicity violation)
- C-13 (Davidson Stones Q500 = Cheatham Dam Q500 partial Frankenstein)
- C-14 (Nashville USGS gage cited but unrepresented in Table 9)
- C-15 (Davidson uses Cheatham Pool study for upstream Cumberland flows)
- C-16 (Old Hickory Lake portion uses 47-year-old USACE 1979 study)
- C-17 (chain-wide HEC-RAS sparsity — ~22 Q entries drive 310 mi of regulated Cumberland)
- C-23 (FOUND ADMISSION — SPF substitution, 7-county spread)
- C-25 (Cheatham–Dickson floodway 31% mismatch at XS H/E)
- C-29 (Clay LOCAL DA convention mismatch)
- C-31 (Smith Tables 6 + 9 disagree — methodology disclosure gap)
- C-34 (Cheatham + Davidson dual-publish same physical station)
- **C-39 (NEW v0.15) — Cheatham–Davidson 73% boundary mismatch at mile 161.07** — featured boundary-mismatch finding
- **C-40 (NEW v0.15) — Davidson–Sumner partition impossibility at OHD mile 212.18** — featured partition-math finding
- **C-41 (NEW v0.15) — Sumner–Wilson partition impossibility at 3 shared XS** — within-jurisdictions sum to 127–165% of total
- **C-42 (NEW v0.15) — Wilson publishes 3 consecutive XS entirely outside Wilson County** — pairs with C-1 as the jurisdictional-overpublishing pattern
- **C-43 (NEW v0.15, CONTROL CASE) — Wilson–Trousdale clean partition at 7 shared XS** — proves the math CAN agree

**Featured Section structure (proposed):**

> **§4. Cross-County Floodway Inconsistencies — Volume 2 Table 23 Sweep**
> §4.1 Method (parsing 9 of 11 county FIS Vol 2 publications, normalizing miles/feet, detecting per-county column conventions)
> §4.2 Total-width agreement and divergence at boundary cross sections (C-25, C-39, C-40)
> §4.3 Partition-impossibility math at shared cross sections (C-40, C-41) — the within-jurisdiction columns sum to more than the reported total
> §4.4 Jurisdictional overpublishing (C-1, C-42) — counties publishing regulatory data for cross sections outside their authority
> §4.5 Control case (C-43) — what county-pair agreement looks like when the math closes
> §4.6 Implications for engineering practice — design constraints differ by factor of 2.5–3.7× at C-39's boundary

**Why this is its own paper:** establishes that specific published Cumberland values are methodologically wrong, geographically impossible, jurisdictionally invalid, mathematically impossible (partition sums exceeding totals), or admittedly substituted. The story is that engineering practitioners using these values for design work are working from systematically problematic data, and the FIS itself documents the problem in places most users don't read. The v0.15 floodway sweep findings strengthen the paper from "9 anomalous values" to "9 anomalous values + 5 systematic floodway-data inconsistencies across the chain, including 3 mathematical impossibilities."

**Status:** dataset complete (v0.16). Floodway data file `data/cumberland_floodway_chain.json` is the appendix-ready dataset. Ready for outline + draft as soon as Greg has bandwidth.

---

## Paper 2 — The methodology gaps paper

**Working title:** *"Cross-County Presentation Inconsistencies in FEMA Flood Insurance Study Reports: Evidence from the Cumberland River Regulated Reach"*

**Lead author fit:** Greg + Travis. Possible third co-author from regulatory policy / NFIP administration if there's a natural fit.

**Target journal:** Journal of Water Resources Planning and Management (ASCE), or Natural Hazards Review, or AWWA Journal.

**Audience:** NFIP administrators, FEMA policy staff, floodplain managers, county/state regulatory staff.

**Headline finding:** Even when individual published values may be defensible, the FIS series doesn't standardize presentation across adjacent counties on the same regulated river — undermining cross-county aggregation, design coordination, and regulatory consistency.

