METAR Daily
Five questions on one real airport weather report, decoded as you answer
METAR Daily puts one real aviation weather report on screen and asks five questions about it. The report is not invented for the puzzle: it is an observation a working station actually filed, pulled from the NOAA feed and frozen into the game's calendar.
How a round goes
The raw report sits on screen from the moment the page loads. Nothing is hidden behind a reveal, because this is not a memory test: the whole game is reading a line of code groups that is right in front of you. Five multiple-choice questions come one at a time, and each is graded on the server, so which of the options is right, and the reasoning behind it, is not in the page until you commit an answer.
After every answer the relevant piece of the decode appears, and only that piece. If the ceiling is its own question, the flight-category explanation will not spell the ceiling out. The full token-by-token decode of the report arrives with the fifth answer, when there is nothing left to give away.
There is a second, shorter mode at /vfr. It shows the same day's report and asks a single question, VFR, MVFR, IFR or LIFR, then tells you which factor decided it: the ceiling, the visibility, or both when they agree.
The crosswind questions use the airport's real runways
Questions are generated from the parsed report together with the airport it belongs to. That matters most for the wind questions. The game carries a curated pool of airports with their runway headings in degrees true, so a prompt like "what is the crosswind component on runway 16" is a real decomposition of the reported wind against that runway's actual heading, not a plausible-looking number.
When the wind is calm or variable the crosswind question cannot be asked at all, so the generator substitutes from a reserve pool covering visibility, the temperature and dew-point spread, the altimeter setting and the wind sector. You always end up with five questions of five different types.
Difficulty on the daily puzzle follows the day of the week, Monday easiest through Sunday hardest, and archive replays rotate the same seven levels by puzzle number. It changes the distractors rather than the questions: on a hard day the wrong numeric options sit closer to the right one.
Where the reports come from
The data sources are NOAA / NWS Aviation Weather Center via aviationweather.gov, which is US federal government work and public domain, and OurAirports, public domain, for airport names, ICAO identifiers and runway headings.
Before a report was allowed into the game its derived answer key was cross-checked field by field against NOAA's own parsed fields, and a disagreeing record would have been rejected rather than patched. The one tolerated difference is NOAA's four-layer cap on cloud groups, which is documented in the parser's exceptions file. The game ships a pool of 365 reports, 45 of them dated into the live calendar and the rest held as an archive.
The decode is a teaching tool, not a briefing
The explanation after each answer, and the full decode at the end, are written by this site's own METAR parser. What you take away is the habit of reading the groups in order: report type, station, time, wind, visibility, present weather, cloud layers, temperature and dew point, pressure.
METAR Daily is a learning game. Its decoded reports are for practice only and must never be used for real flight planning or operational decisions. Always use an official aviation weather source.
Questions
Is it the same puzzle for everyone?
Yes. There is one puzzle a day and every player gets the same report and the same five questions. The day rolls over at midnight UTC, and the end screen counts down to the next one.
Do I need an account?
No. There is no sign-up and no server-side profile. Your streak and stats live in your browser's local storage, which means they are per browser and per device.
Do I need to already know how to read a METAR?
No. Every answer is followed by the part of the decode that explains it, so the report teaches itself group by group. Answering wrong costs you the question but still shows you the reasoning.