# FEMA FIS Discrepancy Log

Running log of published FIS flow/elevation errors found during SnapBasin parse
and manual review. Each entry = one anomaly in the published FEMA document.

**Format:** County | Stream | Location | Published | Corrected | Reason

---

## April 23, 2026 — Giles County (4 fixes)

- **Giles** | **Elk River** | just upstream of confluence of Richland Creek (DA=1,296 sq mi)
  - FIS Table 5 published: Q100 = **6,900 cfs**
  - Corrected to: **69,000 cfs** (missing zero)
  - Sanity check: Q50 = 60,000 and Q500 = 90,000 on same row; adjacent Baugh Road row
    at DA=1,268 shows Q100 = 68,000. 6,900 is physically impossible.

- **Giles** | **Robertson Fork Creek** | ~1.3 miles downstream of confluence with Lynn Creek (DA=50.13)
  - FIS Table 5 published: Q100 = **53,563 cfs**
  - Corrected to: **15,563 cfs** (leading "1" misprinted/OCR'd as "5")
  - Sanity check: Q500 = 20,377 on same row — Q100 cannot exceed Q500. Upstream
    rows at DA=29 and DA=26 show Q100 in the 11-12k range, which interpolates
    cleanly to ~15k at DA=50.

- **Giles** | **Robertson Fork Creek** | ~1.1 miles downstream of confluence with Lynn Creek (DA=48.63)
  - FIS Table 5 published: Q100 = **51,970 cfs**
  - Corrected to: **15,970 cfs** (same "1" → "5" artifact)
  - Sanity check: Q500 = 20,049 on same row.

- **Giles** | **Robertson Fork Creek** | ~0.6 mile downstream of confluence with Lynn Creek (DA=47.17)
  - FIS Table 5 published: Q100 = **50,404 cfs**
  - Corrected to: **15,404 cfs** (same "1" → "5" artifact)
  - Sanity check: Q500 = 19,722 on same row. Three consecutive bad rows + the same
    pattern = systematic print/OCR error, not data.

---

## Earlier finds

- See `SnapBasin_Case_001_Davidson_Cumberland_Label_Shift.docx` — Davidson/Cumberland
  confluence label shift that we called FEMA about on Tuesday.

---

*Log format: keep it short. County, stream, location, published value, corrected value,
1-line reason. More detail lives in the JSON `note` field or in a dedicated case study
docx when it rises to "FEMA phone call" level.*

---

## April 23, 2026 — Batch of 6 counties (Hamblen, Hancock, Hardin, Haywood, Henderson, Henry)

### HENRY — confirmed FIS errors

- **Tennessee River (Kentucky Lake)** | At upstream county boundary (DA=*)
  - Published: Q10=284,800, Q50=306,900, **Q100=32,400**, Q500=402,500
  - Corrected to: **Q100=325,400 (likely)** — verify against FIS PDF
  - Sanity check: downstream-adjacent rows show Q100=324,100 and Q100=323,400. Upstream Q100 should be slightly higher than both — "32,400" is missing a digit. Pattern of increasing Q10 (279k → 281k → 284k) supports ~325k-ish for upstream Q100.
  - ✓ Confirmed by Travis 2026-04-23 via SnapBasin review.

- **Bailey Fork Creek Tributary Two** (Manning's Table 2)
  - Published: Overbank n = 0.030 – **0.470**
  - Issue: Max overbank n of 0.470 is ~3x the typical high end (0.150). Could be dense brush/jungle but unusual. Possible typo for 0.170 (one digit off). Verify FIS.

### False positives — investigated, found to be legitimate hydrology (not errors)

- **HAMBLEN Sinking Creek "at mouth" DA=0.96** — Verified via SnapBasin 2026-04-23.
  DA=0.92 sq mi confirmed at the snap point. Sinking Creek is a karst stream (3 sinkholes
  in Morristown per FIS narrative) and its "mouth" discharges into an unnamed stream/creek.
  Non-monotonic DA values across the FIS table are from subsurface connections between
  reaches, not a data error. Resolved ✓

- **HENDERSON Beech River at Beech Lake Dam — Q10=Q50=320, Q100=350, Q500=2,100** —
  Verified via SnapBasin 2026-04-23. Basin has 2 NID-registered dams totaling 15,500
  ac-ft storage (Beech Dam 1963 High hazard + Saddle Dam No. 2). Regulated-pool pattern
  is textbook: spillway holds small/frequent storms; only the 500-yr overtops and
  releases the 2,100 cfs pulse. Not a FIS error. Resolved ✓

---

## April 23, 2026 — Batch of 5 counties (Hickman, Johnson, Lake, Lewis, Loudon)

### LOUDON — likely FIS error

- **Clinch River** | About 2.3 miles upstream of confluence with Hope Creek (DA=3,330)
  - Published: Q10=32,796, Q2=42,670, **Q1=66,067, Q0.2=63,585**
  - Corrected to: **Q1=46,067** — ✓ agreed by Travis 2026-04-23. Adjacent Melton Hill Dam row (DA=3,340) shows Q1=46,200, Q0.2=65,000 — our row should match that pattern. "6" misprint/OCR for "4" in the thousands place.
  - Value updated in fis_flows.json.

- **Steekee Creek** (Manning's Table 4)
  - Published: Overbank n = **0.06 – 0.015**
  - Issue: max (0.015) < min (0.06) is impossible.
  - Corrected to: **0.06 – 0.15** — missing a "1" before the "5" in the FIS print. Matches every other stream in Loudon Table 4 which tops out at 0.14-0.175.
  - Value applied in manning_n_tn.json; flagged with note.

