Scenario registry — 100-dam programme

For the report thread — read this first

What this is. The single authoritative record of every accepted dam-break scenario in the programme: its breach parameters, how they were derived, the volume errors, the modelled and theoretical peak discharges, and the exact file to read results from. It is generated from the accepted artefacts themselves, never typed by hand.

Where the files are. ~/final_100_dam_hdfs — the primary store. Internal, non-removable, and the only location a report should read from. Every path in the tables below is relative to it. /Volumes/100dams, GSDSO and GSDSO2 hold mirrors written from it; they are removable and an unmounted volume reads exactly like an empty one. Do not read from them.

Each scenario is a complete package. <Dam>/<run_id>__<canonical>/ is a self-contained, openable HEC-RAS project — plan, geometry, flow, terrain, LULC, features. Copy that one folder; nothing needs assembling.

Staleness has been removed, and stays removed. Superseded runs are excluded by construction — the register retires them and they are never written to the store. Every sync prunes any file, shared file or empty directory that does not back a current register row. A scenario being rerun shows pending in every result column rather than its retired values. Both are enforced in code and verified by audit_master_store.py --deep, which hashes all 175 results against the register.

This is the single point of information. Do not look elsewhere. Not the build trees under *_prep/, not accepted_packages/, not the drives, not an older copy of a spreadsheet. If a value you need is not here, it has not been accepted yet.

Assets. Received intake material is in _infofiles/ — infofiles, DEM-correction tables, supplied storage and breach-parameter sheets, breach-location and mesh images, and MASTERFILE.csv consolidating all of it. Storage curves are in _storage_curves/, model curve adopted. Inflow hydrographs — the forcing the runs were driven with — are in _infofiles/ as <Dam>__inflow_flood_hydrograph.csv and <Dam>__inflow_baseflow_hydrograph.csv, with peaks consolidated in INFLOW_HYDROGRAPHS.csv. All 26 dams have both. If an asset you need is missing, ask for it to be generated — do not substitute, improvise, or take one from a build tree. Known gaps: mesh.png is absent for Kalubhar, Utavali-G, Vadi and Vadiya.

Why this exists. Parameter errors were found in accepted runs — breach side slopes left vertical on earthen sections, widths rescaled with the wrong exponents, and scenarios carrying another scenario's formation time. Fifteen scenarios were rerun to correct them. This registry was built so that cannot recur unnoticed: every value is regenerated from the accepted artefact and checked against the relation it claims to follow.

The division of responsibility. This registry is authoritative for the model evidence. The report thread is authoritative for interpretation and presentation. Neither should source the other's material from a third place.

If a number you need sits in the left column, take it from here — do not recompute it from a build tree, and do not carry an older copy forward. If it sits in the right column, it is yours and this registry deliberately stays silent. If it sits in neither, ask.

Deposit your final maps here. _maps/ takes the LULC and DEM maps you generate — <Dam>__lulc.png and <Dam>__dem.png, one per dam. Same rule as everything else in this store: overwrite in place when regenerated, never _v2, _final or a date stamp. A second copy beside the first is how the wrong one gets picked up. Names must match the dam names used here exactly; the sync reports any file that does not. If a map does not exist yet, leave the slot empty and ask — do not take one from a build tree or another release.

Citations must be exact — this is where reports most easily go wrong. The earthen relations are CWC/CDSO 2018 §4.4.1, reproducing Froehlich (2017a). Do not cite Froehlich (2008): that is a different family (0.27 K₀Vw0.32Hb0.04, tf = 63.2√·) which this programme does not use, and the shared DBA methodology chapter currently cites it in error. The masonry relation is §4.4.2 and has no primary source — CWC's own attribution is “Personal Communication to DRIP Project”, and any report reproducing it must say so. The 15 m monolith is IS 6512:1984 cl. 7.1.2. Note also that kM means two different factors: 1.0/1.5 in the width relation, 1.00/1.85 in the peak relation. Carrying one into the other is a category error.

Units. Report all volumes in Mm³ (million cubic metres) — that is how they are given here. Gross storage is stated in the received infofiles in inconsistent units (Mcft, M.Cum, and bare numbers); MASTERFILE.csv carries both gross_storage_Mm3_normalised and gross_storage_unit_as_stated, and the normalised column is the one to quote. Discharges are m³/s, elevations and lengths metres, formation time hours.

