Cut Bangladesh's ~300+ Annual Lightning Deaths with a MoDMR Early-Warning and Shelter Programme

Diagnosis

Lightning kills people in Bangladesh at a rate the curated note describes as "~300+ deaths/yr, highest per capita globally." Unlike floods or cyclones, which build over hours or days, a strike is a nowcast hazard: the window between a storm cell forming and a fatal flash is minutes, not hours. The note also tells us this is a recurring, structural toll, not a one-off event, and that it falls disproportionately on people who cannot get under cover fast, principally farmers, fishers, and day laborers working in open fields and water.

Two things make this urgent now. First, there is no current_state indicator value in the system, which means the Ministry of Disaster Management and Relief (MoDMR) is managing a top-ranked-per-capita killer without a live, verified count to manage against. Second, the response that works elsewhere, fast alerts plus a place to shelter, is cheap relative to the lives at stake and can be built on infrastructure (mobile networks, union-level offices, school buildings) that already exists. The binding gap is organizational, not technological: nobody owns the minute-scale alert-to-shelter chain end to end.

Recommended actions

1. Stand up a verified lightning-death registry (MoDMR). Mechanism: a MoDMR circular requiring every Upazila Nirbahi Officer to report each lightning fatality and injury through the existing disaster-reporting chain, deduplicated against district health and police records. Observable signal: a single official monthly figure that MoDMR publishes, replacing the current null state, so trend and seasonality become measurable.
2. Deliver minute-scale strike warnings to at-risk districts (MoDMR, using the meteorological and telecom agencies it can task). Mechanism: a cell-broadcast and SMS alert keyed to detected storm cells, pushed to handsets in the affected union, plus loud sirens or mosque public-address announcements where coverage is thin. Observable signal: measurable alert lead time before strikes in covered unions, and rising share of fatalities occurring in unions not yet covered (which tells you where to expand next).
3. Build and map field shelters in the highest-toll unions (MoDMR, funded through its disaster-management budget line). Mechanism: simple grounded concrete shelters at intervals across open farmland and landing points, sited using the registry from action 1, with location signage so workers know where to run. Observable signal: shelters constructed per quarter in priority unions, and a falling share of deaths among people caught in the open.
4. Run a seasonal behavior campaign before the peak strike months (MoDMR, with local government). Mechanism: union-parishad-level messaging on the "30-30 rule" of seeking cover, delivered through imams, schools, and agricultural extension officers ahead of the pre-monsoon season. Observable signal: recall surveys showing workers know where the nearest shelter is and when to stop work.
5. Pilot lightning-protection retrofits on rural schools and markets (MoDMR, coordinating with local government engineering). Mechanism: grounded rods on public buildings already used as shelters, so the structures people run to are themselves safe. Observable signal: zero strike casualties recorded inside retrofitted buildings.

Sequencing (first 12 months)

Start with the registry (action 1): without a verified count, MoDMR cannot target shelters, justify budget, or prove the programme works. The registry unlocks shelter siting (action 3) by identifying the highest-toll unions. In parallel, launch the alert pipeline (action 2), because it saves lives immediately even before any concrete is poured. Time the behavior campaign (action 4) to land just before the pre-monsoon strike season. Treat school retrofits (action 5) as a bounded pilot whose results decide year-two scale-up.

Risks and constraints

The binding constraint is fiscal and administrative, not scientific. Shelters and retrofits compete for the same disaster-management budget as flood and cyclone work, so MoDMR must protect a dedicated line rather than fund this from residuals. Alert delivery depends on telecom and meteorological agencies MoDMR does not directly control, so the circular must assign clear obligations or the chain breaks at the handoff. Underreporting is the quiet risk: if the registry stays incomplete, the per-capita toll cited in the note will keep being managed blind.

Bottom line

A killer the curated note ranks highest per capita globally is being managed by MoDMR with no live count, and the fix, fast alerts plus somewhere to shelter, is affordable and buildable on existing infrastructure. Begin with the death registry to make the problem visible, then push alerts and shelters into the unions that registry exposes.