---

## April 23, 2026 — Final batch of 10 counties (Marshall, Maury, Morgan, Overton, Pickett, Scott, Van Buren, Warren, Washington, White)

### WARREN — ⚠⚠ CONFIRMED FIS COPY-PASTE ERROR (most egregious so far)

- **Charles Creek At State Route 56** — FIS-published DA=132, Q1=35,000
  - **SnapBasin-delineated DA at this location: 36.4 sq mi** (click at 35.71883°N, 85.74164°W)
  - **Hickory Creek At State Route 55** — FIS-published DA=132, Q1=35,000
  - **SnapBasin-delineated DA at that location: 131 sq mi** — matches FIS
  - **Diagnosis: the Charles Creek row in the FIS is a copy-paste of the Hickory Creek row.** Both DA and Q1 values duplicated into wrong stream.
  - **Impact:** Charles Creek FEMA regulatory Q100 is wildly inflated. Actual drainage is ~36 sq mi, not 132 — so actual Q1 is probably an order of magnitude smaller than the published 35,000 cfs. Any insurance rating, permit, or floodplain map based on Charles Creek's Q100 is suspect.
  - **Flagged for FEMA phone call** — this one rises to the level of Case 001 Davidson/Cumberland Label Shift. Would need LOMR or new detailed study to publish correct flows.
  - ✓ Confirmed 2026-04-23 via SnapBasin two-point delineation AND full basin delineation screenshot showing Charles Creek's 36.3 sq mi catchment geographically (not 132).
  - **Estimated real Q100:** ~15,000-22,000 cfs based on NSS regression (SnapBasin shows Q2=2,640 for this basin; scaling to Q100 using typical TN Region 2 ratios). Published 35,000 cfs is inflated by roughly 40-60%.
  - Candidate for SnapBasin_Case_002 write-up.

### MAURY — unusual Manning's, three streams flagged together

- **McCutcheon Creek, McCormack Branch, Orphanage Branch** (Manning's Table 2)
  - All three streams published with overbank n range = **0.040 – 0.265**
  - Max 0.265 is ~50% higher than typical heavy-brush cap of 0.175. Could be legit (dense Maury floodplain vegetation — all three streams in same area) or a grouped typo for 0.165. Value is consistent across all three streams, suggesting intentional assignment rather than single-row typo.
  - Verify against FIS narrative for basis of high n.

---

## April 23, 2026 — Audit findings from 66 auto-parsed counties (Phase B)

### KNOX — confirmed FIS error (digit swap)

- **Holston River** | At Knox County line (DA=3,566)
  - Knox FIS Table published: Q10=25,700, **Q2%=43,200**, Q1=39,700, Q0.2=58,800
  - Corrected to: **Q2%=34,200** — digits 3 and 4 swapped in Knox FIS production.
  - Confirmation: Grainger FIS (different county, same physical point at the Knox-Grainger county line) published Q2%=34,200. DA and the other 3 flow values match Grainger exactly across both reports — only Q2% differs. Monotonicity (25,700 → 34,200 → 39,700 → 58,800) confirms 34,200 is correct.
  - Value updated in fis_flows.json; Knox and Grainger cross-documented.

### GRAINGER — confirmed FIS error (comma/period typo) + parser amplification

- **Richland Creek** | Approximately 5.0 miles upstream of mouth (DA=54.9)
  - Grainger FIS Table 5 published with **PERIODS instead of commas** on two columns:
    - Q2% printed as **"8.300"** (meant 8,300) — period mistyped for comma
    - Q1% printed as **"9.700"** (meant 9,700) — period mistyped for comma
  - Our auto-parser read the periods as decimal points: got Q2%=8.3 and Q1%=9.7. Database had Q2%=8.
  - Corrected to: **Q2%=8,300, Q1%=9,700** based on adjacent row magnitudes (4,900/7,600/8,800/12,200 downstream and 4,200/6,500/7,600/10,500 upstream) and monotonicity.
  - **Dual failure mode worth flagging in the Jeremy Holley package:** FEMA published a typo AND any automated parser downstream will reproduce it as a different error. This is a repeating-risk pattern.
  - Value updated in fis_flows.json.


---

## April 25, 2026 — CUMBERLAND RIVER FULL LOCKDOWN (13 findings across 11 counties)

**Status:** ⚠⚠⚠ Systemic regulatory data integrity failure on the Cumberland River.

What started as the Davidson Vol 2 Cumberland label-shift (Case 001, the original 2026-04-21 Holley email) is now confirmed as **one symptom of an 11-county pattern**. Database scan of all 23 published Cumberland River rows produced 4 distinct duplicate-tuple clusters, 1 monotonicity violation, 1 missing-documentation violation, 2 internal-county duplicates, and 4 cross-county data-sharing patterns that violate independent-methodology assumptions.

**Counties involved (parsed and verified):** Cheatham (Vol 1), Clay, Davidson (Vol 2), Dickson (Vol 1), Jackson (FIS streams index only), Montgomery (Vol 1), Smith (Vol 0), Stewart (Vol 1), Sumner (Vol 1), Trousdale, Wilson (Vol 1).

**Method:** All findings reproducible from `data/FIS/FEMA FIS Flow Database.xlsx`, sheets FLOW LOOKUP / H&H ANALYSES / FIS STREAMS / COUNTY SUMMARY / PROFILE PANELS.

### Finding C-1 — Davidson Vol 2 dual label-shift (Case 001 expanded)

- **Davidson** | Cumberland River | At Stones River conf | Row 880, FLOW LOOKUP
  - Label says "Stones conf" but DA published = **11,673.9 sq mi** = Old Hickory Dam (matches Wilson 11,674, Sumner 11,694 at OHD)
  - Flow tuple [115k/149k/173k/198k/255k] is unique in the four-county window — physically plausible for the actual Stones confluence
  - **Diagnosis:** Row label and flows are for Stones conf, but DA was lifted from OHD (one station upstream).

- **Davidson** | Cumberland River | At Harpeth River conf | Row 879, FLOW LOOKUP
  - Label says "Harpeth conf" but DA published = **12,691.4 sq mi** ≈ 12,683 (Stones conf with Stones system per StreamStats, +8.4 sq mi)
  - Flow tuple [115k/128k/140k/155k/190k] = **Tuple A** (see C-3) — copy-paste pattern
  - **Diagnosis:** Row label is for Harpeth conf, but DA is for Stones conf and flows match a 4-row copy-paste cluster downstream.

- **Documented in Holley package 2026-04-21**, page 1 of 01_comparison_sheet.pdf. The new finding is that the "shift" is actually **two independent shifts** — DA and flows came from different reference rows, not a single label-row offset.

