Weather Stations

What is a Weather Station

A weather station is a coordinated set of meteorological sensors, a data acquisition and processing unit, and a power and communication subsystem, all installed at a representative location to quantify the near-surface atmosphere. The output is a time series of observations: air temperature, relative humidity, atmospheric pressure, wind speed and direction, and precipitation amount form the classic synoptic core, and many stations add solar radiation, soil parameters, visibility, and present weather. The value of a station lies not in collecting numbers but in collecting numbers whose uncertainty is known and traceable, which is why the field is governed by formal international guidance rather than convention.

Functionally a station has three layers. The sensing layer holds the transducers that convert each physical quantity into an electrical or digital signal: a platinum resistance thermometer for temperature, a capacitive element for humidity and pressure, a rotating cup or ultrasonic transit-time array for wind, and a tipping bucket or load cell for precipitation. The acquisition layer is the data logger, the brain of the AWS, which samples each channel at the prescribed rate, linearizes and averages the raw signal, applies the required reporting interval, and writes the result to non-volatile storage. The infrastructure layer supplies regulated power, often 12 VDC from a solar-charged battery, and telemetry that pushes data to a server by cellular, satellite, or radio link.

The distinction between a weather station and an isolated sensor is integration and discipline. A standalone anemometer reports wind, but a station reports wind averaged over the WMO-mandated 2-minute or 10-minute window, paired with the highest 3-second gust, time-stamped against a battery-backed clock, and bracketed by metadata describing the mast height and exposure. That processing chain is what makes the data comparable across a national network and admissible for climatology, aviation, hydrology, and energy forecasting.

Historically, surface meteorology began with the manned observatory. The mercury barometer dates to Evangelista Torricelli in 1643, the louvred Stevenson screen that shields thermometers from solar radiation was standardized by Thomas Stevenson in the 1860s, and the rotating-cup anemometer was developed by John Thomas Romney Robinson in 1846. For more than a century an observer read these instruments by hand at fixed synoptic hours. Electronic transduction, low-power microcontrollers, and reliable telemetry from the 1980s onward made the automatic weather station practical, and today the WMO Global Basic Observing Network and the dense regional mesonets it feeds are overwhelmingly automated.

In application scale, weather stations span from a single agricultural field station costing a few thousand dollars to reference climate stations and airport systems (such as the United States ASOS network) carrying redundant sensors, aspirated shields, and rigorous calibration regimes. Industrial users sit across this range: wind and solar developers need resource-assessment grade wind and irradiance data, ports and offshore platforms need marine-rated stations, and chemical plants need stations that drive dispersion models for emergency response. Selecting the right station is therefore an exercise in matching required uncertainty and reliability to the application, not in buying the most sensors.

Station Types and Architecture

Weather stations divide along two axes: degree of automation and intended application grade. The automation axis runs from fully manned conventional stations, through automatic weather stations, to compact all-in-one transmitters that fuse several sensors into a single solid-state housing. The application axis runs from consumer and hobby stations, through professional and industrial stations, to reference and synoptic stations built to feed international data exchange. The table below contrasts the main categories on the dimensions that drive procurement.

Manned conventional stations still exist where human judgement adds value, particularly for present weather, cloud type, and visibility, which automated sensors approximate but do not fully replicate. Their core instruments (mercury or aneroid barometer, paired dry- and wet-bulb thermometers in a Stevenson screen, and a manual storage rain gauge) are read at fixed synoptic hours, typically every three or six hours. Labour cost and the loss of overnight continuity have driven most networks to retire or hybridize these sites.

Automatic weather stations are the dominant modern form. An AWS is architecturally a star network: independent smart sensors connect to a central data logger over SDI-12, RS-485 Modbus, or analog wiring, and the logger handles sampling, averaging, quality flags, storage, and telemetry. The traditional AWS uses discrete, individually replaceable sensors mounted on a 10-metre wind mast and a 2-metre instrument tripod, which suits reference sites where each sensor can be calibrated and swapped on its own schedule.

