Sign inRegisterCheck connection

Why Cumulus Sites Kept /graphs and /graphs2

Why Cumulus sites kept both /graphs and /graphs2: a second collection for longer periods, extra sensors, and a wind rose besides the daily plate.

Back to Historical weather observations and forecasts

/graphs2 is a second graph collection, not a duplicate of the daily plate and not a login form. Sites kept both /graphs and /graphs2 because the first URL had already become a contract: today’s temperature, dewpoint, pressure, wind, and rain. When operators needed a longer window, extra sensors, or a wind rose, they added a second path rather than silently changing what /graphs meant. Two URLs are the honest versioning of two products.

The historical URL /graphs2 on cumulus.tnetweather.com was that second collection. The original images are not rehosted. This article explains what belongs in the overflow set and why merging the two paths would erase a real distinction. The daily plate is /graphs. The second collection’s gate is /graphs2&action=login. Neither login essay is this page.

Historical context

PHP weather sites reused a numbering habit: graphs for the original pack, graphs2 for the rewrite. Sometimes the rewrite was a new JpGraph major version. Sometimes it was extra hardware (solar, UV, soil, extra temperature channels) after the core five shipped. Sometimes it was climatology-scale images beside the 24-hour set. Sometimes it was a wind rose that never fitted the daily strip. The public menu showed two entries. Crawlers kept both.

A single article under /graphs cannot represent both collections without lying to one of them. Preservation keeps two observation articles and two access-control articles. TNET does not restore the scripts, ZIPs, or member cookies. JpGraph remains third-party software (jpgraph.net, last checked 13 August 2026).

Three reasons a second collection exists

1. A longer period is a different experiment

The daily plate answers “what did the last day do?” A 7-day, 31-day, or month-to-date set answers a different question. Amplitudes, missing-day fractions, and what counts as a “spike” all change with window length. A wind gust that dominates a 24-hour panel is a brief mark on a monthly panel. Rain that looks like a burst on /graphs becomes a single bar on a monthly rain chart.

If you put both windows on /graphs without renaming the URL, inbound links that meant “today’s traces” start downloading climate-scale PNGs, or the reverse. /graphs2 as the longer set keeps the first URL stable.

A longer series is still this station, not NCEI normals. Sample size and missing days matter more as the window grows; that methods problem is developed for monthly PWS charts at /trends/monthly. /graphs2 only needs the rule: label the period on every panel, and do not draw a 31-day figure at the same pixel width as a 24-hour figure without saying so.

2. Extra sensors are extra experiments

Solar radiation, UV index, soil temperature, leaf wetness, extra air-temperature channels, and indoor-only traces are real instruments when they exist. They are not part of the standard daily atmosphere plate. Dumping them onto /graphs trains visitors to treat soil at 10 cm as “the weather.”

/graphs2 is the right attic:

  • each extra series named (soil 1 versus extra temp 3);
  • units (W m⁻², UV index dimensionless, °C or °F);
  • height or depth in the caption;
  • a note when the hardware is absent so the panel is not a flatline sold as night.

Unused extra channels in a packet often hold zeros. A second collection that plots zeros as living soil temperature is worse than omitting the panel. Absence should look like absence.

3. A wind rose is not a time-series

A wind rose summarizes direction frequency (and often speed bins) over a chosen period. It is a polar histogram, not a line of gust versus time. It answers “from where did the wind blow during this month?” The daily plate’s wind panel answers “when did it blow today?”

Those are complementary. They must not share a title. A rose with no period, no calm threshold, and no speed bins is a logo. A rose built from 12 samples is not a climatology. State n, the period, the calm class, and whether direction is instantaneous or a vector-mean.

The rose belongs in /graphs2 because it would break the daily plate’s shared time-axis scan. It is a summary graphic, not another 24-hour strip.

Why not overwrite /graphs in place

Operators were tempted to “just add panels” to the first page. That produces three archival failures:

  1. Link rot of meaning. A forum credit to /graphs meant the daily five. After a silent expansion it means a 40-panel attic. The bytes at the URL changed job.
  2. Renderer load. Extra charts, especially PHP-drawn PNGs, multiply CPU. The login-gate article is about that cost for the first gallery; a second collection is a second budget. Putting both on one anonymous URL invites one hotlink to pull everything.
  3. Schema drift. A longer log or a new column map can break the original scripts. A parallel /graphs2 lets v1 keep working while v2 uses a new log. That is also why /graphs2&action=login is a separate authorization object: a cookie for v1 must not silently authorize v2.

Keeping two public paths is therefore a data-source decision: two products, two freshness expectations, two schemas. It is not a cosmetic duplicate.

How to read graphs2 without confusing it with graphs

Start by reading the period and the inventory. If the page is a week of solar plus a monthly wind rose, do not scan it as “today.” Then:

  • Use /graphs for the last day’s co-occurrence of the standard five.
  • Use /graphs2 for extra hardware, longer windows, and directional summaries.
  • Use gauges for the current simultaneous slice.
  • Use the JPGraph project page only if the question is how a PHP pack drew figures, not what the second collection contains.

If /graphs2 actually served the same PNGs as /graphs, the second URL had already failed its job. Preservation still does not merge them: the intent of a second collection is what inbound links to /graphs2 were pointing at.

Practical checklist

  • Period is on every panel and is not silently “today.”
  • Extra sensors are named, sited, and omitted when unused.
  • Wind rose has n, period, calm class, and speed bins.
  • Daily standard five remain on /graphs, not duplicated unlabeled here.
  • No login form on this article; access control is a sibling URL.
  • Longer rain or temperature series do not claim to be climate normals.
  • Missing samples stay missing as the window lengthens.

Modern relevance

A second collection is how you extend a feed without breaking the first contract. Extra sensors and longer series are provenance and coverage questions: what was measured, for how long, and whether the hardware was actually there. TNET’s public note on source quality is data sources and methodology. The Cumulus hub and observations hub list the related pages.

This URL does not restore a second live gallery. It states why /graphs and /graphs2 were both worth keeping.

Sources