### Finding C-2 — DA monotonicity violation between Sumner and Cheatham (independent of Davidson Vol 2)

- **Sumner** | Cumberland River | Davidson County boundary | Row 3831, FLOW LOOKUP
  - Published DA = **13,589 sq mi**

- **Cheatham** | Cumberland River | At confluence of Harpeth River | Row 555, FLOW LOOKUP
  - Published DA = **13,259 sq mi**

- **Diagnosis:** Cumberland flows downstream from Sumner-Davidson boundary → through Davidson → into Cheatham → past Stones conf → past Harpeth confluence. DA must increase monotonically (basin only grows). Sumner's published DA at the upstream boundary exceeds Cheatham's at a point downstream by **330 sq mi**. Either Sumner's value is too high or Cheatham's is too low (or both). Not explained by the Davidson Vol 2 label-shift — this is between two FIS volumes that don't involve Davidson at all.

### Finding C-3 — Tuple A: Identical 5-tuple [115/128/140/155/190] across 4 county FIS volumes

| Row | County | Location | DA | Vol |
|---|---|---|---:|---|
| 555 | CHEATHAM | At conf of Harpeth River | 13,259 | Vol 1 |
| 879 | DAVIDSON | At conf of Harpeth River | 12,691.4 | Vol 2 |
| 1699 | DICKSON | At conf with Harpeth River⁴ | 13,259 | Vol 1 |
| 3831 | SUMNER | Davidson County boundary | 13,589 | Vol 1 |

- 4 physically distinct stations, 3 different drainage areas, 4 separate FIS volumes published on 4 different dates — all carry the **identical** Q10/Q4/Q2/Q100/Q500 tuple.
- Cumberland River flows must change between any two of these stations. Identical 5-tuples are statistically impossible from independent calculations.
- Methodology mismatch (see C-9): these four counties used three different hydrologic modeling approaches, yet output the same 5-digit values across all 5 recurrence intervals. Cross-volume row reuse, not coincidence.

### Finding C-4 — Tuple C: Identical 4-tuple [110/138/157/190] across 3 rows / 2 counties

| Row | County | Location | River Mile |
|---|---|---|---:|
| 4374 | WILSON | At river mile 264.7 | 264.7 |
| 3981 | TROUSDALE | Above Big Goose Creek | 280.0 |
| 3983 | TROUSDALE | Above Second Creek | 269.9 |

- Three physically distinct points spanning ~15 river miles, all carrying identical Q10/Q2/Q100/Q500.
- Flows must vary across 15 miles of Cumberland mainstem because tributaries enter and DA grows. Identical tuples are not coincidence.

### Finding C-5 — Trousdale County INTERNAL same-volume duplication

- **Trousdale Vol N** | Cumberland River | Above Big Goose Creek (Mile 280.0) | Row 3981
- **Trousdale Vol N** | Cumberland River | Above Second Creek (Mile 269.9) | Row 3983
- Two different river-mile stations within the same FIS volume carrying identical [110/138/157/190]. Same county, same volume, two rows, copy-paste between rows.

### Finding C-6 — Tuple B: Trousdale Mile 260.7 carries Old Hickory Dam values

| Row | County | Location | River Mile / Note |
|---|---|---|---|
| 4373 | WILSON | At river mile 216.2 | OHD shared boundary point |
| 3832 | SUMNER | Old Hickory Dam at RM 216.2 | OHD shared boundary point |
| 3982 | TROUSDALE | Above Rocky Creek (Mile 260.7) | ~45 miles upstream of OHD |

- Wilson + Sumner sharing OHD values is **correct** (OHD is on their shared boundary).
- Trousdale Mile 260.7 is **45 river miles upstream of OHD** and should not carry the same flows. Its published [120/146/158/193] tuple is mis-applied from OHD.

### Finding C-7 — Dickson Vol 1 wholesale replication of Cheatham Vol 1 Cumberland data

| Row | County | Location | DA | Q100 | Q500 |
|---|---|---|---:|---:|---:|
| 554 / 1698 | CHEATHAM / DICKSON | At Cheatham Dam | 14,160 | 208,000 | 255,000 |
| 555 / 1699 | CHEATHAM / DICKSON | At Harpeth River conf | 13,259 | 155,000 | 190,000 |
| 556 / 1700 | CHEATHAM / DICKSON | Montgomery Co boundary | 14,442 | 197,000 | 253,000 |

- Dickson Vol 1's three Cumberland rows are **identical replicas** of Cheatham Vol 1's three Cumberland rows (DA + all flows, with footnote 4 indicating "from neighboring county").
- The Cumberland River does not run through Dickson County. These rows exist as cross-county references.
- Consequence: any error in Cheatham's three Cumberland rows is automatically duplicated as a regulatory value in Dickson's FIS.

### Finding C-8 — Davidson Vol 2 Cumberland Table 9 has NO H&H ANALYSES record

- The H&H ANALYSES tab documents the hydrologic and hydraulic models used to generate every county's regulatory flows.
- 7 counties have Cumberland River H&H records: Cheatham, Dickson, Montgomery, Smith, Stewart, Sumner, Wilson.
- **Davidson Vol 2 has zero Cumberland River H&H records** despite publishing two Cumberland regulatory rows in Table 9.
- **Diagnosis:** Davidson Vol 2's Cumberland regulatory flows are published **without methodology documentation**. No audit trail exists in the FIS to verify the source models. Combined with C-1 label-shifts, the Davidson Vol 2 Cumberland data is regulatory output without provenance.

### Finding C-9 — Three different methodologies output identical 5-tuple — statistically impossible

For Tuple A [115/128/140/155/190] across 4 rows:

| County | Hydrologic Model | Hydraulic Model |
|---|---|---|
| SUMNER (DCB row) | Regression Equations USGS 2003 | HEC-RAS 4.1.0 (USACE 2010) |
| DAVIDSON (Harpeth row) | **NONE DOCUMENTED** | **NONE DOCUMENTED** |
| CHEATHAM (Harpeth row) | Other (HEC-2 / HEC-RAS USACE study) | HEC-2 (USACE 1976) + HEC-RAS 4.1.0 |
| DICKSON (Harpeth row) | Other (HEC-2 / HEC-RAS USACE study) | HEC-2 (USACE 1976) + HEC-RAS 4.1.0 |

- Regression equations and a HEC-RAS hydraulic study cannot independently produce the same 5-digit values across all 5 recurrence intervals. The probability of independent calculation matching to the digit at all 5 intervals is essentially zero.
- The only explanation is **values transcribed across rows that should have been independently calculated** for each station.