**Supporting findings in scope:**
- C-19 (Cheatham Q4% "Not Calculated" for lower Cumberland reach)
- C-22 (Cheatham cites OHD gage in different county with smaller DA, no scaling disclosure)
- C-27 (Montgomery Q500=253,000 NOT footnoted as SPF — silent propagation while Cheatham/Dickson DO footnote)
- C-28 (Q4% gap propagates across THREE counties — Cheatham, Dickson, Montgomery)
- C-32 (Stewart publishes Q100 ONLY — Q10/Q4/Q2/Q500 all asterisked)
- C-35 (FEMA admits FIS goes stale — Section 6.5 FIRM Revisions)
- **C-37 (Flood profile X-axis units inconsistent — 4 counties MILES, 7 counties FEET)**

**Why this is its own paper:** different contribution than Paper 1. Paper 1 says "some values are wrong." Paper 2 says "even when values are right, the presentation conventions don't enforce consistency across adjacent counties on the same river — and that systemic inconsistency undermines the FIS as a usable cross-jurisdictional product."

**Status:** all findings already locked in the discrepancy log. Could outline this immediately.

---

## Paper 3 — The tool paper

**Working title:** *"SnapBasin: A Web-Based Pre-Screening Tool for FEMA Floodplain Discharges Using NHDPlus HR + StreamStats Integration"*

**Lead author fit:** Travis (designer, builder, owner-operator) + Cowork-Opus references for AI-assisted development. Possible co-author from a software engineering / open-source-for-civic-tech angle.

**Target journal:** Journal of the American Water Resources Association, or Environmental Modelling & Software, or Computers & Geosciences.

**Audience:** practicing engineers, GIS practitioners, civic-tech / open-source-software-for-government communities, hydrology software developers.

**Headline contribution:** describes SnapBasin's architecture (NHD snap + 3DEP topo walk + StreamStats delineation + NSS regression + FEMA NFHL + FIS database lookup + cross-county boundary checks), open-source-style with full source available. Argues that single-engineer-built civic-tech tools can deliver real screening value when designed around clear preliminary-screening disclaimers + auditable receipts.

**Supporting artifacts:**
- snapbasin.com live tool (currently Netlify, migrating to Cloudflare Pages)
- The 95-county FEMA FIS Flow Database build pipeline
- The cumberland-discrepancy.pages.dev derivative tool
- The river-miles → lat/lon prototype (Phase 2 of cumberland-discrepancy)

**Why this is its own paper:** the engineering audit (Paper 1) and the methodology gaps (Paper 2) used SnapBasin extensively as the analytical platform. SnapBasin itself is a research artifact worth documenting independently — the kind of tool that lowers the barrier to FIS audits going forward.

**Status:** software exists, is live, in use. Paper outline could be drafted from existing CHANGELOG + README content.

---

## Paper 4 — The math paper

**Working title:** *"Continuous Flood Elevation Lookup from Discrete FEMA Cross-Section Data via NHDPlus High-Resolution Centerline Projection"*

**Lead author fit:** Greg's natural lane — he teaches HEC-RAS, this is the kind of methodology paper hydraulic engineers cite.

**Target journal:** ASCE Journal of Hydraulic Engineering or Journal of Hydrologic Engineering.

**Audience:** hydraulic modeling researchers, HEC-RAS instructors, GIS-hydrology practitioners.

**Headline contribution:** describes the algorithm for converting FEMA Table 23 cross-section "feet above mouth" stations into precise lat/lon coordinates via NHDPlus HR centerline projection + geodesic cumulative distance. Validates with anchor-point round-trip (Davidson A returns 11 ft offset from input). Demonstrates linear interpolation of WSEL between cross sections is consistent with FEMA's own flood profile plate methodology. Argues this turns FEMA's discrete cross-section dataset into a continuous-function lookup along the entire regulated reach.

