On stay dates where a typhoon is approaching, room inventory that had once drained away turns around and starts increasing again just before check-in. Operators know the phenomenon anecdotally, but few have the numbers on which lead time, how many properties, and how many rooms. This article measures that directly: for Typhoons No. 9 (Bavi), No. 13 (Dolphin) and No. 15 (Chan-hom), which approached or made landfall in Japan in July and August 2026, we tracked remaining-inventory movement by lead time across 1,723 properties (82,546 rooms) and 9,658 property-stay-date observations in Okinawa, Kagoshima, Ibaraki, Chiba, Shizuoka, Kochi and Miyazaki.

HotelBank has previously published Typhoon-Heavy Aug–Sep 2026: ‘Indoor-Complete’ Hotels Defend Occupancy, but that article dealt with the quality of demand — which property types can hold occupancy through typhoon season. This article is not about that. It isolates one inventory-side phenomenon only: after a typhoon becomes a certainty, how already-sold inventory comes back over time. The goal is not to judge whether demand is strong or weak, but to measure where in the booking curve resale opportunity appears, and at what scale.

The typhoons and stay dates covered — three events in July–August 2026

Start by fixing the meteorological facts. Based on the Japan Meteorological Agency’s “Typhoon Position Table, Reiwa 8 (2026)” and contemporaneous reporting, three typhoons materially affected Japanese lodging demand in July and August 2026. All are past events whose stay dates have already elapsed; this article makes no forecast about future dates.

Table 1: The three typhoon events of July–August 2026 covered here, and the stay dates counted as typhoon-approach dates

Typhoon	Main impact dates	Meteorological course	Stay dates treated as typhoon-approach dates
Typhoon No. 9 (Bavi)	July 11, 2026	Passed near Tarama Island in the Miyako Islands around 09:00 on July 11 at 945 hPa, large and strong, then moved into the East China Sea	Fri July 10 and Sat July 11 / Okinawa
Typhoon No. 13 (Dolphin)	August 6–8, 2026	Closest approach north of the Daito Islands at 945 hPa on the night of August 6 (around 21:00–22:00); passed near Nakijin on Okinawa Island at 935 hPa around 15:00 on August 7. Storm-force winds on Okinawa Island lasted roughly 40 hours, with the storm warning lifted at 13:57 on August 8	Thu August 6 – Sat August 8 / Okinawa and Kagoshima
Typhoon No. 15 (Chan-hom)	August 11–12, 2026	Made landfall in southern Ibaraki around 20:00 on August 11 — the first landfall in Ibaraki since records began in 1951. Crossed Honshu and exited into the Sea of Japan from Tsuruga Bay around 06:00 on August 12	Tue August 11 and Wed August 12 / Ibaraki, Chiba and Shizuoka

Source: Compiled by the HotelBank Editorial Team from the Japan Meteorological Agency’s “Typhoon Position Table, Reiwa 8 (2026)”, Japan Weather Association (tenki.jp), Weathernews and Ryukyu Shimpo (all accessed August 25, 2026)

Comparison normal days were drawn from the same prefecture, same day of the week, within one to three weeks before or after. Matching the day of week matters because booking-curve shape varies sharply by weekday. For Typhoon No. 13 in Okinawa (Thu 8/6, Fri 8/7, Sat 8/8) we used the Thursdays, Fridays and Saturdays between July 23 and August 1; for Typhoon No. 15 (Tue 8/11, Wed 8/12) we used the Tuesdays and Wednesdays of August 4–5 and August 18–19. The Obon period (August 13–16) has a distinctive demand structure and was excluded from both the typhoon and the normal-day sets.

In Okinawa, the T-3 to T-2 return rate ran 3.1x normal days — 48.3% versus 15.6%

The headline first. Comparing the three days directly hit by Typhoon No. 13 in Okinawa (August 6–8; 296 properties, 779 property-stay-dates) against six normal days in the same prefecture (302 properties, 1,415 property-stay-dates), the gap widens as lead time shortens.