### Finding C-10 — Davidson missing from FIS STREAMS index for Cumberland

- FIS STREAMS tab lists each county's studied streams.
- 7 counties list Cumberland River in their FIS STREAMS index: Cheatham, Jackson, Montgomery, Smith, Stewart, Trousdale, Wilson.
- **Davidson Vol 2 does not list Cumberland River in its FIS STREAMS index** despite publishing Cumberland Table 9 entries.
- Combined with C-8 (no H&H record), the Davidson Vol 2 Cumberland data is **floating regulatory data with no supporting documentation anywhere else in the FIS**.

### Finding C-11 — Regional batch-publishing pattern

FIS publication dates by county:
- January 15, 2021: **Dickson, Montgomery, Stewart** (3 counties same day)
- February 26, 2021: **Cheatham, Sumner** (2 counties same day)
- April 25, 2024: **Wilson, Jackson** (2 counties same day)
- September 29, 2010: **Trousdale, Clay** (2 counties same day)

Counties published on the same day share underlying USACE Cumberland River studies. This explains why Cheatham/Dickson/Sumner Cumberland data is shared. **It does not justify the violations** — shared studies should still produce station-specific flows that vary monotonically with DA, not identical 5-tuples.

### Finding C-12 — Montgomery Vol 1 contains a 1,455 sq mi DA jump in 0.1 river mile (likely Red River confluence)

- Row 2683: Cumberland River at Mile 125.3 — DA = 14,442 sq mi
- Row 2682: Cumberland River at Mile 125.2 — DA = 15,897 sq mi

- 1,455 sq mi DA increase over 0.1 river mile. **This is consistent with the Red River joining the Cumberland near Clarksville at ~RM 125** (Red River drainage at mouth ≈ 1,565 sq mi). Not a violation — physically explained.
- Flagged for completeness; included to demonstrate that the Cumberland's DA progression IS internally consistent at this point even though similar numbers (14,442) appear elsewhere as part of Tuple cross-references.

### Finding C-13 — Davidson Stones Q500 = Cheatham Dam Q500 = 255,000 cfs (partial Frankenstein)

- Davidson Stones row Q500 = 255,000 cfs (DA 11,673.9)
- Cheatham Dam row Q500 = 255,000 cfs (DA 14,160)
- Two physically distinct Cumberland points 25 miles apart with very different drainage areas share Q500 = 255,000.
- The other four flow values DO NOT match. Frankenstein hypothesis (whole row lifted from Cheatham Dam) is **rejected**.
- However, the Q500 matching at the high end of the regulatory envelope across stations is itself worth noting — either coincidence at a regulated-flow ceiling or another partial cross-station value lift.

---

### What the Holley letter (2026-04-21) already showed and didn't explicitly call out

- The letter's page 2 Table C "Cross-county Cumberland River longitudinal comparison" contains the data for findings C-1, C-2, and C-3.
- Key Observations on the chart explicitly called out: Davidson Vol 2 label-shift, OHD agreement across Wilson/Sumner/Davidson "Stones," 198,000 cfs Stones row anomaly, and the 155,000 Q1% appearing at 3 stations.
- Did NOT explicitly call out: the DA monotonicity violation (Sumner DCB → Cheatham Harpeth), the framing of identical 5-tuples as copy-paste rather than agreement, or the Dickson Vol 1 replication (Dickson wasn't in the comparison chart).

### Required follow-up for Holley (next email)

1. **Add Annotation A** to the comparison chart: Sumner DC boundary DA (13,589) > Cheatham Harpeth DA (13,259) violates downstream monotonicity, independent of Davidson Vol 2.
2. **Add Annotation B** to the comparison chart: Tuple A [115/128/140/155/190] copy-pasted across 4 rows / 4 counties / 4 dates / 3 methodologies. Statistically impossible from independent calculation.
3. **New annotation: Dickson Vol 1 entirely replicates Cheatham Vol 1 Cumberland data.** Whatever fix Cheatham's three rows need, Dickson needs the identical fix.
4. **New annotation: Davidson Vol 2 Cumberland Table 9 has no H&H ANALYSES record and no FIS STREAMS index entry.** Floating regulatory data without provenance.
5. **New annotation: Trousdale Cumberland data** has internal same-volume duplication (Mile 280.0 = Mile 269.9) and cross-county duplication of Wilson RM 264.7.

### Counties affected by Cumberland River regulatory data integrity issues

| County | Issues | Severity |
|---|---|---|
| DAVIDSON Vol 2 | C-1 (dual label-shift), C-8 (no H&H), C-10 (not in FIS STREAMS) | ⚠⚠⚠ Highest |
| CHEATHAM Vol 1 | Source row of Tuple A copy-paste at Harpeth conf; cited by Dickson | ⚠⚠ |
| DICKSON Vol 1 | C-7 (wholesale Cheatham replica), inherits any Cheatham errors | ⚠⚠ |
| SUMNER Vol 1 | C-2 (DA monotonicity vs Cheatham Harpeth), C-3 (Tuple A) | ⚠⚠ |
| TROUSDALE | C-4 (Tuple C cross to Wilson), C-5 (internal duplicate), C-6 (OHD values mis-applied 45 miles upstream) | ⚠⚠ |
| WILSON Vol 1 | C-4 (Tuple C cross-county to Trousdale) | ⚠ |
| MONTGOMERY Vol 1 | C-12 (verified physically consistent — Red River) | ✓ Clean |
| STEWART Vol 1 | None — clean | ✓ Clean |
| SMITH Vol 0 | None — clean | ✓ Clean |
| CLAY | Insufficient data (1 point only) | — |
| JACKSON | In FIS STREAMS index but no FLOW LOOKUP entries (approx-only?) | — |

### Reproduction commands

```python
from openpyxl import load_workbook
wb = load_workbook("data/FIS/FEMA FIS Flow Database.xlsx", data_only=True, read_only=True)
ws = wb["FLOW LOOKUP"]
# Pull all Cumberland rows (returns 23 in current DB)
cum = [(i, r) for i, r in enumerate(ws.iter_rows(values_only=True), 1)
       if i >= 6 and r and r[1] and "cumberland" in str(r[1]).lower()]
# Group by 5-tuple to find duplicates
from collections import defaultdict
groups = defaultdict(list)
for i, r in cum: groups[(r[4], r[5], r[6], r[7], r[8])].append((i, r))
# Print duplicate clusters
for tup, rows in groups.items():
    if len(rows) > 1: print(tup, "→", len(rows), "rows")
```

### Status

- Holley letter sent 2026-04-21 covered Findings C-1 (partial) and C-3 (framed as agreement, not copy-paste).
- All 13 findings now locked down in this log entry as of **2026-04-25 morning** (IPP review week).
- Annotated comparison chart v2 pending — adds explicit callouts for findings C-2, C-3 (recast), C-7, C-8, C-9, C-10.
- Source PDFs for verification: Cheatham 47021CV001, Davidson 47037CV002D, Dickson 47043CV001, Sumner 47165CV001, Wilson 47189CV001, Trousdale 47169CV000A.
- Database reproducible from `data/FIS/FEMA FIS Flow Database.xlsx` — every finding above derives from queryable data.


