Sign inRegisterCheck connection

Station Configuration as Scientific Metadata

Station configuration as scientific metadata: siting, sensor suite, logger, and published-data access. Why those records matter when interpreting observations.

Back to Personal weather station guides

A temperature on a personal weather website is not self-explanatory. It is a number produced by a sensor in a place, read by a logger, processed by station software, and published through a computer that may be rooms away from the instrument. Station configuration is the metadata that makes that number interpretable: siting, sensor suite, logger and PC, and how the published files can be read. Without those records, a visitor cannot tell a rooftop thermometer from a reference exposure.

The historical TNET Weather page at /configuration.php described one Mesa, Arizona station's equipment, accessibility markup, and network gear. This article does not treat that page as a live equipment boast, does not assert that the same hardware still runs, and does not sell a brand. Davis-class integrated stations appear here as an example architecture. Official siting language comes from WMO guidance, not from a product brochure.

Historical context

Personal-weather-station sites of the 2000s often included a "station" or "equipment" page because neighboring hobbyists, CWOP quality reports, and later researchers needed to know what they were looking at. The historical TNET page mixed several kinds of information that are worth separating:

  • Scientific siting and sensors (where the instruments sat, what they measured).
  • Publication accessibility (whether the website could be read by more than one kind of browser or assistive tool).
  • Software chain (station program, helpers that shared a logger among programs, lightning software).
  • Hosting and network furniture (servers, firewalls, ISP names) that do not belong in a climate record.

This rewrite keeps the first three as metadata problems. It drops the live-site tone, the obsolete browser scorekeeping, and any implication that TNET Weather today operates that backyard.

Why configuration records are part of the observation

Meteorological measurements are always relative to exposure. The World Meteorological Organization's Guide to Instruments and Methods of Observation (WMO-No. 8) is the authoritative reference for how instruments should be sited and operated. WMO and ISO also publish a siting classification for surface stations on land (ISO 19289:2014; WMO-No. 8, Part I, Chapter I, Annex 1.B in the edition cited on the WMO IMOP page). The classification is a 1–5 scale per variable that estimates how well a site matches recommended exposure. Class 1 is treated as a reference exposure; class 5 is an environment poorly suited to representing a wide area.

WMO is explicit that the number is not a moral ranking of the station's worth. A roadside or rooftop site can be the right instrument for a local question and still be a poor stand-in for a region. The class tells a later user how much metadata they must read before using the series. That is exactly the job of a PWS configuration page.

A configuration record should therefore answer, in words a stranger can use years later:

  • What is being measured, with what sensor class?
  • At what height, over what surface, with what nearby obstacles and heat sources?
  • Which values are observed, which are derived (dewpoint, heat index), and which are forecast includes from a weather service?
  • Which software wrote the published files, and at what cadence?
  • How can a machine or a person who cannot see a Flash gauge still get the numbers?

Siting, at a conceptual WMO level

WMO exposure rules exist because nearby objects bias instruments in known directions.

Temperature and humidity want a naturally vegetated, representative surface and a radiation shield that keeps the sensor from reading sun-warmed plastic or asphalt. Pavement, rooftops, exhaust, and walls add heat. Fan aspiration, when present, is an attempt to reduce radiation error, not a guarantee of class-1 siting. Peer-reviewed field experiments cited in WMO siting discussions have documented asphalt-road influence on temperature; the mechanism is ordinary: the sensor measures its immediate environment.

Wind is more demanding. Standard meteorological wind is a 10 m measurement over open terrain. WMO material notes that few sites meet the most open wind class because obstacles affect flow at surprisingly large distances. A rooftop anemometer measures rooftop flow. That can be the quantity the operator wanted (local gusts at the house) and still be a poor regional wind.

Precipitation is reduced by wind undercatch and by obstacles that shelter the gauge. A rain collector next to a wall does not record the same catch as an open gauge. Height and surrounding objects belong in the notes even when the funnel looks level.

Radiation and sunshine need a sky view. A solar or UV sensor under a tree line is a vegetation instrument.

Representativeness versus purpose. A backyard station is often a local climate instrument: garden, school, or neighborhood. That purpose is legitimate. It becomes misleading only when the website implies an official regional observation. Configuration text should state the purpose so that a later user does not have to guess.

Personal stations will rarely be class 1 for every variable. Recording that fact is more useful than claiming "carefully considered" placement without measurements of height, surface, and obstacles.

Sensor suite as an instrument list

A complete PWS configuration lists each measurand as if it were a separate instrument, because it is.

Air temperature and humidity. Usually a combined probe in a shield, sometimes fan-aspirated. Indoor sensors, if published, are building measurements and must be labeled as such.