Table 2: Return incidence rate and median return volume (as a share of total rooms) by lead-time band — Okinawa, Typhoon No. 13 approach dates vs normal days

Lead-time band	Typhoon No. 13 dates return incidence	Normal days return incidence	Ratio	Median return volume (typhoon / normal, % of total rooms)
T-30 to T-15	58.7%	52.1%	1.13x	5.6% / 6.2%
T-14 to T-8	51.8%	42.0%	1.23x	5.0% / 4.3%
T-7 to T-4	54.2%	24.4%	2.22x	8.8% / 3.1%
T-3 to T-2	48.3%	15.6%	3.10x	10.5% / 2.8%
T-1	14.0%	4.5%	3.11x	5.3% / 1.7%

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research / Okinawa; properties whose OTA-listed allotment reached at least 30% of total rooms only. Typhoon No. 13 dates N=296 properties (779 property-stay-dates); normal days N=302 properties (1,415 property-stay-dates)

Between T-30 and T-8 the two series stay within 10 points of each other. But from T-7 onward the normal-day return rate falls to 24.4% while typhoon-approach dates hold high at 54.2%. The gap is widest at T-3 to T-2: 48.3% against 15.6%, a ratio of 3.10x. Median return volume also diverges — 10.5% of total rooms versus 2.8%, roughly 3.8x — so the difference is not only in how often returns happen but in how large each one is.

There is a second, more intuitive measure: the share of properties whose remaining inventory at T-2 was simply higher than at T-7. On Typhoon No. 13 approach dates that was 47.6% — nearly one property in two had more rooms open two days out than a week out. The same measure on normal days is just 10.1%.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research / Okinawa; five typhoon-approach dates (Typhoons No. 9 and No. 13) N=333 properties (1,286 property-stay-dates), six normal days N=304 properties (1,417 property-stay-dates)

Viewed daily, the return incidence rate on typhoon-approach dates rises gradually from T-14, peaks at 31.2% on T-3, and then collapses quickly. Normal days simply track flat in a 7–10% band over the same stretch, forming no discernible peak. The exogenous shock of a typhoon carves a clear spike into the tail end of the booking curve, and nowhere else.

The area-wide remaining-inventory rate reverses — Okinawa runs from 18.5% at T-7 to 25.2% at T-3

What happens when you aggregate the whole area rather than looking at individual properties? Summing remaining inventory across covered properties and dividing by total rooms gives an “area remaining-inventory rate”; plotted by lead time, the contrast between normal days and typhoon-approach dates is sharper still.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research / Okinawa: typhoon-approach dates N=333 properties, normal days N=304 properties. Kagoshima: Typhoon No. 13 approach dates N=171 properties, normal days N=184 properties. All limited to properties whose OTA-listed allotment reached at least 30% of total rooms

On normal days the area remaining-inventory rate declines monotonically — from 29.6% (T-30) to 12.5% (T-1) in Okinawa, and from 38.7% to 19.5% in Kagoshima. A textbook booking curve. On typhoon-approach dates the direction changes midway. Okinawa bottoms at 18.5% at T-7, reverses, and builds back 6.7 points to 25.2% by T-3. Kagoshima does the same, bottoming at 26.3% at T-11 and recovering 5.8 points to 32.1% at T-3.

What matters here is that the reversal occurs at the level of the area aggregate. If it were a one- or two-room swing at individual properties, observation noise would be a reasonable suspicion. But when totals covering roughly 16,000 rooms in Okinawa and 7,900 rooms in Kagoshima are pushed back by several percentage points, a substantial volume of rooms has genuinely returned to sellable status across the market.

Where the return peak forms differs by typhoon — it tracks the moment the forecast firms up

Lining up the three typhoons, the lead time at which returns peak is not the same.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research / Okinawa: Typhoon No. 13 N=296 properties, normal days N=302 properties. Kagoshima: Typhoon No. 13 N=171 properties, normal days N=183 properties. Ibaraki, Chiba and Shizuoka: Typhoon No. 15 N=782 properties, normal days N=981 properties

Table 3: Return incidence rate by lead-time band, by typhoon and prefecture, plus the share of properties with a net inventory gain from T-7 to T-2

Segment	T-14 to T-8	T-7 to T-4	T-3 to T-2	T-1	Properties with net inventory gain T-7 → T-2
Okinawa — Typhoon No. 13	51.8%	54.2%	48.3%	14.0%	47.6%
Okinawa — normal days	42.0%	24.4%	15.6%	4.5%	10.1%
Okinawa — Typhoon No. 9	38.9%	44.1%	36.7%	12.5%	37.7%
Kagoshima — Typhoon No. 13	54.1%	49.7%	29.1%	7.3%	35.0%
Kagoshima — normal days	31.8%	24.2%	23.6%	9.3%	18.5%
Ibaraki, Chiba, Shizuoka — Typhoon No. 15	45.7%	30.6%	18.9%	10.1%	12.3%
Ibaraki, Chiba, Shizuoka — normal days	39.5%	27.3%	12.9%	4.8%	9.9%

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research / properties whose OTA-listed allotment reached at least 30% of total rooms only; same-day (T-0) observations excluded

Typhoon No. 13 in Okinawa builds a long, high plateau from T-7 through T-2. The same typhoon in Kagoshima peaks at 30.1% at T-8 and eases back to 29.1% by T-3 to T-2. Ibaraki, Chiba and Shizuoka, where Typhoon No. 15 made landfall, differ from normal days by only 3.3 points at T-7 to T-4 — yet the gap opens right at the end: 18.9% against 12.9% at T-3 to T-2, and 10.1% against 4.8% at T-1.