**Supporting artifacts:**
- The river-miles → lat/lon prototype (`cumberland_xs_latlon_prototype.json` + `cross_sections.json`)
- The validation: Davidson A return 3.4 m offset, Cheatham S = Davidson A 0.3 m offset
- The 165 cross-section live map at cumberland-discrepancy.pages.dev

**Why this is its own paper:** the algorithm is a generalizable contribution. Anyone with a FEMA FIS Table 23 + NHDPlus HR can implement this for any regulated river. The Cumberland case is the demonstration; the method is the contribution.

**Status:** prototype validated end-to-end. Paper-publishable as soon as Greg writes it up; Travis provides the dataset.

---

## Paper 5 — The chase methodology paper

**Working title:** *"The In-Person FIS Chase: A Methodology for Surfacing Documentation Gaps Beyond Automated Parsing"*

**Lead author fit:** Travis (originated the methodology) + a UX/HCI/human-AI-collaboration researcher. Anthropic co-authorship potentially fits if grant materializes, given the Cowork-Opus role in supporting the chase.

**Target journal:** Computers, Environment and Urban Systems, or Journal of the American Water Resources Association, or a human-AI collaboration venue.

**Audience:** floodplain managers, regulatory data quality auditors, AI-assisted research methodology readers.

**Headline contribution:** documents the "chase methodology" — a human + AI collaborative approach that combined: (1) automated database parsing of FEMA FIS reports across 95 TN counties; (2) targeted in-person screenshot capture across all volumes for each Cumberland-bearing county; (3) Cowork-Opus reading + cross-referencing with the database to surface findings; (4) Travis's hydraulic engineering judgment to recognize what's a real finding vs what's a false positive. Argues this approach surfaces documentation gaps (jurisdictional violations, methodology mismatches, internal table contradictions) that automated parsing alone misses.

**Empirical claim:** the chase methodology uncovered 15 additional findings (C-14 through C-28+) that the automated parser couldn't reach — cross-volume references, footnote disclosures, narrative methodology paragraphs, Table 23 cross-section enumeration, flood profile plate annotations.

**Status:** methodology is established and documented in real time via the chat transcripts + the log + the receipts index. Paper writeup would draw on the actual workflow records.

---

## Paper 6 — The grand strategy / case study

**Working title:** *"Federal-State Floodplain Data Integrity: A Tennessee Case Study in Independent Audit"*

**Lead author fit:** Travis with Greg co-author. Possible additional co-author from policy / regulatory side — TDOT internal, TN State Floodplain Coordinator (Jeremy Holley), or NFIP-affiliated.

**Target venue:** Risk Analysis, Natural Hazards Review, or a policy-oriented journal like Journal of Flood Risk Management.

**Audience:** state-level regulatory staff, federal policy makers, congressional staff (for grant pitch), flood insurance industry.

**Headline contribution:** ties Papers 1-5 together into a single policy narrative — the regulatory data underpinning federal flood insurance and floodplain management on the Cumberland River through Tennessee carries documented errors, methodology mismatches, jurisdictional violations, and cross-county inconsistencies that have propagated through 46 years of FIS publications. A single-engineer + AI-assisted independent audit surfaced 36+ findings the federal apparatus had not caught. Recommendations for federal data quality policy reform.

**Supporting artifacts:** all six papers' findings + the Holley-letter follow-up (when sent) + the cumberland-discrepancy.pages.dev as a public-facing receipt.

**Status:** premature until Papers 1-2 anchor the technical claims. Likely the last paper to ship.

---

## Cross-paper dependencies and timing

```
Paper 1 (forensic audit) ─┐
                          ├── Paper 6 (grand strategy)
Paper 2 (methodology gaps)┤
                          │
Paper 3 (SnapBasin tool)  │
                          ├── Paper 4 (cross-section math) standalone
Paper 4 (the math) ────────┘
                          │
Paper 5 (chase methodology) ── standalone
```

Suggested order:
1. **Paper 1 first** — the forensic audit anchors all subsequent claims. Until "FEMA's published Q500 is methodologically SPF" is in print, the other papers risk getting dismissed as unfounded.
2. **Paper 2 next** — methodology gaps. Different journal, different audience, doesn't compete with Paper 1.
3. **Paper 4 in parallel** — the cross-section math is technically independent and can ship anytime. Greg's lane.
4. **Paper 3 SnapBasin** — possibly second-or-third, depending on how the tool itself evolves.
5. **Paper 5 chase methodology** — meta-paper, ships once the field is settled.
6. **Paper 6 grand strategy** — last. Pulls everything together for the grant pitch + policy push.