All-in-one transmitters collapse several measurements into one compact, often solid-state body. The Vaisala WXT530 series is a representative example: it measures barometric pressure, temperature, humidity, rainfall, and wind speed and direction using ultrasonic wind sensing and capacitive pressure, temperature, and humidity elements, with no moving parts. This architecture minimizes maintenance, suits marine and roadside installations, and simplifies wiring, at the cost of having to replace the whole unit if one parameter fails and of accepting the integrated accuracy rather than best-in-class per-sensor performance.

Grade is the second axis and matters as much as type. A consumer station and a synoptic AWS may both report the same five variables, but only the synoptic station carries the aspirated shielding, the 10-metre standardized wind exposure, the calibration traceability, and the documented siting class that let its data be trusted at the 0.1 K and 0.1 hPa level. Buying a low-grade station for a high-grade application is the most common and most expensive procurement error in this category.

Sensor Technologies and Principles

Each meteorological variable is measured by a distinct physical principle, and the choice of principle sets the achievable accuracy, response time, maintenance burden, and cost. Understanding these principles is the foundation of intelligent selection, because two stations that both claim to measure wind or rain can behave very differently in ice, dust, or heavy precipitation. The table below summarizes the mainstream sensing technologies for the core variables.

Temperature is almost universally measured with a platinum resistance thermometer, a Pt100 or Pt1000 whose resistance rises predictably with temperature per the IEC 60751 curve. Platinum is chosen for long-term stability and traceable linearity. Because a bare element absorbs sunlight and reads high, the sensor must sit inside a radiation shield. The capacitive humidity element is frequently co-located in the same housing, forming a combined thermo-hygrometer, since temperature and humidity are physically linked through dew point.

Humidity is dominated by the thin-film capacitive sensor, a polymer dielectric between electrodes whose capacitance tracks the amount of absorbed water vapour. It is fast, mechanically robust, and shows low long-term drift, which displaced the older hair hygrometer and the maintenance-heavy wet-and-dry-bulb psychrometer. Its main weakness is hysteresis and slow recovery after prolonged saturation near 100 percent relative humidity, which is why datasheets quote a wider tolerance at the top of the range, for example plus or minus 3 percent up to 90 percent and plus or minus 5 percent above 90 percent on the Vaisala WXT530.

Pressure uses a capacitive silicon barometer (a sealed micro-machined cavity whose deflection changes capacitance) delivering high resolution and low drift across the meteorological range of roughly 600 to 1100 hPa. The practical challenge is that wind blowing across the pressure port creates dynamic pressure error, so reference installations use a static pressure head to decouple the barometer from wind. The data logger then converts station pressure to mean-sea-level pressure using the station altitude, the value used on synoptic charts.

Wind is the variable with the clearest technology split. The cup-and-vane anemometer measures speed from cup rotation rate and direction from a separate wind vane; it is inexpensive and well characterized but has bearings that wear, a starting threshold near 0.5 m/s, mechanical lag that limits gust fidelity, and a tendency to freeze in icing conditions. The ultrasonic anemometer measures the transit time of acoustic pulses between transducer pairs, resolving horizontal speed and direction with no moving parts, fast response, and optional heating for de-icing. For unattended, marine, or cold sites the ultrasonic sensor usually delivers lower lifetime cost despite a higher purchase price.

Precipitation is most commonly measured by the tipping-bucket gauge, which funnels rain onto a small see-saw bucket that tips and emits a pulse at each fixed increment (typically 0.1 or 0.2 mm). It is simple and low power but undercatches in heavy rain because water flows during the tip, struggles in high wind, and cannot measure solid precipitation without heating. The weighing gauge records the mass of accumulated precipitation continuously, captures snow as liquid equivalent, and resists evaporation and splash error, but it costs more, draws more power, and needs periodic emptying. Acoustic and optical disdrometers add intensity and drop-size data where a full precipitation profile is required.