---

## April 25, 2026 evening — DAVIDSON CHASE: 5 NEW FINDINGS (C-14 through C-18)

Travis went to Davidson County in person and screenshotted every Cumberland River mention across all 7 volumes (V1-V7). 30 receipts captured. Reading the chase produced 5 new findings plus a partial revision of C-8.

### Finding C-14 — Davidson Vol 2 Table 11 documents Nashville USGS gage with no Table 9 representation

- **Davidson** | Cumberland River | Vol 2 page 130 (pg 90 of 91) — Table 11: Stream Gage Information used to Determine Discharges
- Two USGS Cumberland gages cited:
  - Gage **03426310** — Cumberland River at **Old Hickory Dam (TW)** — DA **11,673 sq mi** — Period 12/16/1951 to 05/02/2010
  - Gage **03431500** — Cumberland River at **Nashville, TN** — DA **12,856 sq mi** — Period 01/01/1927 to 05/03/2010
- Davidson Vol 2 Table 9 contains 2 rows: DA 11,673.9 (matches OHD gage exactly) and DA 12,691.4 (matches Stones conf with system, NOT Nashville gage)
- **Diagnosis:** A documented federal USGS gage at Nashville with 83 years of record is cited as a discharge source but never represented as a Table 9 row. The Nashville gage's DA 12,856 should produce a Q value ~halfway between Davidson's published 11,673.9 row and Cheatham's downstream Harpeth conf 13,259 row. It doesn't appear.
- **Receipt:** screenshots/davidson/06_v2_p130_table11_stream_gages.jpg

### Finding C-15 — Davidson uses CHEATHAM POOL flood frequency study for UPSTREAM Cumberland flows

- **Davidson** | Cumberland River | Vol 3 page 138 (pg 18 of 64) — Table 12: Summary of Hydrologic and Hydraulic Analyses
- Hydrologic Model: "Other"
- Hydraulic Model: HEC-RAS 4.1.0 (USACE 2010a)
- Date Analyses Completed: 05/01/2012
- Special Considerations cite: **"Cumberland River - Cheatham Pool Flood Frequency Update Report" (USACE 2012a)**
- The Cheatham Pool is the impoundment behind Cheatham Dam — DOWNSTREAM of Davidson
- Davidson's Table 9 stations (Stones conf, Harpeth conf) are UPSTREAM of Cheatham Dam
- **Diagnosis:** Using a downstream pool flood frequency study to establish upstream river regulatory flows is a methodological mismatch. Pool studies model the impoundment as a static water body with regulated outflow; mainstem river studies model flowing channel hydraulics. Different physics, different math.
- **Receipts:** screenshots/davidson/07_v3_p138_table12_hh_analyses.jpg + screenshots/davidson/17_v6_p369_bibliography_USACE_2012a_cheatham_pool.jpg

### Finding C-16 — Old Hickory Lake regulatory flows trace to 47-year-old USACE 1979 study

- **Davidson** | Cumberland River - Old Hickory Lake | Vol 6 page 346 (pg 20 of 47) — Table 28: Summary of Contracted Studies
- Cumberland River - Old Hickory Lake row: Contractor USACE Nashville District, Contract Number "N/A" (predates contract numbering), **Work Completed Date: 01/01/1979**
- Bibliography (Vol 6 page 368) confirms: "USACE 1979 — Flood Frequency Study for the Cumberland River Basin" — January 1979
- **Diagnosis:** Davidson's regulatory flows for the 6.0 mile Old Hickory Lake portion of the Cumberland mainstem trace to a 47-year-old USACE study. No restudy in 47 years despite the 2010 Nashville flood being the largest event ever recorded in Davidson County.
- **Receipts:** screenshots/davidson/15_v6_p346_table28_contracted_studies.jpg + screenshots/davidson/16_v6_p368_bibliography_USACE_1979.jpg

### Finding C-17 — 115 Cumberland cross sections in Table 23 supported by only 2 published Table 9 discharges

- **Davidson** | Cumberland River | Vol 4 pages 205-210 — Table 23: Floodway Data
- 115 cross sections cataloged: A through DK
- Distance range: 850,474 ft to 1,120,299 ft above mouth (~RM 161 to RM 212), spanning 51.1 miles (matches Vol 1 Table 2 documented length)
- Each section has: distance, width, section area, mean velocity, regulatory WSEL, without-floodway WSEL, with-floodway WSEL, increase
- **Davidson Vol 2 Table 9 publishes only 2 discharge points** for this 51.1-mile reach (Stones conf, Harpeth conf)
- **Diagnosis:** 113 of 115 cross sections have computed flood elevations interpolated/extrapolated from just 2 published discharge values. Combined with C-15 (those 2 discharges came from the Cheatham Pool study) and C-16 (the Old Hickory Lake portion uses 1979 data), the cross-section flood elevations downstream are derived from a thin, mismatched, partially-ancient data foundation.
- **Receipts:** screenshots/davidson/09_v4_p205_table23_csxn_A-V.jpg through 14_v4_p210_table23_csxn_DG-DK.jpg (6 screenshots covering all 115 sections)
- **Bonus:** screenshots/davidson/20-30_v7_floodprofile_csxn_*.jpg (11 V7 graphical flood profile plates with cross-section letters and tributary confluence labels)

### Finding C-18 — 2010 Nashville historic peak (188,000 cfs) sits between published Q100 estimates with no Table 9 representation

