Accurate Weather Stations: What Accuracy Actually Means

As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.

What accuracy means for each sensor on a weather station

Accuracy is not one number. Each sensor on a weather station has its own error characteristics, and the weakest sensor limits what you can claim about the station as a whole. The listing bullets for consumer stations rarely publish accuracy figures at all. Where they do, the figure usually applies to one sensor under specific conditions, not to the station in your garden.

Temperature accuracy is typically quoted as plus or minus a fraction of a degree, but the dominant error in a backyard station is not the sensor. It is solar heating of the housing. A temperature sensor in a poorly aspirated radiation shield can read several degrees high in still, sunny conditions. The listing for the Ambient Weather WS-2000 states that the integrated sensor array measures temperature, humidity, wind, rainfall, UV, and solar radiation, but it does not publish a temperature accuracy figure. Not stated in the listing bullets. The same is true for the WS-2902 and the Wittboy.

Humidity accuracy is usually quoted as a percentage of relative humidity, for example plus or minus two percent. Consumer humidity sensors drift over time, especially in prolonged high humidity or salt air. A sensor that was accurate when new may be several percent off after a few seasons. None of the listings here publish a humidity accuracy figure, so you cannot compare them on that basis.

Wind speed accuracy depends on the anemometer type. Cup anemometers have a starting threshold, the wind speed below which they do not turn. Below that threshold they read zero. The listing for the Tempest states that it has a built-in wind meter and delivers updates every three seconds, but it does not publish a wind speed accuracy figure or a starting threshold. Not stated in the listing bullets. The same applies to the other stations here.

Rainfall accuracy is limited by the gauge design. Tipping bucket gauges undercount in heavy rain because the bucket cannot tip fast enough. The ECOWITT Wittboy uses a haptic rain sensor, which the listing describes as measuring rainfall volume without vibration and not affected by harsh weather. That is a different measurement principle from a tipping bucket, but the listing does not publish a rainfall accuracy figure, so you cannot say from the listing alone whether it is more or less accurate than a conventional gauge.

Solar radiation and UV sensors are usually the least accurate sensors on a consumer station. They are useful for relative comparisons, such as whether today is sunnier than yesterday, but they are not calibrated instruments. The listings for the Ambient Weather WS-2000, the Ambient Weather WS-2902, and the WeatherFlow Tempest Weather state that these sensors are present, but none publish accuracy figures.

Siting: the error you control

You can buy a station with excellent sensors and still get poor data if you site it badly. Siting is the largest source of error in backyard weather measurements, and it is entirely under your control.

Temperature and humidity sensors should be mounted over a surface that does not store heat, ideally grass or bare soil, at a height of around 1.2 to 1.5 metres above ground. They should be away from walls, fences, and paving, which radiate heat. They should be in a radiation shield that allows air to flow past the sensor but blocks direct sun. Many consumer stations integrate the temperature sensor into the same housing as the wind and rain sensors, which means you cannot optimise the height for both. Wind should be measured high, typically ten metres above ground in open terrain, while temperature should be measured low. A single integrated sensor array forces a compromise. The listing for the WS-2000 describes an all-in-one integrated sensor array, which means one mounting height for all sensors. The same is true for the other stations here.

Rain gauges should be mounted in an open area, away from obstructions that can create turbulence or shelter the gauge. A gauge mounted on a pole next to a wall will undercatch in wind-driven rain. The Wittboy listing mentions bird spikes, which address one common problem, but the listing does not describe the gauge’s exposure requirements.

Wind sensors should be mounted as high as practical and away from buildings and trees. Turbulence from a nearby roof can cause the anemometer to read high in some directions and low in others. The listing for the Tempest states a wireless range of up to 1,000 feet, which gives you flexibility in siting the sensor away from the house, but the listing does not state a recommended mounting height.

If you cannot site a station well, the data will be biased regardless of the sensor quality. A station in a small urban garden will not give you representative wind or temperature for the wider area. It will give you data for that garden, which may be exactly what you want, but you should not present it as regional weather data.

Calibration and the difference between resolution and accuracy

Resolution is the smallest change a sensor can display. Accuracy is how close the displayed value is to the true value. They are not the same, and consumer listings often conflate them. A temperature display that reads to 0.1 degrees has a resolution of 0.1 degrees, but its accuracy might be plus or minus one degree or worse. The listing for the WS-2000 states that calibration is available, which means you can apply an offset to bring the reading closer to a reference. That is useful for temperature and humidity, where a simple offset can correct a consistent bias. It is less useful for wind speed, where the error varies with wind speed and direction.

