Sign inRegisterCheck connection

A Month of Daily NOAA-Style Rows as a Climate Appendix

A month of daily NOAA-style station rows as a climate appendix: completeness, trace rain, and quality flags a personal-station table often omits.

Back to Historical weather observations and forecasts

A monthly NOAA-style grid compresses each civil day into one line. A climate appendix expands those days: twenty-eight to thirty-one blocks with extra statistics, rain detail, and sometimes derived fire-weather indices. The appendix is where completeness, trace precipitation, and quality flags should live. On many personal-station pages they do not, which is why the appendix is easy to over-read as an official daily climate file.

The historical TNET Weather URL /tws-dailynoaa.php published that expanded month for a Mesa, Arizona station. A recovered capture is December 2014, day by day, plus month-to-date averages. Those values are a dated station diary. They are not current Mesa weather. Long recovered tables are not reproduced here.

Historical context

Weather Display can emit both a compact monthly climatological summary and a verbose daily report for each day of the month. The historical site styled the verbose form as “Month Daily Reports,” with a jump list of other available reports. Neighboring tools on the same template family included rain-detail charts and forecast pages. This URL was the appendix: more fields per day than the NOAA-like grid at /parse_noaa.php.

The recovered December 2014 dump listed, for each day, mean temperature, humidity, dew point, barometer, wind, rain for day/month/year, extrema with times, wind run, heat index, and Canadian Forest Fire Weather Index components (FFMC, DMC, BUI, ISI, DC, FWI). That mix is the teaching problem. Some fields are daily observed summaries. Some are accumulators that do not belong inside a single day block. Some are derived indices. Almost none carry a quality flag.

How to parse the compact HTML is /nb-0005.php. How to choose which month to open is /parse_noaa_date.php. This page is how to read the appendix as a completeness and quality document.

Completeness: a month is a set of valid days

A closed month has a known number of civil days. An appendix that lists “Day 1 … Day 6” in early December is a month-to-date file. Treat it as such. Do not compute a December mean from six days and call it the monthly climate.

For each day, completeness means:

  • The logger stored a full day (or the page says it did not).
  • Daily maximum and minimum are extrema over that full day, not over a morning fragment.
  • Daily rain is the catch for that civil day, not a running month total pasted into the day block.
  • Wind run and mean wind use the same valid interval.

Blank, zero, and missing. A dry day is 0.00 (or 0.0 mm) with a unit. A missing day is a token (M, MM, --) or an explicit “no report.” A blank cell in hobby HTML is ambiguous; do not coerce it to zero. Official daily archives such as GHCN-Daily use empty or missing sentinels and, in some products, strings of 9s; they do not ask the user to guess (GHCN-Daily documentation).

Month and year rain inside a day block. If Day 4 prints “rainfall for month 0.59 in” and “rainfall for year 11.11 in,” those are running totals as of that day, not properties of Day 4’s weather. An appendix that repeats running totals in every block invites double-counting. The daily climate value is “rainfall for day.”

Wind run. Daily wind run is a derived distance (mean speed × time) over the valid hours. A short day (logger down) yields a short run that is not comparable to a 24-hour neighbor without a completeness note.

Count valid days before you trust a month-end “average temperature” or “rainfall for month.” If the appendix’s month total disagrees with the sum of daily rains, believe neither until you know whether trace days, missing days, or a reset explain the gap.

Trace rain is not zero and not measurable

U.S. operational practice treats measurable precipitation as 0.01 inch or more. A trace is an amount too small to measure as a daily total, commonly described as less than 0.01 inch (AMS Glossary; NWS product language treats trace as not measurable). NCEI Local Climatological Data prints T for trace; some LCD notes describe trace as usually less than 0.005 inch water equivalent (LCD documentation).

A PWS appendix that only has a numeric rain column will typically print 0.00 both for a truly dry day and for a day when the bucket never tipped. That collapse loses information. If the logger or software can emit T, keep it. If it cannot, do not back-fill traces from a neighboring ASOS. You may note that the gauge resolution (often 0.01 in per tip) cannot represent a trace.

Never convert T to 0.00 to make a chart “numeric.” Never convert T to 0.01 to make a total “conservative.” Summing a month with traces requires a stated rule (count as zero for the arithmetic total, and report the count of trace days separately—the official habit).

A “maximum rain per minute 0.00 on day 01 at 00:00” line in a recovered dump is often a template default, not a meteorological event at midnight. Treat clock-zero extrema on a dry day as suspicious until the software’s empty-field behavior is known.

Quality flags the appendix usually omits

NCEI daily forms and MADIS mesonet feeds carry quality attributes: missing, trace, suspect, estimated, or failed QC. A Weather Display daily HTML page usually carries none. The absence is itself a quality statement: this is an operator diary, not a flagged archive.

When you reuse an appendix, add flags in a separate, labeled layer—do not overwrite the station’s number:

| Flag idea | Meaning | Do not hide it as | |---|---|---| | Missing | No valid daily statistic | 0.00 rain or 0° temperature | | Incomplete day | Outage inside the civil day | A “valid” max/min | | Trace | Precipitation below measurable resolution | 0.00 | | Derived | Heat index, FWI components, dew point | An extra sensor | | Accumulator | Month/year rain, month sunshine | A daily observation | | Suspect | Physically implausible or known siting fault | A silent correction toward the airport |

Canadian FWI components in a desert December file are a derived fire-weather scenario, not observed fuel moisture. If they appear, name the system and keep them out of the meteorological daily line. They are not NWS red-flag warnings. For fire and severe-weather safety, use National Weather Service products. TNET is not a warning service.

Siting still bounds every row. A complete, flag-clean backyard month is still that backyard. It does not become GHCN-Daily because the HTML said “NOAA.”

How to use the appendix next to the monthly grid

Read /parse_noaa.php for the compact month: daily high, low, rain, degree days. Open this appendix when you need the extra daily fields or when the compact rain column disagrees with a rain-detail chart (/tws-raindetail.php).

Reconciliation order:

  1. Header month, station, units.
  2. Valid-day list (which days exist, which are incomplete).
  3. Daily rain series with trace/missing preserved.
  4. Daily Tmax/Tmin series with incomplete days dropped from monthly extrema.
  5. Derived columns last, never mixed into observed extrema.

If you parse the compact file with the notebook workflow, store quality: missing-day count, parse errors on the month object. The appendix is where those counts should be obvious to a human.

Practical checklist

  1. Title the appendix with station ID, validity month, and complete/in-progress.
  2. Keep one true daily rain field; move month/year totals to a summary footer.
  3. Preserve trace as T or as a parallel flag; do not coerce to 0.00.
  4. Mark incomplete days; exclude them from monthly max/min unless you disclose the bias.
  5. Label heat index and FWI as derived.
  6. Do not invent GHCN-style flags you did not compute; it is better to say “unflagged station HTML.”
  7. Do not republish December 2014 Mesa daily blocks as current climate.

Modern relevance

Daily completeness, trace handling, and explicit quality flags are what make a month reusable. Public weather evidence used in later research has the same requirements: known period, known missingness, no silent zeros. TNET’s data sources, quality controls, and methodology page is the modern statement of that standard, without treating a PWS appendix as NCEI.

Observation hub: /legacy/observations/.

Sources