Scan: Will Seoul have less than 75mm of precipitation in September?

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: polymarket_direct polymarket_related kalshi_related claude_news article_search gdelt_news code_execution wikipedia

Sub-questions (Fermi decomposition)

What is the current Polymarket YES price for <75mm and what do the other Seoul September precipitation brackets imply about the crowd's distribution?
Over the past 30-40 years of KMA Seoul records, in what fraction of Septembers did total precipitation fall below 75mm, and what is the median/spread?
How much precipitation has KMA's Seoul station recorded so far in September 2026 (month-to-date), and how many days remain?
What does the KMA 1-month/3-month outlook and ECMWF/other ensemble guidance say about September 2026 precipitation for the Seoul metropolitan area (below/near/above normal)?
Are any typhoons or major frontal systems forecast to affect the Korean peninsula in the remainder of September 2026, and how has the summer monsoon (changma) retreat behaved this year?
What is the current ENSO / Pacific SST state in 2026 and does it historically correlate with drier or wetter Septembers in Seoul?

Planner reasoning

This is a climatological threshold question: Seoul's September normal rainfall is roughly 150-170mm, so <75mm is a distinctly dry outcome that historically occurs in a minority of years. The key drivers are (1) the historical base rate from KMA Seoul September totals, (2) how much rain has already fallen month-to-date if September 2026 is underway, (3) KMA/ensemble outlooks and typhoon/monsoon-retreat patterns, and (4) the current market price and the other precipitation brackets on Polymarket, which reveal the crowd's implied distribution.

2. Research Parallel tool calls

tool	status	time	items	summary
polymarket_direct	OK	0.2s	1	## This Market's Polymarket Data Will Seoul have less than 75mm of precipitation in September? - Current price (probability): 62.00% - 7-day price change: +36.00% - 30-day price change: +37.00% - Total volume: $15,363 (USD notional) - Price range: 12.50% - 62.00% - Data points: 9 days
polymarket_related	OK	3.7s	0	Scanned 100 active Polymarket markets, kept 0 matches. keyword 'Seoul precipitation': 0 markets \| keyword 'Seoul September': 0 markets \| keyword 'Seoul rain': 0 markets
kalshi_related	OK	3.4s	2	2 related markets / summaries. keyword 'Seoul rain': ok \| keyword 'Seoul precipitation': no matches \| keyword 'Korea weather': ok
claude_news	OK	27.9s	6	Based on available search results, here are the key findings: - Typhoon Krovanh (2026) tracking near Japan, not directly threatening Seoul: As of Sept 7, 2026, Krovanh was located 482 km east of Kadena Air Base, tracking north-northeastward, forecast to turn northeastward as it rounds the stee
article_search	OK	0.7s	33	Fetched 33 unique articles across 4 queries (lookback=45d). 'Seoul September 2026 rainfall KMA': 10 articles \| 'Korea typhoon September 2026': 10 articles \| 'KMA monthly outlook September precipitation Seoul': 10 articles \| 'Seoul dry September 2026 drought': 10 articles
gdelt_news	OK	217.4s	0	GDELT: 0 articles across 3 queries (lookback=30d). 'Seoul rainfall September': error GDELT rate-limited after retries (429) \| 'Korea Meteorological Administration September precipitation': error GDELT rate-limited after retries (429) \| 'typhoon Korea September 2026': error GDELT rate-limited after r
code_execution	OK	45.7s	0	## Quantitative Findings: Seoul September Precipitation < 75mm Empirical base rate (2013–2024, n=12 approximate KMA values): - Sample mean = 120.0 mm, std = 46.7 mm (CV ≈ 0.39) - Empirical fraction with total < 75mm: 3/12 = 25.0% (years 2015 (70mm), 2016 (50mm), 2017 (50mm)) **Distribution
wikipedia	OK	0.2s	3	Fetched 3 Wikipedia entries (0 missing pages).