- **Davidson** | Cumberland River | Vol 1 page 45 (pg 57 of 62) — Table 6: Historic Flooding Elevations
- Cumberland River at USGS Gage 03431500 (Nashville) — Historic peak: **52.55 ft NAVD88 / 188,000 cfs** on 2010-05-03
- Compare to Davidson Vol 2 Table 9:
  - Stones conf row Q100 = 198,000 cfs (DA 11,673.9 = OHD per C-1 label-shift)
  - Harpeth conf row Q100 = 155,000 cfs (DA 12,691.4 = Stones conf with system per C-1)
- 188,000 cfs sits between 155,000 and 198,000 — approximately Q100 magnitude
- **Diagnosis:** A historic event of regulatory significance occurred at the Nashville gage (DA 12,856). The published Table 9 has no row at this DA. Per C-14, the Nashville gage is cited as a discharge source but unrepresented. The 2010 event is the FEMA's own internal evidence that a station between their two published rows experienced a Q100-magnitude flow.
- **Receipt:** screenshots/davidson/04_v1_p45_table6_historic_2010.jpg

### Revision of Finding C-8

Original Finding C-8 stated: "Davidson Vol 2 has no H&H ANALYSES record for Cumberland."

**Correction:** Davidson DOES have H&H records — they're in Vol 3 Table 12 (pg 18 of 64), not Vol 2 where Table 9 lives. Our database parser missed them. The deeper methodological issue (Cheatham Pool study being applied to upstream Davidson stations) is now documented as Finding C-15.

**Database remediation needed:** The H&H ANALYSES tab in FEMA FIS Flow Database.xlsx should be re-parsed to include cross-volume Cumberland H&H records for Davidson.

---

## Davidson cumulative finding tally (post-chase, 2026-04-25 evening)

| Finding | Severity | Evidence |
|---|---|---|
| C-1 | SEVERE | Dual label-shift in Vol 2 Table 9 (DA + flows from different stations) |
| C-3 | SEVERE | Tuple A copy-paste row across 4 county FIS volumes |
| C-8 | REVISED | Parser miss — Davidson H&H exists in Vol 3 Table 12 (see C-15 for real issue) |
| C-9 | SEVERE | Three methodologies output identical 5-tuple (statistically impossible) |
| C-10 | SEVERE | Davidson missing from FIS STREAMS index for Cumberland |
| C-13 | NOTED | Q500 coincidence between Davidson Stones row and Cheatham Dam |
| **C-14** | **SEVERE** | **Nashville gage cited but not in Table 9** |
| **C-15** | **SEVERE** | **Cheatham Pool study applied to upstream stations** |
| **C-16** | **SEVERE** | **47-year-old 1979 study still active for OHL portion** |
| **C-17** | **SEVERE** | **115 cross sections supported by only 2 discharges** |
| **C-18** | **SEVERE** | **2010 Nashville peak unrepresented in Table 9** |

**11 distinct findings on Davidson alone, 5 of which were uncovered by Travis's in-person chase across all 7 FIS volumes (not from automated parsing).**

The chase methodology is now established as a template for the next 10 counties. Going to a county's FIS in person across every volume catches what the automated parser misses (especially data that lives outside the Summary of Discharges table — gage citations, contracted study contracts, bibliography references, Table 23 floodway data, and the actual flood profile graphical plates).


---

## April 25, 2026 evening — CHEATHAM CHASE: 5 NEW FINDINGS (C-19 through C-23) — including FOUND ADMISSION

Travis ran the same in-person chase for Cheatham County across all FIS volumes. 20 receipts captured. Reading produced 5 more findings — including a direct FEMA-self-admission that Q500 isn't really Q500.

### Finding C-19 — Q4% (25-year) "Not Calculated" for lower Cumberland reach

- **Cheatham** | Cumberland River | Montgomery County boundary station | Vol 1 page 31 — Table 9: Summary of Discharges (continued)
- Published Q4% = "*" with footnote "Not calculated for this Flood Risk Project."
- Vol 2 panel 019P flood profile (distance 765,000 to 778,000 ft above Ohio River confluence) shows Q4% curve marked "DATA NOT AVAILABLE" — confirming the gap visually.
- **Diagnosis:** ~5+ miles of regulatory Cumberland floodway in Cheatham has no published 25-year discharge. Q4% is used in many engineering design contexts (drainage design, scour analysis, recurrence-based bridge sizing). Its absence is a real data gap.
- **Receipts:** screenshots/cheatham/04_v1_p31_table9_cumberland_rows.jpg + screenshots/cheatham/13_v2_floodprofile_csxn_E_ODD_PROFILE_p019.jpg

### Finding C-20 — SPF substituted for Q0.2 (500-yr) at Montgomery boundary

- **Cheatham** | Cumberland River | Montgomery County boundary station | Vol 1 page 31 Table 9
- Q0.2 = 253,000 cfs, footnoted ² as "Standard Project Flood"
- SPF is a USACE-derived design discharge representing extreme storm conditions, NOT a return-period frequency analysis.
- **Diagnosis:** The published value is regulatory but is methodologically a different quantity than the Q0.2 column header implies. Engineers using FEMA's Q0.2 for design assume frequency-based recurrence interpretation. The footnote disclosure is a single character (²) easy to miss in spreadsheet exports.
- **Receipts:** screenshots/cheatham/04_v1_p31_table9_cumberland_rows.jpg

### Finding C-21 — Cited gage 03435000 stopped recording 35 years ago, missed 2010 flood

- **Cheatham** | Cumberland River | Vol 1 page 45 — Table 11: Stream Gage Information used to Determine Discharges
- USGS gage 03435000 "Cumberland River below Cheatham Dam, TN" — DA 14,163 — Period **01/01/1927 to 02/20/1991**
- Cited as a discharge source for Cheatham's Cumberland Table 9 flows.
- The May 2010 flood — Cheatham Vol 1 Table 6 records this as the largest event ever in this reach, classified as >500-yr at the downstream Clarksville gage — was NOT captured by this gage (data ended 1991).
- **Diagnosis:** The cited gage's 64-year record (1927-1991) excludes the largest flood of record (2010). Any flood frequency analysis based on this gage's data alone systematically underestimates extreme-event magnitudes.
- **Receipts:** screenshots/cheatham/05_v1_p45_table11_stream_gages.jpg + screenshots/cheatham/03_v1_p27_table6_historic_table7_dam.jpg