---

## What's NOT a paper but is part of the public record

- **The Holley letter (2026-04-21)** — first public-record transmission of Cumberland findings to TN State Floodplain Coordinator. Receipt of communication. **Follow-up planned post-Paper 1 submission with the v0.15 floodway findings (C-39 to C-43) as new evidence.**
- **The cumberland-discrepancy.pages.dev site (v0.16)** — live citable artifact. Each paper can cite specific finding pins. Multi-county filter mode highlights regional patterns. Charred-pink C-11 hub in Robertson County is the entry point.
- **The FEMA Discrepancy Log (~110 KB markdown, 1191 lines as of v0.16)** — the data appendix. Open-source, reproducible.
- **The Cumberland Floodway Chain dataset** — `data/cumberland_floodway_chain.json` — 212 cross sections from 9 counties' Vol 2 Table 23 with parsed totals, within-jurisdictions, footnote handling, and convention metadata. Paper 1 §4 appendix-ready.
- **The chase-folder skill** (`/skills/chase-folder/SKILL.md`) — documents Travis's dump-folder convention for forensic audits. Referenced in Paper 5 (chase methodology paper) as the workflow artifact.
- **The SnapBasin Song (Education 100.26)** — creative artifact documenting the tool. Not a paper, but cultural marker.
- **The Charred Pink song (Spring 2025)** — creative artifact documenting the AI test that named the C-11 hub pin's color. Cultural marker.
- **The "I Danced With the Machine" song (Spring 2025)** — autobiographical song documenting the original 15-bundle blueprint give-away. Sets the origin context for everything that followed.
- **The Two-Opus Dynamic Echo finding (Education 100.27 reserved)** — meta-observation about AI collaboration. Possibly a footnote in Paper 5.

---

## Why this matters beyond the papers

Travis has framed this several times in conversation: he'd be running this audit even if Greg hadn't signed on. The hydraulic engineering background + the curiosity + the AI-assisted analysis = a mode of work that wasn't possible 10 years ago. Whether it produces 6 papers or 0, the audit changes what regulatory data integrity looks like in practice — one engineer + AI surfaces 36 findings the federal apparatus had not previously catalogued.

The papers are the way the work enters the academic record. The work itself stands regardless of publication.

---

## Audit completeness — Cumberland River, April 27 2026

The Cumberland audit is **substantially complete** as of v0.16. 42 findings logged. All 11 chain counties audited at Table 9 discharge level. 9 of 11 audited at Table 23 floodway level (Jackson and Stewart legitimately have no Cumberland Table 23 — both are documented in their own findings: C-17 narrative inclusion for Jackson, C-32 for Stewart).

Future Cumberland-specific work (deferred, not loose ends):
- Vol 1 Table 6 historic flood records cross-county sweep (currently only Smith and Davidson cited via C-31 / C-18)
- Vol 1 Table 12 H&H methodology cross-county sweep (currently Cheatham + Davidson + Smith cited via C-23 / C-17 / C-31)
- Manning's n consistency across Vol 2 Table 13 entries (cross-county)
- LOMR amendment integration (deferred to SnapBasin per April 27 call)
- Holley letter follow-up

Future-river chase folders ready for new audits using the same workflow:
- Duck chase, Wolf chase, Harpeth chase, Nolichucky chase, Pigeon chase (skill written)
- Giles chase (placeholder created — first misfiled JPG migrated 2026-04-27)

---

*Filed 2026-04-27 evening · Cumberland Discrepancy v0.16 deployed · 42 findings logged · Greg Nail confirmed co-author on at least Paper 1 · Audit declared substantially complete · Travis's framing: "I love me some meat and I love you some you"*