3. Evidence Brief Sonnet 5 · 6306 chars

# Current state September 2026 is in progress with the market resolving on final monthly KMA precipitation total for Seoul; as of the last confirmed news check (Sept 5-7, 2026), Seoul had recorded ~0mm with only a 10% rain chance and a dry, sunny short-term forecast, while Typhoon Krovanh remained well offshore (near Japan, not Korea). No official KMA month-to-date figure was directly retrieved in research. # Timeline of key events - 2026-08-04: South Korea sets national heat record (41.4°C, Yangsan) amid intense heatwave — confirmed (KMA/CBC). - 2026-08-14: NPR reports on Asia/Europe weather extremes, mentions Seoul context generally, no specific rainfall data — reported. - 2026-09-01 to 09-05: Seoul recorded ~0mm precipitation, 10% rain chance, mostly sunny/overcast forecast through Sept 10 — reported (aqi.in). - 2026-09-07: Typhoon Krovanh (No. 24) tracked ~482km east of Kadena, forecast to curve toward southern Japan (Shikoku/Kansai), staying ~300-480km from Korean coast; KMA monitoring but no direct landfall threat to Seoul — reported (zoom.earth, stormatlasx, KMA typhoon page). # Event Will Seoul's total September 2026 precipitation (per KMA) be less than 75mm? # Outcomes to forecast Yes (<75mm) / No (≥75mm) # Kalshi market anchor No direct Kalshi YES price was returned in this research pass (kalshi_direct tool absent/silent). Only a synced Polymarket market for the identical question is available: **YES = 62%**, up sharply from a 30-day low of 12.5% and a 7-day rise of +36 points, on modest volume ($15.4k total, 9 data points). This is our best available consensus proxy; treat Kalshi's actual price as unconfirmed but likely broadly aligned or slightly divergent given thin liquidity. # Sub-question answers 1. **Polymarket YES price / distribution** — 62% for <75mm bucket; no other Seoul precipitation bracket markets found on Polymarket or Kalshi (polymarket_related, kalshi_related: no matches), so no cross-bracket distribution is directly observable. 2. **Historical base rate (past ~12 KMA years, 2013-2024)** — 3/12 years (25%) fell below 75mm (2015: 70mm, 2016/2017: 50mm each); sample mean ≈120mm, std ≈47mm. Distribution-fit models (gamma/lognormal) anchored to this recent sample give P(<75mm) ≈16-19%. If anchored to the older 1991-2020 normal (~170mm), P(<75mm) drops to ≈4%. Blended central estimate ≈10-20% (code_execution). 3. **Month-to-date 2026 precipitation** — No official KMA MTD figure retrieved; secondary source (aqi.in) reports ~0mm through Sept 5, 2026 with low rain chance into Sept 10 — suggestive of a dry start but not authoritative. 4. **KMA/ECMWF outlook** — Not retrieved; KMA English 1-month outlook and ENSO pages did not yield usable September 2026 probability data (claude_news notes this gap explicitly). 5. **Typhoon/changma activity** — Typhoon Krovanh (Sept 7, 2026) tracking toward Japan, staying 300-480km offshore from Korea; no direct landfall threat identified. No other typhoon or major frontal system reported approaching the peninsula in remaining September window. Changma (monsoon) retreat behavior not specifically detailed in sources. 6. **ENSO state** — Not confirmed in research; one Chicago-focused article mentions a "developing El Niño" pattern in NH summer 2026, but no confirmed ENSO/SST state for the Korea region was found, and no direct correlation data to Seoul September rainfall was retrieved. # Key facts (high-confidence, factual) 1. [code_execution] Recent 12-yr KMA sample: mean ≈120mm, 25% of years <75mm. 2. [Wikipedia] Most of Seoul's annual precipitation falls June-September via monsoon/typhoon activity — high variance driver. 3. [claude_news] 1991-2021 climatology: September rainfall front-loaded (first 10 days avg 6.3mm/day-equivalent), days 11-20 driest (~1.0mm/day), suggesting typical monthly totals well below 75mm absent typhoon events. 4. [claude_news/aqi.in] Seoul recorded ~0mm through Sept 5, 2026, low rain probability into Sept 10. 5. [zoom.earth/stormatlasx] Typhoon Krovanh (Sept 7, 2026) tracking away from Korea toward Japan. 6. [CBC] Record national heatwave (41.4°C) in South Korea Aug 4, 2026, indicating an unusually hot, possibly dry summer pattern preceding September. # Cross-market signals - Kalshi related: No matching Seoul precipitation market found; unrelated Korea politics market (Lee Jae-myung full term, 51%) offers no weather signal. - Polymarket: Same-question market at 62% YES, sharp +37pt 30-day rally suggests recent dry conditions/reports have shifted sentiment strongly toward "under 75mm," but thin volume ($15k) limits confidence. - Sportsbook implied: None available (not a sports market). # Analyst opinions and speculation - No direct meteorologist commentary found on September 2026 Seoul outlook; code_execution model treats this as base-rate/statistical extrapolation rather than expert forecast. - Heatwave-heavy summer 2026 narrative (record temps) may correlate with drier atmospheric pattern, but this is inferential, not sourced directly to precipitation forecasts. # Directional lean per outcome - **Yes (<75mm)**: Supported by early-month dryness (~0mm through Sept 5), Krovanh remaining offshore, historical climatology showing typical totals often under 75mm, and strong Polymarket momentum (62%). - **No (≥75mm)**: Supported by high year-to-year variance (std ≈47mm on 120mm mean), remaining 3+ weeks of September (including typhoon season) unaccounted for, and lack of confirmed dry outlook from KMA official sources; single typhoon landfall could easily push totals over threshold. # Gaps / unknowns - No confirmed Kalshi YES price for this specific ticker. - No official KMA month-to-date total or 1-month outlook data retrieved. - ENSO state for 2026 unconfirmed. - Uncertain how many days into September the "current" evidence reflects (research dated ~Sept 5-7); ~23+ days of September risk remain unaccounted for. # Calibration anchors - Polymarket proxy YES price: 62% (recent, low-liquidity, rapidly rising). - Historical base rate: ~25% raw (3/12 yrs), model-implied 10-20% central estimate. - Early dry start (0mm through day ~5) meaningfully raises conditional probability of finishing <75mm per code_execution scenario table (e.g., X=0mm, D=30 → 18%; increasing sharply if dryness persists through mid-month).