Siting, Exposure, and Standards

The single largest source of error in surface meteorology is often not the sensor but its surroundings. A laboratory-perfect thermometer reads the wrong temperature next to a sun-baked wall, and a calibrated anemometer reports the wrong wind beside a building. International practice therefore couples sensor accuracy requirements with formal siting and exposure rules, and a credible station documents both. The governing references are WMO-No.8, the CIMO Guide, and ISO 19289:2014, the first jointly adopted ISO and WMO siting standard.

Radiation shielding is mandatory for the temperature and humidity sensors. The classic louvred Stevenson screen and the modern multi-plate Gill shield are naturally ventilated: cheap and maintenance-free, but capable of a radiation error above 0.5 K in strong sunlight with low wind. A mechanically aspirated shield draws air across the sensor at roughly 3 to 5 m/s with a fan, cutting the radiation error below 0.2 K, which is why reference and climate-grade sites accept the added power draw and fan maintenance. The shield is not optional dressing; it is part of the measurement.

Standardized exposure means putting each sensor where its variable is representative. WMO specifies the temperature and humidity sensors at 1.25 to 2 metres above short grass, and the wind sensors at the reference height of 10 metres above open ground, with obstacles ideally at least 10 times their height away. Wind measured at a non-standard height must be corrected with a logarithmic wind profile before it is comparable. The rain gauge orifice is set at a defined height with the rim level, and wind shields or pit gauges are used where wind-driven undercatch is severe.

The ISO 19289 siting classification grades how well a real site meets these ideals on a scale of 1 to 5, separately for temperature and humidity, precipitation, wind, and radiation. Class 1 is a reference site whose surroundings add negligible bias; class 5 is a site where nearby obstacles create an inappropriate environment with large added uncertainty over an area of tens of square kilometres. Reporting the class alongside the data tells downstream users how far to trust the headline accuracy. The table below summarizes the standards landscape a buyer should recognize.

Calibration traceability closes the loop. WMO-No.8 requires periodic inspection and calibration against standards traceable to national metrology institutes, with stations inspected at intervals appropriate to the network. A station that ships with a calibration certificate, a stated drift specification, and a documented recalibration interval is auditable; one that does not is a black box. For regulated applications such as aviation, port operations, and environmental compliance, this traceability is not optional.

Key Specification Parameters

Reading a weather station datasheet means separating the headline marketing accuracy from the parameters that actually constrain the application. The dominant specifications are the per-variable measurement range and uncertainty, the sampling and averaging behaviour, the operating environment, the output interfaces, and the power budget. The table below sets the WMO-No.8 Annex 1.D operational requirement against the published Vaisala WXT530 sensor-element accuracy so the gap between an ideal target and a real compact product is visible.

Measurement uncertainty is the headline number, but read it carefully. WMO-No.8 Annex 1.D states required and achievable uncertainties as end-of-chain field figures: 0.1 K for air temperature above -40 degrees Celsius, 0.1 hPa for pressure with 0.15 hPa achievable, 1 percent for relative humidity, 0.5 m/s for wind speed at or below 5 m/s rising to 10 percent above that, 5 degrees for wind direction, and the larger of 0.1 mm or 5 percent for daily precipitation. A compact transmitter that quotes a sensor-element accuracy looser than these targets is not WMO-compliant for the corresponding variable, which may be perfectly acceptable for an industrial site that does not feed international exchange.

Sampling and averaging determine whether two stations are even comparable. WMO requires pressure, temperature, and humidity to be reported as roughly 1-minute averages after linearization, wind as 2-minute and 10-minute averages, and the wind gust as the highest 3-second average. A station that reports an instantaneous spot reading rather than a defined average is not following the standard, and its data cannot be meaningfully merged with a network that does. The data logger sampling rate and the configurable averaging interval are therefore first-class specifications, not footnotes.

Operating environment governs survivability. The operating temperature range (for example -52 to +60 degrees Celsius on the WXT530), the storage range, the ingress protection rating, and standards such as IEC 60945 for marine use tell you whether the station survives where it will live. Heated sensor options matter wherever ice forms: an unheated ultrasonic head or tipping bucket can read zero in freezing fog or snow, silently losing data exactly when weather matters most.