Map that difference onto the character of each storm and it reads clearly. Typhoon No. 13 formed on July 27 and did not pass the Daito Islands until August 6 — more than ten days during which a track toward the Nansei Islands was signalled early. Travellers could decide to change plans early, and returns build from T-7. Typhoon No. 15, by contrast, travelled westward from the sea east of Japan — the reverse of the usual direction — and its landfall in Ibaraki was the first there since records began in 1951. When a track firms up late, the decision to move an itinerary is pushed back by the same amount. The lead time at which the return peak forms appears to be tied closely to the point at which the typhoon’s track was communicated with confidence.

The practical implication is straightforward. Rather than assuming uniformly that “inventory comes back from T-3 when a typhoon hits”, it fits reality better to take the day the track information firmed up as the starting point, and expect the resale peak two to four days after that. In practice that means preparing well in advance for Nansei Islands tracks, where forecasts settle early, and staying ready right up to the day before for Honshu-crossing tracks, where the path is harder to read.

Inventory comes back in blocks — 10-plus-room releases account for 14%

Look at the size of each individual return and inventory on typhoon-approach dates comes back in a completely different shape from normal days.

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research / return events observed between T-7 and T-1. Okinawa: typhoon-approach dates N=1,951 events, normal days N=710 events. Ibaraki, Chiba and Shizuoka: Typhoon No. 15 N=1,016 events, normal days N=1,696 events

On normal days in Okinawa, 77% of returns are small — one or two rooms. Cancellations trickle in day by day and inventory trickles back: an entirely ordinary cycle. On typhoon-approach dates the one-to-two-room share falls to 55%, and in its place five-to-nine-room returns take 15% and 10-plus-room returns take 14%. Since blocks of 10 or more make up just 2% on normal days, that is a seven-fold increase in frequency. Among the observed cases: a 205-room property where 48 rooms returned to sellable status at T-9 and 51 more at T-10, and a roughly 220-room property where 83 rooms came back at once at T-3.

For Typhoon No. 15 in Ibaraki, Chiba and Shizuoka, the size mix barely differs from normal days (10-plus-room returns at 4% versus 4%). The storm took about ten hours from landfall to exit, so the impact was confined to a single night — it never escalated into group or multi-night block cancellations. Even within the category “typhoon”, the granularity of returning inventory differs entirely between a long-stalling Nansei Islands event and a Honshu-crossing one.

69% of returned inventory is relisted without touching the price

At what price does returning inventory come back to market? We compared the lowest listed rate at the moment a return was observed against the rate at the immediately preceding observation.

Table 4: Direction of change in the lowest listed rate at the moment of return — typhoon-approach dates vs normal days

Segment (returns observed T-7 to T-1)	Observations	Price lowered	Unchanged	Price raised	Median change
Okinawa — typhoon-approach dates	1,951	24%	69%	7%	±0.0%
Okinawa — normal days	710	23%	69%	8%	±0.0%
Ibaraki, Chiba, Shizuoka — Typhoon No. 15	1,016	28%	59%	13%	±0.0%
Ibaraki, Chiba, Shizuoka — normal days	1,696	23%	68%	9%	±0.0%

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research / lowest listed rate (two guests per room, per-room rate, tax included) compared against the immediately preceding observation. Changes within ±1% are treated as unchanged

In Okinawa, on both typhoon-approach dates and normal days, roughly 70% of inventory is relisted at an unchanged price. Even in the middle of an unmistakable demand shock, pricing behaviour looks no different from an ordinary week. Only Typhoon No. 15 in Ibaraki, Chiba and Shizuoka shows the unchanged share falling to 59%, with movement in both directions rising — 28% lowered and 13% raised — and even there, a majority still leaves the price alone.

This is where the revenue opportunity sits. T-3 to T-2 on a typhoon-approach date is the moment when inventory that was never in the original sales plan comes back — and comes back in a block on the order of 10% of total rooms. The price at which that inventory is returned to market is exactly the kind of decision where planning ahead converts speed directly into results. Even within the single category “return”, the right resale price is completely different for a stay date under an active storm warning versus a post-typhoon stay date after the warning has been lifted. As set out in Revenue Management vs Dynamic Pricing: 19,365 Hotels, Median 4 Moves, automating price moves and designing which conditions call for which price are two separate things. Typhoon resale pricing belongs firmly to the second — it is something to design in advance.