4. Forecasters Fable 5.1 + GPT-5.6 Sol · in parallel

claude-fable-5-1

40% conf

Yes	55%
No	45%

No direct Kalshi price was available; the Polymarket proxy for the identical question sits at 62% YES after a sharp rally, consistent with a dry start (~0mm through Sept 5, low rain chances into Sept 10) and Typhoon Krovanh tracking toward Japan rather than Korea. However, the historical base rate is only ~25% raw (3/12 recent years below 75mm) with high variance (std ≈47mm), and roughly 23 days of typhoon/frontal-season September remain unaccounted for, so a single landfall or frontal event could easily push totals past 75mm. Conditioning on a dry first ~10 days and climatology showing days 11-20 as the driest stretch, the probability of finishing under 75mm plausibly rises to ~45-60%. I shade slightly below the thin-liquidity Polymarket price given the lack of official KMA month-to-date or outlook data.

gpt-5.6-sol

54% conf

Yes	43%
No	57%

The 62% Polymarket YES price is the best consensus anchor, and the reported near-zero rainfall through September 5 plus Krovanh tracking toward Japan support an unusually dry start. However, the strongest historical evidence points lower: only 3 of 12 recent years finished below 75mm, while fitted models estimate a 10-20% unconditional probability. The dry start raises that base rate, especially because September rainfall is normally front-loaded, but more than three weeks of volatile frontal and typhoon-season weather remain and no official KMA outlook confirms persistent dryness. I therefore place YES below the thin, sharply moving market price while still substantially above the historical prior.

5. Devil's Advocate Sonnet 5 · red-team critique

Both forecasters land far below the Polymarket anchor (62%) despite explicitly calling it the "best available consensus proxy," but neither justifies why they discount it by 7-19 points beyond vague "thin liquidity" hand-waving—thin volume ($15k, 9 points) cuts both ways and could just as easily mean the price is stale/uninformative rather than systematically too high.

Both treat the ~0mm-through-Sept-5 dry start as strongly diagnostic, but the brief's own climatology fact (#3: "first 10 days avg 6.3mm/day-equivalent, days 11-20 driest ~1.0mm/day") implies the first 10 days are normally the WETTEST period—so a dry first 5-10 days is arguably more informative of a below-normal month than either forecaster's reasoning fully credits, yet both land near 45-57% No, oddly close to a coin flip given this front-loaded-then-dry pattern.

Neither forecast explicitly reconciles the tension between the low unconditional base rate (10-20% model-implied, 25% raw) and their much higher conditional estimates (43-55%) with an actual updating mechanism (e.g., Bayesian combination of prior with dry-start likelihood); the jump is asserted narratively rather than computed, leaving the magnitude of the update unjustified.

Both ignore the explicit gap flagged in the brief that "23+ days of September risk remain unaccounted for" in a way that meaningfully changes tail risk: a single typhoon landfall (unlike Krovanh) could push monthly totals well past 75mm in one event, given monsoon/typhoon activity drives "most of Seoul's annual precipitation" — this asymmetric tail risk for the No outcome is mentioned but not quantitatively incorporated into either probability.

Neither forecaster addresses the ENSO uncertainty (a "developing El Niño" is only loosely mentioned) despite it being a known driver of East Asian monsoon/typhoon variability; this is simply dropped as a non-factor rather than treated as a genuine unknown that should widen uncertainty bands.

The two forecasts split on the same evidence to land on opposite sides of 50% (0.55 Yes vs 0.43 Yes), essentially symmetric around a coin flip, suggesting neither has strong conviction—yet their stated confidence levels (0.40 and 0.54) don't clearly reflect this near-disagreement or the wide plausible range (10-62%) spanned by the cited anchors.

6. Aggregator Fable 5.1 · final calibrated forecast

58% (47% confidence)

Yes: 58% No: 42%

The Polymarket proxy at 62% YES is the only market anchor, and the critique correctly notes both forecasters discounted it without a clear mechanism. The brief's own climatology shows Seoul's September rain is heavily front-loaded (first 10 days normally the wettest stretch, days 11-20 the driest), so ~0mm through Sept 5 with a dry outlook to Sept 10 removes the largest expected chunk of the month's rainfall and makes finishing under 75mm more likely than the unconditional 10-25% base rate suggests. Against that, roughly three weeks remain in typhoon/frontal season and a single landfalling system could clear 75mm in one event, which keeps a substantial fat tail for No. Balancing the dry-start update against the unhedged tail risk and the lack of official KMA data, I land near but slightly below the market at 58% YES.

Pipeline Timing

Total pipeline time: 337.6s

Per-tool research timings shown in the Research section above.