Weather Data Market
BETA

About Weather Data Market

A consensus weather oracle and autonomous data market - built for humans and AI agents alike.

Our purpose

VERIFIABLE PUBLIC DATA ASSET

Weather moves markets, insurance, logistics and energy - yet most weather feeds ask you to trust a single provider. Weather Data Market exists to make weather a verifiable public data asset: national met-agency sources are combined into one consensus index, the same way exchanges combine order books into a reference price, so no single source - or its failure - can distort the truth.

Every hourly run collects all sources per city, rejects outliers by median + MAD, computes the WDI/TDI indexes with a fixed rule, issues a cross-check verdict against the METAR reference, and anchors the snapshot hash on-chain. What you buy is exactly what anyone can re-verify.

How it works

COLLECT → CONSENSUS → ANCHOR

Collect

Every hour, 11 source feeds - METAR airport observations plus 7 national weather-agency models, seven of them via one aggregator - report every city.

Consensus

Median + MAD outlier rejection produces one manipulation-resistant index; a verdict oracle flags any source that drifts.

Anchor & sell

Each run's snapshot hash is anchored on Flare. Data, oracle feeds and forecast/verdict contracts settle in XRP - your wallet is the account.

Key features

MANIPULATION-RESISTANT · VERIFIABLE · AUTONOMOUS
CONSENSUS
Multi-source consensus

11 source feeds - METAR airport observations, KWeather and MET Norway direct, plus seven national weather-agency models retrieved through one aggregator - merged with median + MAD outlier rejection. Anomalies are excluded, never averaged in.

PROOF
On-chain proof

Every index snapshot is SHA-256 hashed, and the hash is anchored on-chain - retried each cycle until it lands, with current coverage shown on the snapshots page. Re-download, re-hash, verify - trustlessly.

OPEN
Fixed, anchored rule

WDI and TDI are deterministic: same inputs, same number, every time. The rule does not change between runs, and every run's hash is anchored on-chain. Its parameters stay proprietary - what you get is the verdict and its reasons.

AUTONOMOUS
Humans + AI, equal access

People buy on the web with a wallet; AI agents buy over REST with the same quote → pay → claim flow. Members keep history on the server.

What we sell

30-DAY 25 USD · YEARLY 200 · AGENT 0.04 / CALL · PAID IN XRP

Every product ships with the verdict oracle, on-chain anchored proof and the rain (Y/N) criterion by default - a Web3 weather oracle for parametric climate insurance, weather-linked hedging and prediction-market settlement.

Accounts & wallets

WALLET PAYMENT · OPTIONAL MEMBERSHIP · AGENT-READY

Payment always happens from your crypto wallet. Create a member account to keep purchase history on the server and manage subscriptions across devices - or stay wallet-only. AI agents trade fully autonomously with no account at all.

An account is optional

You can buy everything here without one. Connect a wallet, pay, download - no email, no sign-up. Buying without an account still records the wallet address and the order (transaction hash, cities, period, amount, dataset hash): that is how the data is delivered to you and re-delivered later, and it is a transaction record we are required to keep. An account adds an email address, and keeps that history reachable from another device.

Never collected: name, phone number, postal address, date of birth, national ID, or card details. No identity verification. No advertising, profiling, tracking cookies or third-party trackers, and your data is never sold or shared. If you do create an account, the email address is the only item added beyond the wallet and order records above - see the Privacy Policy.

Vision, together

DATA × CHAIN
DATA · KWEATHER
Weather as everyday intelligence

KWeather, Korea's leading private weather company, opens meteorological data beyond the forecast screen - re-providing Korea Meteorological Administration (KMA) observations through its gateway and operating its own nationwide fine-dust IoT sensor network. Its direction is clear: turn weather from a daily headline into operational intelligence that businesses, cities and individuals can act on.

KWeather does not define a city by one station. Its city indices combine the city's official observation stations, ICAO METAR, and GTS bulletins with numerical models from around the world (ECMWF, GFS, JMA, KIM), AI techniques, and terrain information such as land cover. As observation networks expand and measurement density rises, the indices become more accurate. In this market KWeather is a first-class observation source - KMA-derived ground truth for Korean cities, world realtime coverage abroad - and one of the observation class the consensus is checked against.

CHAIN · FLARE NETWORK
The blockchain for data

Flare's mission is to give decentralized applications secure access to high-integrity real-world data. Its enshrined protocols - FTSOv2 price feeds and Flare Confidential Compute (FCC) with TEE-attested execution - make external data provable on-chain instead of merely claimed.

Here, Flare is the intended proof layer: FLR settlement priced by FTSOv2, and the consensus computation inside an FCC enclave whose signature the consumer contract verifies on chain. That path is built and tested, but it runs on the Coston2 testnet with a test key - FCC is not on Flare mainnet yet. See the Technical Proof page for what is verifiable today and what is not.

Official oracle of KWeather Co., Ltd.

Terms →
PROVIDER · CONTRACTING PARTY
KWeather Co., Ltd.

Weather Data Market is the official Web3 weather oracle of KWeather Co., Ltd. (Business Registration No. 110-81-37628, Seoul, Republic of Korea), the provider of the data and the contracting party. The platform is developed and operated under delegation by Wellbian Labs Pte. Ltd. (UEN 202604584N, Singapore), a separate company that is not a subsidiary of KWeather and not a party to your contract. The two are related parties under common control - see section 1 of the terms.