The effect spills into prefectures off the track — Kochi and Miyazaki hit 50.4% at T-7 to T-4

Typhoon No. 13 passed over Okinawa Island and exited into the East China Sea; Kochi and Miyazaki lay outside the central track of the storm-force wind area. Yet in those two prefectures too, stay dates of August 6–8 recorded a T-7 to T-4 return incidence rate of 50.4% — 20.2 points above the 30.2% on the same prefectures’ normal days (July 30 – August 1).

That gap cannot be explained by the presence or absence of storm-force winds alone. During Typhoon No. 13, flight and ferry cancellations were widespread, centred on services to and from the Nansei Islands, and transport and logistics on Okinawa Island came to a halt. Once part of an itinerary becomes impossible, the accommodation nights attached to it move in sympathy, even outside the wind field. When sizing up a typhoon’s impact, looking only at the storm-force wind area on the track map is not enough. Prefectures that share a transport network with the affected area will see inventory return even without a direct hit.

One caveat: in those same two prefectures, T-3 to T-2 came in at 28.3% against 32.7% on normal days — below the normal-day level. The spillover concentrates in the T-7 to T-4 band and does not persist to the final days. That is the opposite of Okinawa, on the central track, which stayed elevated right through the end.

Sizing the resale opportunity — 2.0 to 5.1 rooms per 100, with volume the stronger lever

Everything above is observed fact. To set an operational posture, though, it helps to convert that fact into how many rooms it means for your own property. What follows uses only the return incidence rates and median return volumes measured in this article; no new observations or external forecasts are added. We assume a single property with 100 total rooms and estimate the rooms available for resale at T-3 to T-2 on a typhoon-approach date.

Table 5: Expected resale rooms at T-3 to T-2 by typhoon type (per 100 total rooms; converted from observed values)

Scenario (corresponding observed event)	Return incidence (observed)	Median return volume (assumption)	Expected resale rooms / 100 rooms
Pessimistic: Honshu-crossing type (Ibaraki, Chiba, Shizuoka / Typhoon No. 15)	18.9%	10.5%	2.0 rooms
Mid-case: Nansei Islands, smaller event (Okinawa / Typhoon No. 9)	36.7%	10.5%	3.9 rooms
Optimistic: Nansei Islands, long-stalling type (Okinawa / Typhoon No. 13)	48.3%	10.5%	5.1 rooms

Source: Compiled by the HotelBank Editorial Team from MetroEngines Research / incidence rates are the observed T-3 to T-2 figures from Table 3. Median return volume uses the observed Okinawa / Typhoon No. 13 T-3 to T-2 value of 10.5% as a common assumption across all scenarios (median return volumes for the other events were not calculated for this article). Applying the same conversion to Okinawa normal days (incidence 15.6%, return volume 2.8%) yields just 0.4 rooms

The same calculation on normal days gives 0.4 rooms per 100. Whatever the typhoon type, the final band before a typhoon-approach stay date produces 4.5 to 11.6 times the resale volume of a normal day (the ratio of incidence × volume). That said, since the conversion is nothing more than the product of two observed values, it is worth knowing which of the two does the work.

Table 6: Two-way sensitivity, return incidence × return volume — expected resale rooms at T-3 to T-2 (per 100 total rooms)

Return incidence \ Median return volume	3%	5%	8%	10.5% (observed)	13%
20%	0.6	1.0	1.6	2.1	2.6
30%	0.9	1.5	2.4	3.1	3.9
40%	1.2	2.0	3.2	4.2	5.2
48.3% (observed)	1.4	2.4	3.9	5.1	6.3
55%	1.7	2.8	4.4	5.8	7.2

Source: Compiled by the HotelBank Editorial Team / the shaded cell is the observed Okinawa / Typhoon No. 13 T-3 to T-2 value (incidence 48.3% × return volume 10.5%). Each cell is incidence × return volume × 100 rooms

Read the table across and, holding incidence at 48.3%, moving return volume from 3% to 13% swings expected rooms from 1.4 to 6.3 — a factor of 4.5. Read it down and, holding return volume at 10.5%, moving incidence from 20% to 55% takes expected rooms from 2.1 to 5.8, a factor of only 2.8. What drives the scale of resale is less “at how many properties it happens” than “how many rooms come back at each one”. The observed data says the same: the typhoon-versus-normal gap was 3.1x on incidence but 3.8x on median return volume — the latter is the wider of the two.

