Sign inRegisterCheck connection

Publishing a Weather Site’s Generator Stack as Scientific Metadata

PHP, Cumulus, template name, and last-updated are scientific metadata on a personal weather site—not a credits page and not a live software inventory.

Back to Cumulus weather software guides

A “website info” page on a personal weather site is not a vanity footer. It is a generator stack published as metadata: which station program built the files, which PHP skin wrapped them, which template name to cite, and when the public HTML was last produced. At /aboutus.websiteinfo the historical Cumulus host kept that stack next to the rest of the About section.

This article treats those fields as scientific metadata for a published view of observations. It is not a live inventory of what TNET runs today, not a credits roster, and not a substitute for the outpost identity page. Modern TNET Weather does not maintain Cumulus and does not claim the historical generator as its production stack.

Historical context

Cumulus processes web tags on the operator’s computer and uploads HTML, images, and optional data files. The public page a visitor sees is therefore a generated artifact. Its meaning depends on the generator the same way a plotted figure depends on the software that drew the axes.

Community skins made that explicit. Footers printed a Cumulus version, a template name, a PHP mention, and a last-updated stamp. Some skins collected those lines on a Website Info or System page so the home view stayed uncluttered. HTML also has a long-standing meta name="generator" convention for naming the producing software (WHATWG HTML Standard; last checked 13 August 2026). Weather sites were doing the same job in visible prose.

The Cumulus Wiki Webtags page documents the current tag vocabulary, including tags that emit software version and related status into processed pages (last checked 13 August 2026). Those tags are how a template could print the stack without typing it by hand every hour. The original info screen at this URL is not rehosted. The fields still matter.

The stack as a metadata record

Write the generator the way you would label a figure in a notebook.

| Field | What it is | What it is not | |---|---|---| | Station program | Cumulus or Cumulus MX, with version | A claim that TNET ships Cumulus | | Build or tag set | The web-tag dialect that produced the HTML | A live forecast engine | | Template or skin name | The PHP/HTML family | The logger | | Runtime | PHP (and major version if known), GD, plotting library | The atmosphere | | Last-updated (page) | When the HTML or include was generated or uploaded | The time of the meteorological event | | Last-updated (data) | When the packet or tags were filled from the logger | The page’s cache time | | Hosting note | Shared host, home server, or other, if published | A security audit |

Two clocks are easy to confuse and fatal to mix.

Data time is when the logger (or Cumulus) last wrote the values. A stale realtime.txt can sit on the server with yesterday’s afternoon in it. That is a quality-control event, not a design flourish.

Page time is when the template was processed or when the file landed on the host. A skin can rebuild hourly while the logger is stuck. A skin can also be static HTML that still shows a fresh packet in an iframe. Website info must say which clock it is printing.

Native units, latitude, and elevation belong with station metadata. They may appear on a website-info page, but they are not PHP. Keep them labeled as station fields so a later scrape does not file “PHP 5.3” as a climate attribute.

Why a published view needs a generator

Personal weather HTML is often reused as if it were a data file. People screenshot gauges, scrape tables, or cite a URL as “the station.” Without a generator stack, that citation is incomplete.

  • Version drift. Cumulus field lists and tag behavior changed across versions. MX documents unknown tags as error strings rather than silent blanks (Webtags). A table copied from a 2009 skin is not the same object as a table from a current MX build.
  • Unit and locale formatting. The skin may have converted or locale-formatted numbers. Generator plus native units tells a later parser whether a comma is a decimal.
  • Derived chrome. Banners, JPGraph traces, and translated labels are derived. The stack names the deriving tools.
  • Stale-page detection. A last-updated stamp that never moves while the operator believes the site is live is evidence. It is not weather.

WMO and national services encode instrument and processing metadata because a number without a method is not an observation you can reuse. A hobby website-info page is a coarse version of that habit for the web view. It does not make the station official. The NWS remains the warning authority; NCEI remains the place for official U.S. climate archives.

What to publish, and what to omit

Publish enough to interpret the HTML:

  • software name and version;
  • template name;
  • whether pages are tag-processed uploads or host-side reads of realtime.txt;
  • the two clocks, named;
  • a date on the info page itself.

Omit from a public info page:

  • passwords, FTP accounts, and panel URLs;
  • exact internal directory layouts that only help an attacker;
  • claims of “100% uptime”;
  • a live dump of phpinfo() or full module lists.

The historical culture sometimes printed more runtime detail than a 2026 operator should. Metadata is not a request to expose the server.

Do not treat this URL as a current TNET production manifest. The modern service’s public methodology lives elsewhere. This path documents the idea of a generator stack on a Cumulus About page.

Practical checklist

  1. Name the station program and version on a stable page, not only in a footer that a redesign will delete.
  2. Name the template. “Custom” is not a name.
  3. Split data time from page time. If you cannot, say so.
  4. Keep native units with the station, display units with the view.
  5. Follow current tag docs if you still process Cumulus pages (Webtags; Software; MX GitHub; last checked 13 August 2026).
  6. When you migrate, archive a copy of the website-info page with the HTML capture so later readers know the generator.

The About-section project narrative explains why tools were built. Website info explains which tools produced this view. Keep those jobs separate.

Modern relevance

MX JSON endpoints and static site generators still produce views. A generator meta tag, a README, or a small /status page that names software, template, and freshness is the same metadata. Raspberry Pi stations that publish to the public web without a stack note produce numbers that look authoritative and cannot be versioned.

TNET does not operate this hostname as a Cumulus version tracker.

TNET research bridge

Source identity, units, and freshness of weather records used in connection research are discussed publicly in data sources, quality controls, and methodology. That page is not a Cumulus footer. The Cumulus legacy hub lists related historical pages.

Sources