A weekly climate table is a small climatology: seven civil days, each with temperature, rain, and wind statistics, plus a line that says which days are missing. It is not a monthly NOAA-style summary, and it is not the latest current-conditions packet. The scientific problems are compactness and honesty. Seven rows are easy to misread as a complete week, and a backyard mean is easy to misread as a departure from an official normal.
The historical TNET Weather file /weather/weekrep.htm was titled Daily Weather Report. Weather Display used that name for a short-period climate extract. A recovered capture covered April 2005 through the 25th as month-to-date averages, extrema, daily rain, and all-time logger records. Those figures are a dated station product. They are not current Mesa weather and are not copied here as a climate of the city.
Historical context
Station software of that era emitted several HTML climate products from the same archive: a monthly form that imitated a NOAA climatological summary, a verbose day-by-day appendix, and a shorter “week” or “daily weather report” aimed at a glance. This URL was the short extract.
The recovered April 2005 file mixed three timescales in one text page: month-to-date means, daily rain on the days that had any, sunshine and evapotranspiration listed day by day, and all-time records with dates in 2004–2005. That mix is why a modern reader still needs a rule: name the period of every block. A week table that silently contains month-to-date means is not a week table.
This article treats the URL as the place to learn a seven-row (or daily-in-a-week) PWS climate table. It does not restore the 2005 numbers.
What seven rows must contain
Each row is one civil day in the station’s local calendar.
Day. Calendar date, not a row index. If the table is “week of 18–24 April,” the dates must say so. A heading that only says “Daily Weather Report” is not a period.
Temperature statistics. Daily maximum, minimum, and a mean if the logger publishes one. State units (°F or °C). Know whether the mean is the average of max and min or a true interval mean. They differ on dry, high-range desert days.
Precipitation. Depth for that civil day, in inches or millimetres. Trace, zero, and missing are three different states (see below). A rain event that straddles midnight splits across rows.
Wind. Daily mean and daily high gust, with unit and, if present, direction of the high or a dominant direction. Dominant direction is a summary of the day’s distribution, not automatically the direction at the time of the gust.
Completeness flag. Whether the day has a full archive. A day with a logger outage from noon to midnight is not a valid daily maximum in the climate sense even if a number exists for the morning.
A weekly product may add a seventh-day summary line: week mean temperature, week rain total, highest max, lowest min. Those summaries must skip missing days or they invent climate.
Missing days are not zeros
The most common weekly-table error is filling a blank with 0.0.
- Missing temperature is not 0°F.
- Missing rain is not 0.00 in. Zero rain means the gauge reported a dry day. Missing means the archive has no valid day total.
- Missing wind is not calm. Calm is an observed zero with a unit.
Official climate products use explicit missing tokens (NWS climate instructions use MM for missing in CLI/CLM products). Personal-station HTML often uses a blank, a dash, or a leftover zero from a template. Before you chart a week, count valid days. A “weekly mean” from five valid days and two zeros is not a weekly mean.
If you publish a week before Sunday (or before the station’s chosen week boundary), trailing days should be absent, not forecast and not zero-filled. Incomplete weeks are allowed. Silent padding is not.
The recovered April 2005 extract listed daily rain only on days that had measurable rain and omitted the dry days from that list. That is a legitimate sparse rain list. It is not a seven-row completeness table. A reader who treats “two rain lines” as “only two days existed” has misread a filter.
Comparison to a normal
A normal is a 30-year average for a named official station and a named period. NOAA’s current U.S. Climate Normals are the 1991–2020 set published by NCEI (U.S. Climate Normals). Weekly or daily normals exist for official sites; they do not automatically exist for a backyard logger.
An honest comparison therefore has three names:
- The PWS ID and location.
- The normals station (for Mesa-area work, an official Phoenix-area climate site—not the PWS itself).
- The normals period (1991–2020, or an earlier period if you are reading an old product).
Without those names, “two degrees above normal” is an invented climate. A rooftop or backyard exposure can run warmer than an airport ASOS on still nights and cooler on irrigated lawns. The difference is siting, not a forecast skill score.
If you still compute a departure, label it derived and state the mismatch: different instruments, different height, different surface, different day boundary. Do not title the week table “NOAA weekly climate.”
NWS Climatological Report products (CLI daily, CLM monthly) include departures when the site is a designated climate location. A Weather Display week report is not those products. It is a station extract that sometimes looks tabular enough to be mistaken for them.
Distinct from the monthly NOAA-style HTML
/parse_noaa.php is the monthly climatological summary: up to 28–31 daily lines plus monthly totals, in a NOAA-like layout. This URL is the short window—seven rows, or a week-to-date, or a daily-report glance.
Use the weekly table to answer: did this station have a wet spell, a heat burst, or a missing-data hole in the last few days? Use the monthly HTML to answer: what did the calendar month look like as a whole? Do not average seven PWS days and call the result the monthly climate of Mesa.
The identifier page /weather/KAZMESA12.html is the provenance key for the same station. A week table without that ID is seven anonymous rows.
How to read a PWS weekly table in practice
Check the week boundary. Meteorological weeks and ISO weeks differ. A station that resets “this week” on Sunday morning will split a Saturday-night storm.
Check units once, in the header. Mixed °F means and °C extrema in hobby HTML are a known corruption. Do not convert in place without labeling a derived column.
Do not sum sparse rain lists as if missing days were listed. If only wet days appear, the weekly total is the sum of those lines only when the operator also asserts that omitted days were dry. That assertion should be written, not inferred.
Treat all-time records as logger memory. A “record high gust” on a weekly page is the maximum in the station archive since the logger was last reset or installed, not an NCEI statewide record.
Keep forecasts off the climate table. A WXSIM or NWS period forecast is a different URL family on this site (/wxsim.php). Climate rows are observed (or missing). Forecast rows are not.
Practical checklist
- Title the table with start date, end date, station ID, and units.
- Show seven date-labeled rows, or fewer with an explicit incomplete-week note.
- Use distinct tokens for trace, zero, and missing.
- Compute week summaries only over valid days; publish the valid-day count.
- If you show a departure from normal, name the normals station and period.
- Do not paste a month-to-date block under a “week” heading without relabeling.
- Do not present April 2005 Mesa extracts as current climate.
Modern relevance
Short climate windows are how people judge a storm week or a heat episode. They are also how missing data hides: two blank days vanish in a monthly mean more easily than in seven rows. Provenance and completeness for public weather records are the same discipline TNET describes in data sources, quality controls, and methodology.
Observation hub: /legacy/observations/.
Sources
- NOAA National Centers for Environmental Information, U.S. Climate Normals: ncei.noaa.gov
- National Weather Service Instruction 10-104, climate products (CLI/CLM; missing as
MM): weather.gov directives (PDF) - American Meteorological Society, Glossary of Meteorology, “trace”: glossary.ametsoc.org
- TNET, Data sources, quality controls, and methodology