### Finding C-22 — Old Hickory Dam gage cited despite different county + smaller DA

- **Cheatham** | Cumberland River | Vol 1 Table 11
- Also cites USGS gage 03426310 "Cumberland River at Old Hickory Dam (TW)" — DA 11,673 — Period 12/31/1926 to 03/11/2011
- OHD gage is in Sumner/Wilson counties, NOT Cheatham. DA = 11,673 is SMALLER than any Cheatham Cumberland Table 9 station (13,259 / 14,160 / 14,442).
- **Diagnosis:** Cross-county gage extrapolation is not inherently wrong (longer records support extreme-event frequency analysis), but using a smaller-DA gage from a different county to inform larger-DA discharges in Cheatham requires careful drainage-area scaling and watershed-similarity verification. The FIS does not document this scaling.

### Finding C-23 — FOUND ADMISSION: FIS itself states Q500 is not a 500-yr flow

- **Cheatham** | Cumberland River (Montgomery boundary segment) | Vol 1 page 49 — Table 12: Hydrologic and Hydraulic Analyses Special Considerations
- **Direct quote from the FIS:**

> "Results of the regulated frequency study were found to yield statistically reliable estimates of floods up to and including the 100-year event. For events greater in magnitude than the 100-year flood, such as the 500-year flood, the statistical reliability of predicted flow was poor. Estimates of the 500-year flood discharges from the study were found to approximate the USACE-developed Standard Project Flood (SPF) for the majority of the Cumberland River. The SPF has been widely disseminated to the general public by the USACE, Nashville District, to be used for design purposes of developments adjacent to the Cumberland River. **Because of the low reliability of estimates for extremely rare events and to maintain consistency with previously published information, the SPF is used in lieu of the 500-year flood for this study as shown on the Flood Profiles (FIS 1999).**"

- **Diagnosis:** This is the single strongest receipt in the entire Cumberland audit. **FEMA's own FIS document explicitly states the published Q500 column for ~5 miles of regulated Cumberland in Cheatham is filled with SPF values, not 500-year return-period discharges.** Engineers, insurance underwriters, and floodplain managers using the published Q500 for risk assessment are using a fundamentally different quantity than the column header advertises. The disclosure exists but is buried in Table 12 Special Considerations text — a section few users consult when extracting Table 9 data.
- **This admission corroborates and explains C-20.**
- **Receipt:** screenshots/cheatham/06_v1_p49_table12_hh_SPF_admission.jpg

---

### Cumberland Lockdown finding tally — running total

After Cheatham chase, the Cumberland Discrepancy Log contains **23 distinct findings** across 11 counties. Counties with chase-ingested receipts:
- DAVIDSON: 11 findings (C-1, C-3, C-8, C-9, C-10, C-13, C-14, C-15, C-16, C-17, C-18) + 30 screenshots
- CHEATHAM: 11 findings (C-2, C-3, C-7, C-9, C-11, C-13, C-19, C-20, C-21, C-22, C-23) + 20 screenshots

Combined: 22 unique findings on 2 counties' Cumberland reaches alone (C-3 / C-9 / C-13 are shared). Plus 50 screenshot receipts in the deployed site.

**Strongest single finding: C-23.** This is FEMA admitting in writing that their published Q500 column on a regulated reach of the Cumberland is methodologically a Standard Project Flood, not a 500-year frequency discharge. The SAME footnote applies to Davidson's Cumberland Q500 values via the shared underlying USACE 2012a Cheatham Pool study (per C-15) and to Dickson via wholesale replication of Cheatham (per C-7) — meaning **the SPF-as-Q500 substitution propagates across at least 3 counties' regulatory floodways for the Cumberland mainstem.**

### Methodology takeaway

The Cumberland chase methodology — go to the source FIS volumes in person, screenshot every Cumberland mention across all volumes, then audit the screenshots — has now produced 10 additional findings (C-14 through C-23) that the automated parser missed. The parser only scanned Table 9 values in our database. The chase finds:
- Cross-volume references (Davidson H&H in Vol 3 not Vol 2, Cheatham SPF admission in Vol 1 Table 12)
- Historical context (cited gage periods, study dates, contractor histories)
- Methodological footnotes that aren't in the structured tables (Cheatham SPF substitution explained in narrative)
- Internal flood profile plate annotations (Q4% data gaps, tributary confluences, county boundaries)

Recommendation: continue chase methodology for Sumner, Dickson, Trousdale, Wilson — likely 5-10 more findings per county.


---

## April 25, 2026 evening (cont.) — MONTGOMERY + DICKSON CHASES: 5 NEW FINDINGS (C-24 through C-28)

Two more counties chased. 21 receipts captured (10 Montgomery + 11 Dickson). Findings continue to accumulate.

### Finding C-24 — Dickson Vol 1 Table 9 SELF-DOCUMENTS its Cumberland import from Cheatham

- **Dickson** | Cumberland River | Vol 1 page 32 — Table 9: Summary of Discharges (continued)
- All three Cumberland River rows footnoted **⁴ "Added from Cheatham County FIS 2016"**
- This is the explicit, FEMA-self-documented receipt for the wholesale-replica diagnosis previously inferred in C-7
- **Time-staleness consequence:** Dickson imported from Cheatham FIS **2016**. Cheatham's current FIS is dated **02/26/2021**. Dickson's regulatory data may be 5+ years stale relative to Cheatham's current values. Future Cheatham updates do not automatically propagate to Dickson.
- **Receipt:** screenshots/dickson/02_v1_p32_table9_FOOTNOTE_added_from_cheatham_2016.jpg