Calibration requires a reference. You cannot calibrate a sensor against itself. For temperature and humidity, you can compare against a known-good thermometer or a sling psychrometer. For pressure, you can compare against a nearby official station at a known elevation, but you must correct for the difference in altitude. The listing for the Tempest states that it monitors barometric pressure, but it does not state whether the pressure is corrected to sea level or reported as station pressure. Not stated in the listing bullets. That matters because station pressure and sea-level pressure differ by roughly 1 hPa per 8 metres of elevation, so a station at 100 metres elevation will read about 12 hPa lower than sea-level pressure. If you compare an uncorrected station pressure to an official sea-level pressure, you will see a large apparent error that is not a sensor error.

Relative humidity calibration is best done at two points, for example over a saturated salt solution, but that is beyond what most homeowners will do. A simpler approach is to compare against a trusted reference in stable conditions and apply an offset if the error is consistent. The listings do not publish calibration procedures, so you are relying on general knowledge of the sensor types.

Rain gauge calibration is usually done by dripping a known volume of water through the gauge and checking the count. The listing for the Wittboy describes a haptic rain sensor, which does not have a tipping bucket to calibrate in the traditional way. The listing does not state how the haptic sensor is calibrated or whether user calibration is possible. Not stated in the listing bullets.

What the data is good for, and what it is not

A consumer weather station is good for local, relative measurements. It tells you whether it is warmer today than yesterday, whether the wind is stronger than this morning, and whether it rained more than last week. It is useful for gardening, for deciding whether to water, for logging conditions for a hobby, and for feeding a smart home system. The listing for the WS-2902 states that it supports IFTT, Google Home, and Alexa, which means you can use the data to trigger actions, such as closing blinds or turning on a fan.

It is not good for official records, for legal or regulatory purposes, or for research-grade measurements. The World Meteorological Organization publishes standards for siting and sensor accuracy that consumer stations do not meet. If you need data that will stand up to scrutiny, you need a station that publishes its accuracy specifications and is sited according to those standards, or you need a different class of instrument entirely.

It is also not a substitute for a dedicated instrument when the measurement is safety-critical. A weather station does not measure air quality, and it does not detect gas leaks. If you need to monitor indoor air quality, you need a different sensor. If you need to detect carbon monoxide, you need a carbon monoxide alarm, not a weather station.

For agriculture and frost protection, a weather station can be useful, but you need to understand its limits. Frost prediction depends on temperature, humidity, wind, and cloud cover, and a station that measures these variables can help you decide whether to protect crops. But the station itself does not predict frost; it measures current conditions. You need to combine the data with a forecast model or your own experience. The best weather stations for farms page discusses this in more detail.

When a weather station is the wrong instrument

There are cases where a weather station is not the right answer, and it is worth saying so plainly.

If you need to measure wind at a specific height for structural or safety reasons, such as for a crane or a mast, you need a calibrated anemometer with a known accuracy and a mounting arrangement that meets the relevant standard. A consumer weather station is not that instrument.

If you need to measure rainfall for hydrological or engineering purposes, you need a tipping bucket or weighing gauge with a published accuracy and a known orifice size, sited according to the relevant standard. The consumer stations here do not publish those specifications.

If you need to measure temperature and humidity for building science or HVAC commissioning, you need a calibrated instrument with a certificate. A weather station is not a substitute.

If you need to monitor indoor conditions, a weather station is usually the wrong choice because its sensors are designed for outdoor use and its display is not intended for indoor monitoring. There are dedicated indoor air quality monitors that measure temperature, humidity, carbon dioxide, and particulate matter.

If you need a portable instrument for spot measurements, a handheld weather meter is more appropriate than a fixed station. The best portable weather meters page covers that category.

For most homeowners, a weather station is a useful and interesting instrument, but it is not a precision measurement device. If you want to compare specific models, the best weather stations page and individual reviews such as the Ambient Weather WS-2902 review and the WeatherFlow Tempest review go into more detail on the trade-offs.

The honest answer on accuracy

An accurate weather station is one whose sensors are adequate for your purpose and whose siting does not undermine them. For backyard monitoring, the stations here are adequate, but none of them publish the accuracy figures that would let you compare them on that basis. If you need published accuracy, you need a different class of instrument. If you want hyper-local data for your own use, choose a station with the sensors you need, site it as well as you can, and treat the readings as relative rather than absolute.

Frequently Asked Questions