State conventions, not received data, as such. k = 4 failed bays is a programme convention — CWC fixes the unit that fails, never how many. Wavg is a trapezoidal section at 3.0:1 u/s and 2.5:1 d/s, worth about ±8 %. The breach invert rule — curve floor in-channel, local foundation on a flank — is ours; CWC defines Vw as volume above the final breach bottom but does not say where that bottom lies.

Dams one row per dam — values that genuinely do not vary by scenario

Earthen breach scenarios one row per scenario

Rows under rerun show pending for every result-derived column — errors, MSA and modelled peak. Their previous artefact is being superseded, so its values are withheld rather than displayed. Breach parameters on those rows are the corrected values now running. MSA is the maximum stage attained — the highest reservoir-side water surface reached during the run, read from the connection's Stage HW. initial is the reservoir level the model started from, taken from the unsteady-flow file. inflow is the forcing encoded in the canonical code. modelled is measured from the run. Q_emp, Q_semi, Q_max are CWC/CDSO 2018 estimates validated against the handbook's solved example. Q_semi is the one comparable with the modelled value — Q_max assumes instantaneous formation and is an upper bound by construction. Hover any HDF path for its sha256.

Where Q modelled is measured. Breach outflow — Breaching Variables column 6 ("Breach Flow") on the breached SA/2D connection, maximum over the run. It is not the total at the dam (Structure Variables column 0), which also carries weir and spillway flow: the two differ by 43 % at Jaspara-Mandava S_EG_03 (1,955.10 against 2,795.11 m³/s). Q_emp, Q_semi and Q_max are breach-peak relations, so comparing them against a total reads as a validation failure when nothing is wrong. Boundary outflow is not a substitute — where the flood does not reach a downstream boundary inside the window it reports zero (Kaniyad does exactly that).

Structural breach scenarios spillway / masonry — a different rule

Non-failure scenarios routing, references and gate operations — no breach

These carry no breach by design: no-failure references, design-flood routing screens, and gate-operation cases. They are accepted artefacts with full packages in the primary store — they simply have no breach parameters to publish, so they are listed rather than tabulated with the breach scenarios.

Breach locations as marked by the supplier — one per dam, all 26

The station each breach is placed at is in the scenario table above; these are the supplier's marked locations the placement was derived from. Source files are in _infofiles/.

Show all 26 breach-location figures

Defect ledger — every quarantined scenario, forever

36 scenarios across 6 dams were quarantined on 2026-08-16 for running with a single blanket Manning's n of 0.060 instead of spatially varying roughness from land cover. Mass balance is blind to roughness, so every outcome gate had passed them. Rows in this ledger are permanent and only move forward — a fixed scenario stays listed as DEFECT FIXED with its replacement's sha, so anyone holding earlier numbers can see, in this table, that they are superseded. If a scenario you are using appears here, compare your sha against the replacement's before quoting anything. 10 fixed · 26 replacements in flight.

