Odisha Got 30% Excess Rain and UP Saw Deadly September Floods in a National Drought Year: Why Flood Cover Is Not Optional

The 2026 southwest monsoon closed on 30 September as a below-normal season by every national measure. Skymet classified it as a 'mild drought' with a 13 per cent seasonal deficit. The India Meteorological Department's end-of-season figures, as reported the same day, put all-India rainfall at about 759.4 mm against a normal of about 868.6 mm, or roughly 87 per cent of the long-period average (LPA).

That is the number that will appear in board papers, renewal notes and broker presentations over the next six months. It is also the wrong number to use when deciding whether a commercial property programme needs flood cover in April 2027.

Two states make the point from the same season. Skymet's tally shows Odisha finishing with a 30 per cent surplus over its LPA in a year when the country as a whole fell short by 13 per cent. And in the last week of September, days of heavy rain and flooding in Uttar Pradesh killed at least 62 people and damaged more than 1,000 houses, with Kanpur among the hard-hit cities and schools closed in Lucknow (AP, 28 September 2026). In Odisha, more than 75,000 people were evacuated and floods and landslides damaged 194 houses over the same period.

In June, our piece on [the dry-monsoon underwriting trap](/underwriting-risk/below-normal-monsoon-2026-dry-year-flood-underwriting-trap-india) argued at the forecast stage that a below-normal outlook should not loosen flood underwriting. This post closes that argument with the outcome data. The season did what the forecast-stage argument said it could: it came in dry nationally and still produced flood events severe enough to kill people, damage buildings and disrupt business in two large states.

An all-India seasonal figure is an average across 36 meteorological subdivisions and four months. Commercial flood loss happens at one address, over a few days. The two measures are related, but loosely, and 2026 shows how loosely.

IMD's end-of-season distribution, as reported on 30 September, had 18 of 36 subdivisions in the normal band and 17 deficient. Read quickly, that sounds like a country where half the map was dry. Read as an underwriter, it says something different: half the subdivisions received normal rainfall, and a normal monsoon still brings the heavy spells that flood the usual places. The national deficit was carried largely by the other half.

Three ways a deficit year still floods a factory

1. Regional surplus inside a national deficit. Odisha's 30 per cent surplus is the clearest case. A risk in coastal or deltaic Odisha experienced a wet year, regardless of the all-India figure.
2. Late-season concentration. The Uttar Pradesh floods arrived in the final week of September, at the very end of the season. A monsoon that underdelivers for three months can still deliver its heaviest spell at the close, and a seasonal total says nothing about when the rain fell.
3. Urban drainage limits. Kanpur and Lucknow are large urban and industrial centres. In cities, flood damage depends on how much rain falls in a short window relative to drainage capacity, not on whether the season as a whole was above or below normal.

None of these mechanisms is visible in a 13 per cent national deficit. All three were visible in the September news.

The risk now is not in underwriting rooms first. It is in the buyer's finance function. A CFO looking at a 'drought year' headline and a property premium that includes flood loading will ask a reasonable-sounding question: why are we paying for a peril that did not show up?

For many buyers in the 17 subdivisions that ended the season deficient, the honest answer may be that their own locations did not flood this year. That is not a reason to drop the cover. Flood is a low-frequency, high-severity peril, and one dry season says almost nothing about the next one. The premium buys protection against the year when the national average is irrelevant because a single spell over one catchment does the damage.

For buyers in Uttar Pradesh and Odisha, the 2026 data cuts the other way and should end the conversation quickly. In a season labelled a drought, these are the states where people died, homes were damaged and schools shut.

On the insurer side, the same headline creates a different temptation: to discount flood loading in competitive tenders on the basis that the season was dry. Our analysis of how the monsoon loss tally feeds the April 2027 treaty renewal explains why reinsurers will be reading the event-level losses, not the seasonal average. Direct underwriters who price off the average will find their pricing out of step with their treaty costs.

The useful work for April 2027 is to replace the national figure with location-specific evidence. For each insured site, underwriters and buyers should be able to answer a short set of questions before the sum insured and flood terms are agreed.

