Meltwater Exposure Tracker

1. What this tracker is, and what it is not

It is a public, basin-resolved register of Himalayan cryosphere condition set against the economic and institutional exposure sitting downstream of it.

It is not a forecast, a hazard warning system, an early warning service, or a substitute for statutory monitoring by CWC, NDMA or state agencies. The tracker must carry this disclaimer on every page. HEF must never occupy a position where a user could reasonably construe its output as operational safety guidance.

Editorial rule: the tracker reports condition and exposure. It does not predict events. Any language implying event prediction is a publication error and should be treated as one.

2. Scope

2.1 Geographic scope — v1

Scope exclusions are deliberate and should be stated publicly in the methodology note, so that scope is understood as a design decision rather than an oversight. The tracker covers basins; it does not cover jurisdictions upstream of the Indian Himalayan Region, and it does not model transboundary allocation.

2.2 Temporal scope

3. Indicator dictionary

Every indicator carries: id, name, definition, unit, basin_resolution, source, source_type, publication_lag, last_verified, confidence.

3.1 Sub-index A — Cryosphere Condition (CC)

3.2 Sub-index B — Economic Mass (EM)

3.3 Sub-index C — Institutional Absorption (IA)

3.4 Event log

Separate table. One row per recorded cryosphere-driven loss event.

event_id, date, basin, sub_basin, event_type (GLOF / avalanche / debris flow / flash flood / other), trigger_mechanism, peak_discharge_m3s, sediment_volume_m3, generation_capacity_affected_MW, capacity_destroyed_MW, asset_owner, ownership_class, fatalities_range, persons_affected, roads_damaged, bridges_damaged, subsequent_ownership_change (bool), source_refs[], figure_conflicts (free text).

The figure_conflicts field is mandatory and is a feature, not a caveat. The October 2023 Sikkim event alone carries divergent published casualty figures. A tracker that silently picks one number is less credible than one that shows the divergence. Reference entry for that event should record: 1,200 MW Teesta-III destroyed; peak discharge approximately 7,355 m³/s at Chungthang; approximately 270 million m³ sediment eroded; flow travelled approximately 385 km; subsequent majority-stake sale by the state developer.

4. Data schema

{
  "basin_id": "IN-GNG-BHG",
  "basin_name": "Bhagirathi",
  "parent_basin": "Ganga",
  "states": ["Uttarakhand"],
  "reporting_year": 2026,
  "data_completeness_pct": 71,
  "cryosphere_condition": {
    "glacierised_area_sqkm": null,
    "glacier_count": null,
    "area_change_pct_since_1990": null,
    "snow_persistence_anomaly_pct": null,
    "snow_persistence_5yr_mean_pct": null,
    "consecutive_below_normal_seasons": null,
    "snowmelt_runoff_share_pct": null,
    "source_refs": []
  },
  "economic_mass": {
    "hydro_operational_mw": null,
    "hydro_under_construction_mw": null,
    "hydro_allotted_mw": null,
    "lean_season_generation_mu": null,
    "lean_season_plf_pct": null,
    "irrigated_command_ha": null,
    "population_exposure_zone": null,
    "asset_value_exposure_zone_inr_cr": null,
    "source_refs": []
  },
  "institutional_absorption": {
    "sdrf_allocation_inr_cr": null,
    "centre_state_share": "90:10",
    "ownership_split": {"state": null, "central": null, "private": null},
    "early_warning_coverage": null,
    "insurance_penetration_pct": "not_disclosed",
    "water_levy_status": null,
    "source_refs": []
  },
  "provenance": {
    "last_verified": "YYYY-MM-DD",
    "verified_by": "",
    "changelog_ref": ""
  }
}

Null discipline. null means not yet collected. "not_disclosed" means sought and unavailable. "not_applicable" means structurally absent. These three must never be collapsed into one value, and the front end must render them differently.

5. Publication protocol

5.1 Annual cycle

5.2 Front-end requirements

1. Basin matrix as landing view. Three sub-index columns, never one composite score.

2. Completeness bar per basin, visible without interaction. A basin at 40% completeness must look different from one at 90%.

3. Every figure hover-traceable to source, date and licence.

4. Download in CSV and JSON, unrestricted. Citation reach is the objective; friction defeats it.

5. Permanent per-basin URLs — the citation unit is the basin page, not the tracker.

6. Versioned snapshots. A reader citing the tracker in 2027 must be able to retrieve exactly what it said in 2026.

7. Static server-side rendering of all headline figures. Values rendered only client-side are invisible to static fetchers and search crawlers, which forfeits the citation and indexing value the tracker exists to generate.

8. Changelog page, human-readable, listing every value revision with date and reason.

5.3 Correction policy

Published, linked from every page. Corrections are versioned, never silent. A visible correction record is an asset for an institution whose product is credibility.

6. Build sequence

Phase 1 — Aggregation (weeks 1–8). Ingest all published cryosphere and fiscal series. This produces a useful tracker with no proprietary component. Ship it.

Phase 2 — Concordance (weeks 6–16). Build the state→basin generation concordance (EM-01 to EM-05). This is the first genuinely proprietary output and the first thing that cannot be replicated by aggregating public sources. Publish the concordance table itself as a standalone dataset.

Phase 3 — Exposure zones (months 4–12). Population and asset value in modelled exposure zones (EM-07, EM-08). Highest effort, highest defensibility. Requires an explicit, published methodology and named technical review before any value is released.

Phase 4 — Institutional layer (months 6–18). Insurance and absorption data via direct engagement. Expect partial success. Publishing the shape of the gap is itself a finding.

Sequencing principle: never let Phase 3 delay Phase 1. A partial tracker that publishes on schedule builds more institutional standing than a complete one that publishes late.

Annexe A — Source register

Annexe B — Comparative reference basins

Two non-Indian mountain basins to be selected for methodological validation, on the criteria that they (a) have long instrumented hydrological records, (b) carry material downstream generation or irrigation exposure, and (c) sit in jurisdictions outside the region, so that the MEI methodology can be shown to be portable rather than bespoke. Candidate ranges: Andean and Alpine. Final selection to be made and published with the methodology note; not pre-committed here.