Sign inRegisterCheck connection

Monthly Trends from a Single Personal Weather Station

Monthly PWS charts need sample size and missing-day handling. NCEI climate normals are context for a backyard series, not proof of climate change.

Back to Cumulus weather software guides

A monthly trend chart from one personal weather station (PWS) is a summary of that station’s logger for calendar months. It can show whether this July was warmer or wetter than last July at this mount. It cannot, by itself, demonstrate regional or global climate change. Sample size, missing days, siting changes, and comparison to official normals decide whether the picture is a station diary or a misleading poster.

The historical URL /trends/monthly published that monthly view for a Cumulus site. The original plots are not rehosted, and this page does not present current monthly climate for any city. It is a methods article for reading a single-station monthly series.

Historical context

Cumulus and similar loggers accumulated daily highs, lows, means, and precipitation, then rolled them into monthly tables and simple trend graphics. Operators wanted a page that answered “how unusual was this month?” Incoming links treated those graphics as climate. The honest product is smaller: a backyard monthly climatology with error bars made of missing days.

Daily temperature charts on this host—see temperature time-series—are the usual input. A sample graph is not an input. If the monthly PNG was actually a demo, it has no climatological meaning.

Sample size is the first number on the chart

A month is not “30 numbers.” It is however many valid days you actually have, for each variable.

  • Temperature means need a rule: mean of daily means, or mean of all logger samples. Those differ when sampling is irregular.
  • Extremes (monthly maximum and minimum) are sensitive to a single bad spike from a sun-loaded sensor.
  • Precipitation totals are sums. A missing day is not a zero-rain day. Treating missing as zero dries the month.
  • One year of months is a seasonal cycle, not a trend. A trend is a change over many years at a stable site.

State n on the figure: days used, days possible, days missing. A July mean built from 18 days is a different statistic from a July mean built from 31. Do not connect monthly points with a line that implies a continuous climate process through months that were not observed.

A few years of a rooftop station are a short, biased series. Official U.S. climate normals are 30-year statistics from quality-controlled network stations, currently the 1991–2020 normals issued by NOAA NCEI. That mismatch of length and quality is the point, not a footnote.

Missing days and other quality holes

Before drawing a trend, mark days that fail a basic quality screen:

  • logger outages and clock jumps;
  • physically impossible values;
  • known ventilation or wetting failures;
  • days after an undocumented station move.

WMO observing practice treats exposure and continuity as part of the measurement. A PWS that moved from a roof to a shaded yard in March has a discontinuity that will look like a climate shift on a monthly chart. Split the series at the move. Do not run a single regression across the break and then caption it as “warming.”

Trace precipitation, unit changes (°C vs °F, mm vs inches), and reset rules for “rain today” all leak into monthly totals if you ignore them. Document the logger’s midnight and timezone. A station that changed from local time to UTC will shift which samples fall into which calendar month.

NCEI normals are context, not a verdict

It is reasonable to plot this month’s station mean next to the 1991–2020 normal for a nearby NCEI station. That plot answers: “how does my backyard compare to a 30-year official baseline nearby?” It does not answer: “has the climate changed?”

Reasons the comparison is context only:

  1. Different instruments and siting. A Stevenson-class screen over grass is not a roof vent. WMO siting classes exist because those differences are systematic.
  2. Different periods. Three years of PWS data versus thirty years of network data.
  3. Different quality control. NCEI normals account for missing values and station changes with documented methods. A hobby average often does not.
  4. Spatial mismatch. The nearest official station may be an airport, a valley, or a city park. Your yard is not that site.
  5. Normals are not a trend analysis. NOAA publishes normals as a “ruler” for typical conditions. Detection of climate change uses other, longer, homogenized datasets and formal methods. One PWS monthly page is not that literature.

If you show a departure from normal, label it as a departure from the named official station’s normal, for the named month, with the PWS n and siting note attached. Do not title the page “climate change.”

What a monthly PWS chart can honestly claim

It can claim:

  • this station’s monthly mean, extreme, or total, with n and units;
  • a year-to-year difference at the same mount, if siting was stable and missingness is comparable;
  • a seasonal cycle at this site (summer warmer than winter) once you have at least one full year of adequate days;
  • a comparison-to-normal as a local curiosity.

It cannot claim a regional trend, a cause, or an official climate monitoring result. It cannot substitute for NWS forecasts or warnings. A hot monthly mean at a rooftop PWS is not a heat advisory.

Practical checklist

  • Caption names station, location, variable, unit, and month definition.
  • Days used and days missing are printed.
  • Missing precipitation days are not zeros.
  • Station moves and shield changes split the series.
  • Any official normal is cited to NCEI (or another national climate service) with station name and normal period.
  • Language says “this station,” not “the climate.”
  • No sample or demo file is used as a monthly source.

Modern relevance

Single-site monthly summaries remain a good station diary and a bad climate paper. Regional weather and performance studies need many places, stated periods, and labeled evidence types. TNET’s public collection of that work is regional performance and weather studies. How TNET groups atmospheric observations among other inputs is described under evidence families. The Cumulus legacy hub lists the related publishing pages.

This URL does not update monthly climate. It states how to read the historical monthly-trend product without overclaiming.

Sources