Business Reg.
110-81-37628
Corporate Reg.
110111-1427586
Representative
Kim Dong-sik
Since
1 July 1997
Office
4F Room 401, 5 Digital-ro 26-gil, Guro-gu, Seoul, Republic of Korea
Contact
khelp@kweather.co.kr · +82 1644-0902

Contract, licence, disputes and personal-data requests go here. Technical and API questions go to the operator - see the next card.

DEVELOPER · OPERATOR (DELEGATED)
WELLBIAN LABS PTE. LTD.

Builds and runs this platform under delegation from KWeather. A separate legal entity, not a subsidiary - KWeather holds no shares in it - and not a party to your purchase contract.

UEN
202604584N
Type
Private company limited by shares (Singapore)
Incorporated
29 January 2026
Activity
Software and application development (SSIC 62011)
Office
111 Somerset Road #06-01W, Singapore 238164
Contact
admin@wellbianlabs.io (technical / API only)

Related-party disclosure: the two companies are under common control. Kim Dong-sik is Representative Director of KWeather and majority shareholder of Wellbian Labs. Both registries are public; a service built so you need not trust its operator should not leave that unstated.

INTENDED USE
Live API subscription

Intended for parametric climate insurance, weather-linked hedging, prediction-market settlement, planning and research. Operating an activity that requires a licence under Korean gambling law without holding one is prohibited by the terms.

Section 8 of the terms makes this a condition of the licence, not a recommendation: data supplied to a service of that kind may be cut off without refund. Section 10 sets out what is warranted - the process, not the weather.

Frequently asked questions

Where does the number come from?

Eleven weather source feeds are collected for each city, then reduced to a single value by taking the median and rejecting outliers. Which sources are rejected is decided by a fixed rule whose parameters are not published. No single provider can move the published figure on its own. Eleven feeds is not eleven observations of the city: two are measurements - METAR from airport stations, and KWeather, carrying KMA observations in Korea and GTS SYNOP bulletins elsewhere - and the remaining nine are global numerical models. Only the two measurement feeds are counted when a value is marked verified.

Which sources are used, and what does each actually cover?

Two measurement feeds and nine models. The measurements: METAR airport observations from NOAA/AWC, taken at ICAO stations mapped to every city in the catalogue; and KWeather, which supplies Korea Meteorological Administration surface observations for Korean cities and GTS SYNOP bulletins - measurements from WMO synoptic stations, distributed over the Global Telecommunication System - everywhere else. Those are two different station networks, so they corroborate each other independently. Note that KWeather is not observing the non-Korean cities itself and neither is the KMA, which observes South Korea only; it is re-providing the SYNOP measurements those countries publish, and SYNOP normally reports three-hourly while METAR reports hourly or more often. Then nine global numerical models: ECMWF, DWD ICON, JMA, Meteo-France, the UK Met Office, NOAA GFS, ECCC GEM, MET Norway and Open-Meteo's own blend. MET Norway is the national authority for Norway and the Nordics; outside that its API returns model output like the rest. Each source carries its own licence and attribution, listed on the technical proof page. Four feeds are fetched directly (KWeather, METAR, MET Norway, DWD MOSMIX); six of the national-agency models and Open-Meteo's own blend arrive through Open-Meteo, so an outage there removes seven of the eleven at once. DWD is the exception: MOSMIX is fetched straight from DWD open data.

How can I verify a value after I have bought it?

Every collection run is serialised to a canonical JSON snapshot, hashed with SHA-256 and anchored on Flare mainnet in the SnapshotAnchor contract (the anchor of record since 23 September 2026), and, for runs before 23 September 2026, as a first-generation copy on BNB Chain. Each row you receive carries its run ID, the snapshot hash and the anchor transaction, so you can recompute the hash and compare it with the chain without asking us - or simply call SnapshotAnchor.isAnchored(hash).

What does it cost and how do I pay?

Pricing is per city per term, quoted in USD and settled in XRP on the XRP Ledger: the quote converts the price at the FTSOv2 XRP/USD rate and fixes the XRP amount for ten minutes. There is no account, subscription tier or card. You send the payment with the quote's DestinationTag, paste the transaction hash, and the data is released against that transaction.

Can an AI agent buy data without a human?

Yes - that is the primary design target. An agent calls the oracle endpoint, receives HTTP 402 with the payment requirements, approves the meter contract once, and every subsequent call settles on-chain automatically. The full flow is documented for agents to read directly.

What am I allowed to do with the data?

Use it in your own products, services and analyses, commercially included, and show it to your own end users. Attribution is required wherever it is displayed - "Powered by Weather Data Market", linked; every API response carries an attribution object with the exact text and markup. What you may not do is hand it on to third parties as a feed, API or dataset, or resell it substantially as delivered under another brand. Anything you derive from it - analyses, indexes, forecasts - is yours. Data already received stays usable after a subscription ends. Section 7 of the terms is the licence.

Do you sell historical weather data?

No. Contracts run forward from the purchase date, up to 365 days, and days fill in as they are collected and anchored. This is deliberate: the value here is a verifiable record produced under consensus, which cannot be manufactured retroactively for dates that were never collected this way.