AO1, AO2, AUTO, and the $ Maintenance Flag

Most METARs you read are generated by machines, and a few small tokens tell you exactly how much that machine can sense and whether it is healthy. AO2 marks an automated station with a precipitation discriminator, while a trailing dollar sign flags that the station needs maintenance. Between those two signals sits the AUTO modifier near the front of the report and its quieter sibling AO1, and together they let you judge how far to trust an unattended observation before you ever look at the weather it reports.

This guide covers the AUTO and COR modifiers, the difference between AO1 and AO2, why an AO1 weather group deserves a second look, and the maintenance $. It is part of the METAR field guide; for where these station-type codes live in the report, start with inside METAR remarks.

AUTO and COR: how the report was produced

Right after the date-time group, a METAR can carry one of two report modifiers before the wind:

• AUTO means the observation was produced entirely by automated equipment with no human augmentation. Nothing in it was added, corrected, or confirmed by an observer.
• COR means the report is a correction that replaces an earlier observation for the same time. It can apply to a manual or an automated report.

KXYZ 291953Z AUTO 27012KT 10SM CLR 24/09 A2998 RMK AO2

The AUTO here tells you a sensor package, not a person, decided that the sky was clear and the visibility was 10 statute miles. That matters because a human observer can see things a sensor cannot: a distant thunderstorm, a thin cloud layer outside the sensor's narrow look angle, or blowing dust on the horizon. AUTO is not a warning, but it does tell you the report has limits, and the station-type remark tells you what those limits are.

AO1 versus AO2: the precipitation discriminator

Down in the remarks, a US automated station identifies itself as either AO1 or AO2. The single difference is a piece of hardware called a precipitation discriminator.

• AO1 is an automated station without a precipitation discriminator. It can detect that precipitation is falling, but it cannot reliably tell what kind. It does not distinguish liquid from frozen.
• AO2 is an automated station with a precipitation discriminator. It can distinguish rain from snow, so its precipitation reports carry far more information.

That hardware difference is why the two stations describe the same weather differently. An AO2 site can populate a present-weather group with a precise phenomenon code such as RA for rain or SN for snow, and it can report freezing precipitation. An AO1 site, lacking the discriminator, often reports only that something is falling, with less confidence about its type.

Why AO1 weather groups are less trustworthy

The discriminator gap has a direct operational consequence: the present-weather group from an AO1 station is the weakest part of its report. Because the station cannot classify precipitation type, a winter scenario where the difference between rain and freezing rain is the difference between a wet runway and an icing event is exactly where an AO1 site has the least to offer.

This is not a reason to ignore an AO1 report. The wind, pressure, and temperature sensors are unaffected, and the temperature and dewpoint still tell you whether conditions favor frozen precipitation. It is a reason to cross-check. If an AO1 station near your route shows precipitation and the temperature is close to freezing, treat the precipitation type as unknown rather than assuming rain, and corroborate with a nearby AO2 station or a forecast before you lean on it.

The trailing $ maintenance flag

The last signal is the smallest. A $ at the very end of the remarks means the automated station has run its self-check and flagged that it needs maintenance.

KXYZ 292353Z AUTO 18004KT 10SM CLR 19/12 A3001 RMK AO2 SLP163 T01890122 $

The $ does not invalidate the observation. The report is still transmitted and still usable. What it tells you is that one or more sensors or systems have detected a fault, so the data may be degraded in ways the station itself has noticed. When you see the maintenance flag, give the rest of the report a more skeptical read, especially any group that looks inconsistent with neighboring stations.

Putting it together when you trust an automated report

Read these four tokens as a confidence chain. AUTO tells you no human reviewed the observation. AO1 or AO2 tells you whether the station can classify precipitation. The $, if present, tells you the station has flagged itself as unhealthy. None of them changes a single value in the report, but together they tell you how much weight each value can bear.

A clean AO2 report with no $ is about as solid as automated data gets. An AO1 report with the maintenance flag, in winter, near freezing, is one to corroborate before you act on it. When you want every one of these tokens pulled out and labeled in plain English next to the raw text, paste the report into the decoder, and read inside METAR remarks for the rest of the remarks section. Whether the report arrives as a routine observation or an unscheduled one is a separate question, covered in why you sometimes see SPECI. wxdecoder is for situational awareness, not an official briefing, so use these flags to decide how hard to cross-check, not as the final word.

Last updated June 29, 2026