In practical terms, that makes it more valuable to identify the properties — and the dates — with large return volumes first. Blocks of 10 or more rooms accounted for 14% of returns on typhoon-approach dates, seven times the 2% on normal days. Within a single property, the days with a high share of group and multi-night business are the days most likely to produce large returns. The first thing to check once a typhoon approach comes into view is your own property’s group and multi-night mix for that date.

What this means for revenue practice — three things to design in advance

Pre-define resale pricing for last-minute inventory by scenario

At T-3 to T-2 on a typhoon-approach date, roughly half of properties in Okinawa get back inventory on the order of 10% of total rooms. Working out the price from scratch in the moment costs time, and every hour of delay shrinks the opportunity. In the observed data, about 70% of returned inventory was relisted at an unchanged price — a clear sign of unused design headroom. At minimum, set pricing policy in advance for three cases: a stay date under an active storm warning, a post-typhoon stay date after the warning has been lifted, and a stay date where transport is disrupted but conditions around the property are normal. The post-typhoon date in particular tends to absorb rescheduled itineraries, so real demand returns — and a mechanical discount is not necessarily the right answer.

Expect the return peak at “the day the track firmed up, plus two to four days”

For an event like Typhoon No. 13, where the track was signalled early, returns built from T-7; for one like Typhoon No. 15, where the track stayed hard to read until the last moment, returns concentrated in T-3 to T-1. Rather than fixing a calendar lead time, it fits reality better to set the resale posture from the day forecast confidence improved. In practice that means preparing well ahead for Nansei Islands tracks and keeping the window open until the day before for tracks approaching Honshu.

Design cancellation policy and date-change offers around block returns

On typhoon-approach dates, blocks of 10 or more rooms return seven times as often as on normal days. That calls for different moves than the one-or-two-rooms-at-a-time pattern of an ordinary week. If you start from the assumption that a large number of rooms will open at once, it becomes well worth preparing date-change offers in advance. When a date change lands, the inventory on that stay date still returns — but the revenue is simply moved to another date. Typhoon-approach dates are usually flanked by post-typhoon demand-recovery dates, and having an alternative date of your own to offer is the decisive difference from an ordinary cancellation.

On the question of supply-side capacity, see also 43% of Chiba’s 75,065 Rooms Sit in 25 Disaster-Relief Municipalities, which addresses lodging demand during a disaster directly.

Method and caveats

Conclusion

Measuring remaining-inventory movement across 1,723 properties and 9,658 property-stay-dates for Typhoons No. 9, No. 13 and No. 15 in July–August 2026 confirms four things. First, on stay dates where a typhoon approached, the return incidence rate in the final two to three days (T-3 to T-2) reached 48.3% in Okinawa — 3.1 times the 15.6% on normal days. Second, the area-aggregate remaining-inventory rate reversed upward in Okinawa, from 18.5% at T-7 to 25.2% at T-3, meaning a substantial volume of rooms across the market returned to sellable status. Third, the lead time at which the return peak forms varies by typhoon: returns built from T-7 for Typhoon No. 13, whose track firmed up early, and concentrated in T-3 to T-1 for Typhoon No. 15, whose unusual track stayed unclear until the last moment.

And fourth, roughly 70% of returned inventory was relisted without any price change. A typhoon is an exogenous factor no one can avoid — but when, how much and at what granularity inventory comes back is well within the range of what can be anticipated. And if it can be anticipated, how that inventory is handled becomes a matter of design. Last-minute inventory in typhoon season is better understood not as a situation to be managed after the fact, but as a resale opportunity you can prepare for in advance.

Further reading

• Typhoon-Heavy Aug–Sep 2026: ‘Indoor-Complete’ Hotels Defend Occupancy
• Nagasaki Early Sep Booking Curves: 9/12 OCC 82.7%, Ryokan 24.2pt Gap
• Wakayama Silver Week Peaks Midway: Ryokan 93.7% Sun, 82.9% Sat
• Revenue Management vs Dynamic Pricing: 19,365 Hotels, Median 4 Moves
• 43% of Chiba’s 75,065 Rooms Sit in 25 Disaster-Relief Municipalities
• Chiba Floods: Narita 2.29x, Urayasu 2.40x, Disaster-Hit Chuo 0.83x
• Okinawa 176 Resorts: Obon vs Early Sep, 3.0pt Gap at 30 Days Out
• Booking Curves Across 47 Prefectures: +4.1pt Median Pickup from T-45