Pressure. Often inside the console. Elevation and reduction to sea level, if applied, are software settings, not extra sensors. Publish the station elevation with the pressure series.

Rain. Tipping-bucket or other collector; funnel height; any heating; known splash or tree-drip sources.

Wind. Anemometer and vane height above the true local surface, not above a roof if the scientific question is 10 m open-field wind. Wireless links add dropout as a data-quality mode.

Solar and UV. Optional on Davis-class plus packages and other suites. They are radiometric observations with their own siting class.

Lightning. A separate instrument class (directional antenna or other detector plus dedicated software). A hobby lightning map is not an official warning. For life-safety decisions, use National Weather Service alerts. TNET is not a warning service.

Davis-class integrated stations are a convenient example because they package several of the above into an Integrated Sensor Suite plus a separate anemometer and a console/logger. The architecture matters: one wireless ISS on a mast, wind often elsewhere, console indoors. Configuration notes should say which physical boxes exist and where each sits. Naming a commercial family is identification of an instrument class, not a recommendation to buy it.

Logger, computer, and software chain

The logger is the clocked memory of the station; the PC is the publisher. Configuration should name:

  • the station software (Weather Display, Cumulus, or another) and its version;
  • helpers that multiplex one logger to several programs, because two programs opening one serial port is a classic data-loss mode;
  • the upload path (FTP/SFTP, interval, atomic rename);
  • extra processes (webcam, lightning PC) that do not share the meteorological logger.

A number on a website that passed through two PCs and a lightning workstation is still an observation, but the failure modes multiplied. The Weather Display publishing article treats the file side of that chain.

Derived indices deserve an explicit line. The NWS heat index is a derived apparent temperature from temperature and humidity, after Steadman (1979) and the operational Rothfusz regression. Publishing it next to air temperature without saying so invites users to treat it as a second thermometer.

Accessibility of published data

The historical page devoted long copy to XHTML, WCAG, and § 508. The scientific remainder of that concern is still valid: an observation that exists only inside a plugin, a canvas gauge, or a webcam is not fully published. A configuration record should say how a person or a parser gets the numbers:

  • text or HTML tables with units and timestamps;
  • compact files (ClientRaw, realtime.txt) with a documented schema;
  • captions and alternatives for images that encode values;
  • a statement that some media (video, older plugin viewers) are not data.

Accessibility here is data-access, not a compliance trophy. If the only "live" view requires a deprecated runtime, the durable product is the text file.

Independent quality feedback

Many U.S. operators contribute to the Citizen Weather Observer Program. NOAA's MADIS documents CWOP as a public-private partnership that collects citizen observations, makes them available, and returns quality-control feedback to contributors. That feedback is an independent check against nearby stations. It does not make a backyard site official, and a good QC email does not freeze siting forever: trees grow, a new shed appears, a fan fails.

A configuration page should record whether the station participates in CWOP or similar networks, the identifier if it is a public ID, and the date that participation was last confirmed. Do not paste an old quality URL as if it were today's score.

Practical checklist: a configuration record worth keeping

  1. Coordinates, elevation, and timezone of the instruments, not of the web host.
  2. Photo or sketch of exposures (even a described sketch): surface, mast height, nearest buildings and pavement.
  3. Per-sensor list: manufacturer class, height, units, extra notes (aspiration, wireless, heating).
  4. What is derived: dewpoint, wind chill, heat index, software forecasts.
  5. Software and build, logger model, upload interval, file names published.
  6. How to read the data without the pretty page.
  7. Hazard disclaimer: the site is not an official warning service.
  8. Change log: when the anemometer moved, when the shield was replaced, when the station software changed. A move is a break in the series.

Modern relevance

Metadata standards have only become more important as PWS data flow into MADIS, peer-reviewed urban-climate papers, and machine-readable APIs. A 2005 equipment page that only said "wireless Davis in the backyard" is thin by WMO siting-classification standards, but it is still better than a gauge with no notes. Operators who write configuration as if a stranger will ingest the series in 2032 are doing the same work national networks do with station histories.

TNET Weather does not operate the historical Mesa backyard as a current showcase, does not claim WMO certification for any PWS, and does not replace NWS warnings. The /configuration.php URL remains the place to teach why those records matter.

TNET research bridge

Siting, units, software version, and publication cadence are quality-control inputs. They decide whether a public weather record is usable as evidence. TNET's data sources, quality controls, and methodology page is the modern statement of how public weather and network records are treated in connection research, without disclosing proprietary assembly rules.

Related historical pages: the weather-station hub, Weather Display as a publishing system, and notes on how the station published data.

Sources