Dam	Run id	status	new vol err %	new distinct n	replacement sha256	old sha256 (do not use)
Bagad	S_EG_05_SBL	running again	—	—	—	2fc9fb9d1afa41f7
Goma	S_EG_00	running again	—	—	—	10d56ed9058e4d35
Goma	S_EG_00_REF	running again	—	—	—	a5a369c0fb04929d
Goma	S_EG_01	running again	—	—	—	f545df60e6b42a7a
Goma	S_EG_03	running again	—	—	—	555a7a9ab7b88930
Goma	S_EG_05_SBL	running again	—	—	—	7b5b974eb459c85d
Goma	S_EG_05_SBR	running again	—	—	—	f17969d51c1b155f
Khodiyar	S_EG_00	running again	—	—	—	18a65eaa0321baaa
Khodiyar	S_EG_00_REF	running again	—	—	—	0e83a9aea78a3617
Khodiyar	S_EG_01	running again	—	—	—	e1907260c5a67290
Khodiyar	S_EG_02	running again	—	—	—	c7a90d8e90ad8f75
Khodiyar	S_EG_05_SBL	running again	—	—	—	088cd430a226575a
Khodiyar	S_EG_05_SBR	running again	—	—	—	d9976f1b24fb4b01
Khodiyar	S_EG_06	running again	—	—	—	8d82d8b5d07cab68
Munjiyasar	S_EG_00_REF	running again	—	—	—	3eb90e0a4c05cfb5
Munjiyasar	S_EG_01	running again	—	—	—	e60571113f291e30
Munjiyasar	S_EG_03	running again	—	—	—	125657789a83685f
Munjiyasar	S_EG_05_SBL	running again	—	—	—	0d87bcfc6e3816d0
Munjiyasar	S_EG_05_SBR	running again	—	—	—	7e2141c7904ee5b6
Varanshi	S_EG_00	running again	—	—	—	17ddd31c291f00a2
Varanshi	S_EG_00_REF	running again	—	—	—	057f5eb179571658
Varanshi	S_EG_01	running again	—	—	—	364dddd9d910d6f5
Varanshi	S_EG_02	running again	—	—	—	6f52e1111a73e9b3
Varanshi	S_EG_05_SBL	running again	—	—	—	540310fbae316bc3
Varanshi	S_EG_05_SBR	running again	—	—	—	c6614ac795bbdb72
Varanshi	S_EG_06	running again	—	—	—	90f2d2777e1f2406
Bagad	S_EG_00	DEFECT FIXED	0.2647	7	549e34b4af9b3065	2d362b4ba7f19eba
Bagad	S_EG_00_REF	DEFECT FIXED	0.0037	7	931f18ef6853e7fb	ac1e5efa28ce5f71
Bagad	S_EG_01	DEFECT FIXED	0.0155	7	8e8a1f5a8e6aa042	f1a004fdabb6189e
Bagad	S_EG_03	DEFECT FIXED	0.0484	7	0c5893bdd8e426b4	6dea657cae9124a4
Bagad	S_EG_05_SBR	DEFECT FIXED	0.0450	7	7e8fc6932959e0b4	507d1dbc6c24d79c
Kaniyad	S_EG_00_REF	DEFECT FIXED	0.0031	8	2f0c96e9a05d96fd	83552135c5e4663d
Kaniyad	S_EG_01	DEFECT FIXED	0.0703	8	e1d8163f764b118c	fff8c4fb2e0612ec
Kaniyad	S_EG_05_SBL	DEFECT FIXED	0.0803	8	295cabfb372d4db1	04882b3d8611cc9f
Kaniyad	S_EG_05_SBR	DEFECT FIXED	0.0804	8	df44db750e4fe56e	f45e302c00d7fecc
Kaniyad	S_EG_06	DEFECT FIXED	0.0712	8	4220cddec6e83c0e	5cb8663627536274

Superseded — never report from these

Retired in the register's band column and excluded from the primary store.

Dam	Retired run id	width m	Disposition
Bhimdad	S_E_05	61.70	replaced by k=4 CWC envelope, 2026-08-13
Dhatarwadi-II	S_EG_05	20.43	replaced by k=4 CWC envelope, 2026-08-13
Hamirpara	S05	23.18	replaced by k=4 CWC envelope, 2026-08-13
Hamirpara	S06	23.18	replaced by k=4 CWC envelope, 2026-08-13
Hanol	AS-01	81.13	replaced by k=4 CWC envelope, 2026-08-13

Relations and citation

EARTHEN            B_avg = 0.23 · k_M · V_w^(1/3)        k_M 1.0 piping / 1.5 overtopping
                   W_b   = B_avg − m · H_b                m   0.6 piping / 1.0 overtopping
                   t_f   = 60 · √(V_w /(g·H_b²))          seconds → hours,  g = 9.81

PEAK               Q_emp  = 0.0175 · k_M · k_H · √(g·V_w·H_w·H_b² / W_avg)   k_M 1.00 / 1.85
                   Q_max  = (8/27)(L_a/B_avg)^0.28 · [B_avg − m(H_b − 0.8H_w)] · √(g·H_w³)
                   Q_semi = Q_max · [1/(1 + 0.000045·t_f·√(g/H_b))]^β,  β = 500(W_avg·H_b²/V_w)^(2/3)

STRUCTURAL         gated   envelope = k · bay + (k−1) · pier,   k = min(4, n_bays)
                   ungated 4 monoliths × 15 m = 60.00 m
                   fixed   t_f 0.300 h · vertical sides · C_w 1.44 · trigger at FRL

Derived, not received: L_a is the centreline length of the SA/2D connection carrying the breach, read from the accepted model. W_avg is a trapezoidal section at 3.0:1 u/s and 2.5:1 d/s — a declared programme convention worth about ±8 %. Both are per scenario: W_avg = W_c + 2.75·H_b so it moves with the invert, and L_a follows the connection that is breached. k = 4 failed bays is a programme convention; CWC fixes the unit that fails, not how many. Breach inverts are an input, not an output.

† Note on L_a — reported, with a caveat