Understanding Temperature Data

Official ASOS (Automated Surface Observing System) temperatures are 2-minute averages of sensor readings, reported in whole-degree Fahrenheit. This is the standard used for official weather records and market settlement.

However, the publicly available 5-minute data that most weather apps and websites display goes through a multi-step conversion that introduces rounding errors:

1. The sensor reads a precise Fahrenheit value (e.g., 77.6°F)
2. It gets converted to Celsius (25.33°C)
3. Rounded to the nearest whole Celsius (25°C)
4. Converted back to Fahrenheit (77.0°F)
5. Rounded to whole Fahrenheit (77°F)

The original 77.6°F reading could end up displayed as 77°F or 78°F depending on exactly where the rounding falls. That 1-2°F difference can determine whether a market contract pays out.

Most weather websites and apps source their data from ASOS stations, but the data they receive often lacks the proper 2-minute averaging required for official records. Some sources pull from whole-degree Celsius values and convert back to Fahrenheit, compounding the rounding error.

No known near-real-time public source combines correct 2-minute averaging with proper Fahrenheit conversion. This means every real-time temperature you see online is an approximation — not the value that will appear in the official record.

The Iowa Environmental Mesonet (IEM) deliberately marks some 5-minute near-real-time temperatures as missing specifically because they don't meet the official 2-minute averaging standard. They'd rather show no data than misleading data.

Not all observation intervals are equal in precision:

• •5-minute intervals — Most frequent but least precise. Subject to the full rounding pipeline described above. Good for tracking trends, unreliable for exact values.
• •Hourly observations (XX:51 to XX:54) — Official hourly readings with less rounding ambiguity. Celsius values are reported to one decimal place, reducing conversion error.
• •SPECI reports — Special observations triggered by significant weather changes. These contain exact temperature readings without the standard 5-minute rounding ambiguity.

You might expect the "top of the hour" temperature to be measured at exactly :00. In reality, hourly reports arrive several minutes early due to collection procedures dating back to the 1970s.

ASOS sites have reset times between :50 and :59 after each hour. If the temperature changes by 1°F between the reset time and the actual top of the hour, that change may go unreported. The METAR T-group temperatures can also differ from mandatory integer Celsius values, creating additional discrepancies.

For traders watching a temperature approach a strike price in the final minutes before the hour, this timing gap matters.

minuteTemp does not treat every temperature-shaped data point the same way. The main observation line is built from non-report sensor observations first, so it preserves a consistent real-time station trend instead of mixing in every later bulletin or promoted correction.

That means a METAR T-group, CLI, or DSM-derived temperature can still matter for official interpretation without becoming part of the primary observation line. If a report-derived point lands at a time when the canonical sensor observation is missing, the chart may visually bridge across that minute using the surrounding non-report observations.

• •Observation line: optimized to show the live sensor timeline traders are watching minute to minute.
• •Report-derived values: can still influence official context such as settlement interpretation or promoted highs and lows, even when they are not drawn as a normal line point.
• •Fallback behavior: if a report-derived value is all we have, it can still be surfaced as the latest available reading rather than dropped entirely.

In practice, this means a chart line, a latest displayed value, and the eventual official record can briefly disagree without any one of them being "wrong." They are answering slightly different questions about the same station.

minuteTemp treats each observed temperature as a range rather than a single exact value. Because many public readings originate from integer Celsius, each reading implies a Fahrenheit window that is roughly 1.8°F wide.

• •Chart overlay: red line = estimated max, blue line = estimated min
• •Tooltip labels: hover shows Actual, Min, and Max together
• •External payloads: REST and WebSocket observations includetemp_min_fandtemp_max_f

This makes uncertainty explicit in both the UI and the API, so internal and external consumers can reason about the same bounds.