Underwriting Above the Snowline: GLOF Exposure for Himalayan Hydropower, Transmission and Project Cargo in 2026

On the night of 3 October 2023, the South Lhonak glacial lake in north Sikkim breached. A wall of water and moraine raced down the Teesta, overtopped and washed away the 60-metre Chungthang dam, and submerged the 1,200 MW Teesta-III project. The asset was insured by a consortium reportedly led by IFFCO-Tokio with a sum insured in the range of Rs 11,400 crore, and the loss became one of the largest single property events the Indian market has handled. Then the wording did what poorly structured wordings do: while flash flood and cloudburst losses reportedly carried full cover, the glacial lake outburst flood (GLOF) peril sat sub-limited at a figure widely reported near Rs 500 crore. The entire dispute that followed, with the Sikkim operator arguing cloudburst and a central committee finding GLOF, was a fight over which clause applied to a single physical event.

That is the lesson for every broker placing Himalayan risk in 2026. GLOF is no longer a remote academic peril. The NDMA National GLOF Risk Mitigation Programme, a roughly USD 20 million initiative, is now assessing and treating 195 high-risk glacial lakes across Sikkim, Ladakh, Uttarakhand, Himachal Pradesh and Arunachal Pradesh, expanded from an initial 56. Published hydrology placed 177 of 259 assessed Himalayan hydropower plants on potential GLOF run-out tracks, with up to a third facing discharges well above their local design floods. When the design flood itself understates the peril, the insurance contract has to carry the residual, and right now most of them carry it badly.

For a desk, the practical signal is simple. If your insured asset sits in a glaciated catchment above roughly 1,500 metres, GLOF is a named peril you must price and word on purpose, not a sub-clause you inherit from a generic STFI extension.

The Indian engineering and property markets price flood through the Storm, Tempest, Flood and Inundation (STFI) cluster, and the underlying actuarial picture assumes a rainfall-driven, catchment-scale event with some warning and partial damage. GLOF violates almost every one of those assumptions. The trigger is a moraine or ice collapse, not rainfall, so a dry-season burst is entirely possible. The hydrograph is a near-instant debris-laden surge rather than a rising river, which means velocity and scour, not just inundation depth, drive the damage. And the loss profile is binary: a powerhouse in the run-out path is rarely partially damaged, it is destroyed.

That mismatch creates three wording failure modes brokers should hunt for at every renewal.

• Silent peril. Many EAR, property and mega-risk wordings neither name nor exclude GLOF. Underwriters assume STFI captures it; insureds assume it is covered as flood. Neither has priced it, and the gap surfaces only at claim.
• Causation traps. Where GLOF is sub-limited but cloudburst or flood is fully covered, you have manufactured the exact Teesta-III dispute. The peril definitions overlap physically, so the placement invites a coverage fight that a forensic hydrologist, not the policy, ends up deciding.
• Sub-limit illusion. A Rs 500 crore GLOF sub-limit against a Rs 11,000 crore-plus asset is not cover, it is a token. For a total-loss peril, a sub-limit set far below the probable maximum loss leaves the developer and its lenders carrying the real risk while believing they transferred it.

A desk that wants to write Himalayan hydropower confidently needs an explicit GLOF assessment in the file, separate from the STFI rate. The good news is that the public hydrology and the NDMA inventory now make this possible without inventing numbers. Structure the underwriting around four questions.

Is the asset in a run-out path, and from which lakes? Map the insured location against the NDMA and ISRO glacial lake inventories upstream in the same catchment. A powerhouse 30 km below a Category A lake is a different risk from a diversion weir in a non-glaciated tributary. This is binary exposure screening before any rate is discussed.

What is the GLOF peak discharge versus the design flood? The structure was engineered to a design flood. If credible studies suggest a plausible GLOF surge exceeds that figure, the hydraulic margin of safety is negative and the loss given occurrence trends toward total. Ask the developer for the GLOF-specific dam break or surge study; its absence is itself an underwriting finding.

What protects the asset between lake and turbine? Distance, intervening reservoirs, sediment-trapping reaches and the height of the powerhouse above the riverbed all attenuate a surge. These are physical mitigations you can credit.

What early warning exists, and does it buy evacuation or shutdown time? Automated Weather and Water Stations and downstream Early Warning Systems under the NDMA programme relay lake-level and rainfall data at short intervals. Warning rarely saves the concrete, but it saves lives, can trigger protective gate operation, and materially cuts the liability and workers-compensation tail.

The output is not a single STFI loading. It is a documented GLOF position: exposed or not, sub-limit justified against an estimated maximum loss, and a defined set of credits the insured can earn. That file is what lets you defend the rate to a reinsurer and the cover to a lender.

Wording is where this risk is won or lost, and the Teesta-III causation fight is the cautionary tale. The objective is to remove the incentive and the room to argue about which natural event caused a single physical loss.

Start with a standalone peril definition. Define GLOF on a physical basis: a flood or debris flow originating from the partial or complete failure of a glacial or moraine-dammed lake, however the failure is initiated, including by rainfall, seismic activity, avalanche or moraine collapse. Drafting it on the source of the water, rather than on the meteorological trigger, closes the cloudburst-versus-GLOF gap that an adjuster would otherwise exploit. A debris flow that began as a moraine breach should not be re-characterised as ordinary flood merely because rain was also falling.