What to put in the renewal submission

• District and subdivision rainfall outcome for 2026. Whether the site sat in a normal, deficient or surplus subdivision this season, and whether there was a short, heavy spell even in a deficient year.
• Flood history at the address, not the state. Any water entry, access disruption or drainage backup in the last several seasons, including events that did not lead to a claim.
• Site elevation and drainage relative to surroundings. Whether the plant sits in a low-lying industrial area, near a river or nala, or in an urban zone with known drainage limits.
• Values at ground level. Stock, raw materials, finished goods and machinery stored at or near floor level, which is where flood losses concentrate.

Matching values to exposure

The point of the exercise is to set flood terms that reflect where the money actually sits. A warehouse in Kanpur with most of its stock on the floor has a different flood profile from an office block on raised ground in the same city, even if both are in a subdivision that ended the season close to normal. Sums insured should reflect current replacement values, because flood claims are where underinsurance shows up most painfully: under an average clause, a site that is underinsured at the time of loss recovers only a proportion of the claim.

For underwriters, the same data feeds the flood loading and the accumulation picture. A portfolio with several sites in the Kanpur-Lucknow belt or in coastal Odisha is carrying correlated exposure that a single event can hit at once, which is exactly what the September floods did.

Buyers who want to reduce premium after a dry year have a better lever than deleting flood cover: the deductible. A higher deductible lowers the premium by moving smaller, more frequent losses onto the insured's balance sheet while keeping protection against the severe event.

That trade-off works only if the buyer understands how the deductible is applied. Our earlier piece on STFI deductibles, sub-limits and inundation wording set out how these three levers stack on a single claim. The short version for April 2027:

1. Check whether the STFI deductible is a percentage of the claim or a fixed amount. A percentage deductible grows with the loss, which matters most on exactly the large events the cover is meant for.
2. Check whether a flood sub-limit applies, and what it is. A sub-limit set below realistic flood values defeats the purpose of buying the cover.
3. Read the inundation wording. Whether water entry from blocked urban drains, rather than a river breaking its banks, is treated as covered inundation is often the question that decides a city-flood claim.

A buyer who raises the deductible, keeps the full flood limit and confirms the inundation wording has made a rational trade. A buyer who drops flood because the season was dry has not.

For small and medium enterprises in Uttar Pradesh and Odisha, the immediate priority is not renewal. It is the claim from the September event. Flood claims are won or lost on documentation gathered in the first few weeks, while water marks are still visible and stock records are still fresh.

The evidence file

• Dated photographs and video of water levels inside the premises, damaged stock, machinery and building elements, ideally showing the height of the water line on walls or racking.
• Stock records as of the date of loss. Purchase invoices, sales records and stock registers that let the surveyor reconstruct what was on site when the water came in.
• Intimation records. The date and mode of claim intimation to the insurer, and copies of any acknowledgement.
• Local evidence of the event. District administration notices, school-closure orders and news reports that establish the date and severity of flooding in the area.
• Mitigation steps taken. Records of pumping, moving stock to higher ground or securing machinery, which support the claim and show the insured acted to limit loss.

Our commercial flood claims guide covers the claim process in more detail. The same evidence file also becomes the strongest renewal submission a year from now, because it documents the site's actual flood behaviour in a real event.

In June, the case against loosening flood underwriting rested on mechanism: seasonal totals and flood losses are only loosely linked, so a below-normal forecast should not drive flood terms. In October, the case rests on outcome. The country finished at about 87 per cent of LPA and Skymet called it a mild drought. Odisha finished 30 per cent above normal. Uttar Pradesh lost at least 62 lives to late-September flooding, with damage across Kanpur and disruption in Lucknow.

For April 2027, the practical rules are simple:

1. Do not use an all-India deficit as evidence for dropping or cutting flood cover.
2. Use district-level rainfall, site flood history and ground-level values to set sums insured and flood terms.
3. If premium needs to come down, use a higher deductible rather than a lower limit or a deleted peril.
4. In UP and Odisha, close out the September claims with full documentation first, then bring that evidence to the renewal table.

The national rainfall figure is a good measure of agricultural stress and water supply. It is a poor measure of whether a specific factory, warehouse or shop will flood next year. Renewal decisions should use the second question, not the first.