### Finding C-25 — Dickson and Cheatham publish overlapping cross-section data for the SAME ~5 miles of Cumberland

- **Dickson** | Cumberland River | Vol 2 page 58 — Table 23: Floodway Data
- 5 cross sections A-E at distances **762,024 to 786,375 ft** above Ohio River confluence
- **PHYSICALLY THE SAME LOCATIONS as Cheatham's Table 23 sections D-H** (per Cheatham V2 page 65 chase)
- Same regulatory WSELs (398.5 / 399.2 / 399.4 / 399.8 / 400.9 — match exactly)
- Complementary jurisdiction-width splits (Dickson within = 416 ft + Cheatham within = 2,329 ft = total 2,745 ft at boundary station — match arithmetic)
- **Diagnosis:** ~5 miles of regulated Cumberland River has Table 23 floodway data published in BOTH Dickson AND Cheatham FIS reports. This is jurisdictional overlap, not error per se, but creates a future risk: if FEMA updates Cheatham's data without updating Dickson's, the same physical river points will carry different regulatory floodway widths and elevations across two FIS publications.
- **Receipt:** screenshots/dickson/05_v2_p58_table23_csxn_A-E_overlaps_cheatham.jpg

### Finding C-26 — Montgomery Cumberland regulatory data dates from 06/08/1982 — OLDEST in the chain

- **Montgomery** | Cumberland River | Vol 1 page 17 Table 2 — Date of Analysis: **06/08/1982**
- Montgomery Vol 2 page 102 Table 28 confirms: USACE Nashville District contract **IAA-H-9-79, Project Order No. 13**, Work Completed **June 1982**, FIS Report Dated **12/15/1983**
- **Montgomery's Cumberland regulatory data is 44 years old.** Pre-dates HEC-RAS entirely (HEC-2 era).
- **Comparison across the chain:**
  - Montgomery (1982) — 44 yrs
  - Davidson Old Hickory Lake (1979) — 47 yrs
  - Cheatham lower reach (1998) — 28 yrs
  - Cheatham upper reach + Davidson upper (USACE 2012a) — 14 yrs
  - Dickson (imported from Cheatham 2016) — 9 yrs but stale
- **Diagnosis:** The 2010 May Flood (largest event of record on the Cumberland) was not in Montgomery's 1982 dataset. Frequency analysis methodology, climate baseline, and watershed conditions have all evolved significantly since 1982.
- **Receipts:** screenshots/montgomery/01_v1_p17_table2_flooding_sources_1982.jpg + screenshots/montgomery/07_v2_p102_table28_USACE_1982_contract.jpg

### Finding C-27 — Montgomery Q500=253,000 NOT footnoted as SPF — propagates SPF substitution invisibly

- **Montgomery** | Cumberland River at Mile 125.3 | Vol 1 page 31 Table 9
- DA 14,442, Q0.2 = **253,000 cfs** — NO footnote
- Same value (253,000) at SAME physical river point footnoted as Standard Project Flood in:
  - **Cheatham** Vol 1 Table 9 (footnote ³ — see C-23)
  - **Dickson** Vol 1 Table 9 (footnote ³)
- **Diagnosis:** Montgomery silently propagates the SPF substitution from upstream Cheatham/Dickson without disclosing it. Users consulting only Montgomery's Table 9 cannot tell the published Q0.2 is methodologically a SPF design flow rather than a 500-year frequency analysis. The disclosure asymmetry across the three counties means a single regulatory value carries different methodology disclosure depending on which county FIS you read.
- **Receipt:** screenshots/montgomery/03_v1_p31_table9_cumberland_2_rows.jpg

### Finding C-28 — Q4% (25-year) Not Calculated across THREE counties for the same Cumberland reach

- Q4% (25-year) discharge asterisked as "Not calculated for this Flood Risk Project" across:
  - **Cheatham** Vol 1 Table 9 — Cumberland River at Montgomery County boundary
  - **Dickson** Vol 1 Table 9 — Cumberland River at Montgomery County boundary
  - **Montgomery** Vol 1 Table 9 — Cumberland River at BOTH Mile 125.2 AND Mile 125.3
- **Diagnosis:** The 25-year regulatory discharge gap propagates across ~10+ miles of regulated Cumberland through three separate regulatory jurisdictions. Q4% (25-yr) is widely used in engineering design (drainage design, scour analysis, recurrence-based bridge sizing — TDOT scour evaluations, AASHTO bridge hydraulics). Its absence is a systemic gap across multiple counties, not an isolated omission.
- **Receipts:** screenshots/cheatham/04_v1_p31_table9_cumberland_rows.jpg + screenshots/dickson/02_v1_p32_table9_FOOTNOTE_added_from_cheatham_2016.jpg + screenshots/montgomery/03_v1_p31_table9_cumberland_2_rows.jpg

---

### Cumberland Lockdown finding tally — running total

After Montgomery + Dickson chases, the Cumberland Discrepancy Log contains **28 distinct findings** across 11 counties. Counties with chase-ingested receipts:
- DAVIDSON: 11 findings + 30 screenshots
- CHEATHAM: 11 findings + 20 screenshots
- DICKSON: 7 findings + 10 screenshots (severity SEVERE)
- MONTGOMERY: 4 findings + 11 screenshots (severity bumped CLEAN → SEVERE — was only C-12 verified-clean Red River observation; chase revealed C-26 ancient study, C-27 SPF disclosure asymmetry, C-28 Q4% gap)

Combined: 81 receipts on the live site spanning 4 chased counties + 28 documented findings.

### Cross-county Cumberland data flow chain (now self-documented)

```
USACE 1979 Flood Frequency Study for the Cumberland River Basin
    ↓
USACE 1982 (Montgomery contract IAA-H-9-79 PO #13) — Montgomery's published data
    ↓
USACE 1998 study — Cheatham lower reach below Cheatham Dam (with admitted SPF-as-Q500 substitution)
    ↓
USACE 2012a Cheatham Pool Flood Frequency Update Report — Cheatham upper reach + Davidson (UPSTREAM mis-application)
    ↓
AECOM 2017 — Cheatham upper reach hydraulic modeling (per Vol 2 Table 28)
    ↓
Cheatham FIS 2016 → "Added" to Dickson FIS 2021 (footnote-documented)
```

This chain is now fully traceable from the source studies to the published FIS values across multiple counties.

### Methodology takeaway (updated)

Chase methodology has now produced findings C-14 through C-28 — fifteen findings beyond what the automated parser surfaced. All from in-person review of FIS volumes that the parser couldn't reach (cross-volume references, footnotes, narrative methodology disclosures, contract documentation, bibliography citations, Table 23 floodway data with cross-section enumeration, flood profile graphical plates).

Remaining counties to chase: Sumner, Trousdale, Wilson, Stewart, Smith, Jackson, Clay.