Then make three drafting choices deliberately. First, set any GLOF sub-limit against a genuine estimated maximum loss for the exposed asset, not a round number that looks reassuring on a schedule. For a powerhouse in a direct run-out path, the honest EML is close to a total loss, and the sub-limit should reflect that or the cover should be repriced to remove it. Second, align the material damage and the business interruption or advance loss of profits triggers so a GLOF that triggers the property section also triggers the consequential-loss section without a separate causation argument. Third, write any early-warning or mitigation condition as a clear warranty or condition precedent with defined consequences, so both sides know whether a failed siren voids cover or merely reduces a credit. Vague best-endeavours language helps nobody at claim.

Himalayan reconstruction is where GLOF business interruption cover goes wrong on the indemnity period. A washed-out powerhouse at altitude is not rebuilt in twelve or eighteen months. The site is remote, the construction season is short, access roads are often destroyed in the same event, the original glacial hazard still sits upstream, and regulators may demand a fresh dam-safety and environmental review before reconstruction is even permitted. Realistic reinstatement of a major Himalayan hydro asset runs in years, not quarters.

This matters in two distinct phases. During construction, advance loss of profits (ALOP), also called delay in start-up (DSU), attaches to the erection-all-risks programme and pays the developer's lost revenue if a GLOF delays commercial operation. Underwriters routinely sell DSU indemnity periods that assume normal site logistics. For a glaciated catchment, that assumption is wrong, and an indemnity period that runs out while the access road is still being rebuilt leaves the lender exposed exactly when the project economics are most fragile. We cover the indemnity-period mechanics in detail in our DSU and ALOP guide for renewable and industrial projects and in the contractors and erection all-risks ALOP framework for infrastructure.

Once the plant is operating, the same logic flows into the property and business interruption programme. The maximum indemnity period must reflect Himalayan reinstatement timelines, not a generic figure copied from a plains thermal plant.

Press the developer and the lender to model the realistic reinstatement clock: demolition, debris clearance, access restoration, fresh hydrological clearance, the short build season, and re-commissioning. Then set the indemnity period above that, with margin. An indemnity period that is shorter than the physical reality is a silent retention the borrower discovers only after the lake has already burst.

The hydropower asset is most fragile before it ever generates a rupee. Turbines, generators, transformers and penstock segments are oversized, high-value and effectively irreplaceable on any short timeline. They move by sea to an Indian port, then overland on long, single-route Himalayan roads, and finally sit on a riverbank construction site for months during erection. A GLOF that hits during the construction window can destroy uninstalled critical-path equipment and reset the entire project schedule.

This is where the marine and engineering programmes must be read together rather than as separate silos.

• Project cargo and transit. The marine cargo or project cargo policy covers the equipment in transit, but the storage and pre-installation phase at site is the GLOF-exposed window. Confirm the wording covers storage at the project site and aligns with the EAR attachment, so there is no gap on the day a transformer sits on the riverbank waiting to be set.
• Marine delay in start-up. If the lost item is on the critical path, a transit or storage loss can delay commercial operation just as surely as a construction accident. Marine DSU bridges that revenue gap, and its trigger and indemnity period should be set with the same Himalayan reinstatement logic as the engineering ALOP. Our marine DSU and project cargo guide sets out how to structure it.
• Erection all-risks handover. The seam between the cargo policy ending and the EAR policy attaching is a classic gap. Make sure GLOF is a named, consistently treated peril on both sides of that handover, so a surge during erection does not fall between two insurers each pointing at the other.

For a lender financing the project, the cargo and erection phase is the period of highest loss-given-event with zero revenue to absorb it. A broker who maps the GLOF peril continuously from sea voyage through storage, erection and operation, rather than at each policy in isolation, is selling something most desks cannot.

The strongest moment to fix GLOF cover is before financial close, when the lender still holds real bargaining power over the developer and the insurance structure is being set. A broker who arrives with a worked GLOF position changes the conversation from price to risk quality. A practical sequence:

1. Screen the location against the glacial lake inventories. Establish, in writing, whether the asset sits in a GLOF run-out path and from which catalogued lakes. This single finding drives everything downstream.
2. Obtain the GLOF-specific surge or dam-break study. If the developer cannot produce one, that is a finding the lender needs, not a detail to wave through. The absence of the study is itself a risk signal.
3. Force the wording question at quotation. Is GLOF covered, sub-limited or excluded, on what peril definition, and does it align material damage with business interruption and ALOP. Get the answer before binding, not at claim.
4. Right-size the sub-limit to a real EML. For a powerhouse in a direct run-out path, refuse a token sub-limit. Either price the full exposure or document that the developer and lender are knowingly retaining the gap.
5. Credit genuine mitigation. Early warning systems, upstream lake-lowering works under the NDMA programme, protective structures and powerhouse elevation are real risk reducers. Build them into the rate and the conditions so the insured has an incentive to maintain them.
6. Stress-test the indemnity period. Model the Himalayan reinstatement clock and set ALOP and operational BI periods above it.

The market for this is thin precisely because almost no Indian desk prices GLOF with confidence. That is the opportunity. India's hydropower pipeline in the eastern and western Himalaya is large and lender-driven, and every project needs a defensible GLOF answer before the banks release funds. The broker who can structure the peril, the wording, the sub-limit and the indemnity period as one coherent position owns a niche that the next ordinary STFI quote cannot touch.