Output and protocol is the integration interface. Smart sensors and transmitters typically offer SDI-12, RS-232, RS-485 or RS-422, and Modbus for digital integration, plus analog 4-20 mA or 0-10 V where a legacy controller demands it. The WXT530, for example, provides SDI-12, RS-232, RS-422, RS-485 and two milliamp current outputs. Confirm that the station speaks a protocol your data logger or SCADA already supports, and that documentation and configuration tools are available, before committing.

• SDI-12: The dominant serial standard for environmental sensors, addressable multi-drop, low power, simple wiring.
• RS-485 / Modbus RTU: Robust multi-drop bus for industrial integration over long cable runs.
• 4-20 mA / 0-10 V: Analog outputs for direct connection to legacy PLC or controller analog inputs.
• Telemetry uplink: Cellular, satellite (Iridium, GOES), LoRaWAN, Ethernet, or Wi-Fi from the data logger.

Power budget closes the specification. Most field stations run on 12 VDC from a sealed battery charged by a solar panel, so the worst-case current draw, especially with sensor heaters engaged, sizes the panel and battery. A station whose typical draw is a few milliamps can balloon to hundreds of milliamps or amperes when heating, so the winter heating load, not the fair-weather average, must drive the energy design for any site that sees ice.

Selection Decision Factors

To turn the preceding chapters into a model on a purchase order, follow the decision sequence below. Most selection failures come not from a single wrong sensor but from skipping the application-definition step and then over- or under-buying. These eight steps form a reusable RFQ template for any weather station procurement.

1. Define the application and required uncertainty: Decide first whether the data feeds international exchange or climate reference (must meet WMO-No.8 Annex 1.D), regulated operations (aviation, ports, environmental compliance), or internal industrial use (looser tolerances acceptable). The required uncertainty per variable drives every downstream choice.
2. Fix the variable set: Specify exactly which variables you need at what accuracy. Avoid paying for solar radiation, soil moisture, or visibility you will never use, but do not omit a variable that a forecast or dispersion model depends on.
3. Choose architecture: Discrete sensors on a mast for reference sites with per-sensor calibration and serviceability, or an all-in-one transmitter for marine, roadside, and distributed mesonet sites where low maintenance and simple wiring outweigh best-in-class per-sensor accuracy.
4. Select sensor technologies: Ultrasonic versus cup-and-vane wind, tipping versus weighing precipitation, naturally ventilated versus aspirated radiation shield. Match each choice to the climate, especially icing, dust, and salt exposure.
5. Plan siting and exposure: Target an achievable ISO 19289 class for each variable. Secure the 10-metre wind exposure and the 1.25 to 2-metre shielded temperature height, and document obstacles. Siting often constrains achievable accuracy more than the sensor does.
6. Specify environment and ruggedness: Operating and storage temperature, ingress protection, IEC 60945 for marine, lightning protection, and heated options for icing climates. Confirm the station survives the worst case at the actual site.
7. Define output, telemetry, and power: Confirm the protocol (SDI-12, RS-485 Modbus, analog) matches your logger or SCADA, choose the telemetry uplink for the site's connectivity, and size solar and battery for the worst-case winter heating load, not the average.
8. Cost the full lifecycle: Purchase price plus installation, calibration interval and cost, spare-sensor strategy, and the cost of data gaps. A station that saves money upfront but drifts out of tolerance or freezes in winter loses far more in lost or corrupted data.

One dimension that buyers consistently underweight is serviceability and long-term support: local calibration and repair service, spare-sensor availability years after purchase, firmware and configuration tool maintenance, and documented calibration traceability. A station lives for a decade or more in the field, and the response time to a failed sensor in year seven matters more than a small accuracy edge at purchase. Established meteorological instrument makers such as Vaisala, Campbell Scientific, OTT HydroMet (Lufft), Gill Instruments, Thies Clima, and Kipp & Zonen maintain calibration laboratories and long support horizons, which makes them reliable anchors for network-scale projects, while regional suppliers can serve cost-sensitive industrial sites where reference-grade traceability is not required.