Siting, exposure and calibration
Automatic Weather Stations (AWS) are becoming increasingly popular in many applications owing to their becoming more affordable, reliable, having improved data logging capabilities and through the growth in personal computing bringing sophisticated data manipulation and archiving within the reach of all.
This guide is of relevance to all who manage AWS, but it is primarily aimed at the observer who cannot be tied to a station at 0900 UT, and brings the opportunity to contribute data recorded between the standard climatological times. AWS have the potential to add enormously to the historical weather data in the UK by recording in fine detail meteorological parameters from a wide range and great number of sites. They also have the potential to set meteorology back many decades, however, if poorly sited and poorly calibrated.
These two factors are well recognised as fundamental when dealing with traditional instruments but AWS bring new challenges. Although the principles of traditional instruments apply equally to AWS, often the practicalities brought on by the design of equipment mean that compromises are unavoidable and calibration of sensors more difficult. Too often the precision of AWS sensors is mistaken for apparent accuracy.
With careful consideration of exposure prior to installation, however, and initial and ongoing calibration of sensors, many models of amateur quality AWS can give equipment many times their price a run for their money. Without this application, though, data cannot be relied upon – poor exposure often invalidates it and lack of calibration casts doubt on its accuracy.
Buying an AWS and getting it to work is just the first step on the road to producing climatological and meteorological data which will hopefully be not only interesting in itself but also a future asset to the meteorological community. Siting, exposure and calibration are the essential next stages.
Siting and exposure
The issues of siting and exposure are concerned with ensuring that everyone is recording the same thing, the same way, with the same limitations to allow intercomparison of data. Siting refers to the physical characteristics of the recording location while exposure refers to the deployment of the instruments in relation to the elements. The standard siting and exposures for the different types of recording instrument can be found in The Observer’s Handbook, published by The Met Office. If you can meet the criteria within for your own equipment, congratulations! For those unable to meet the ideal, the following details should allow a good compromise siting.
Temperature and relative humidity
Air temperature and humidity should be measured at between 1 and 1.5 metres above the ground in a shaded and ventilated environment; such an environment is usually provided by a white-painted Stevenson Screen, which acts to keep the temperature sensor out of the warming effects of direct or indirect solar radiation, infrared radiation from all objects in sight and rain, but permits free air flow around the sensor. The problem with relying on the AWS screen is that there are varying patterns, each with their own exposure, and with widely varying responses to solar radiation, wind flow or rainfall.
The Stevenson Screen has been the standard exposure in the UK and many other countries for more than 120 years and, while not perfect in every respect, does provide a benchmark reference by which readings can be compared, not only between stations, but over a long period of climatological record. It is often impossible with makes of AWS which have the raingauge attached to the sensor screen to mount it other than in the open, without extensive modification.
Whichever screen is used, ideally it should be fixed in an open place with good airflow on a level surface above short grass at the standard height. It should be no nearer than 30 metres from extensive concrete, aggregate or a road surface. Measure the height of surrounding objects above screen level: the distance from the screen to these objects should be at least two times these heights. These are the ideals but what of the amateur siting his equipment in a suburban garden? There may be trees in the garden and if it is possible to move the screen far enough away from them it approaches too closely to the walls of the house, garden fence or hedging, or even off the lawn and into the herbaceous border.
This is a typical problem and no hard and fast rules can be laid down, but an assessment of the various obstructions is necessary to enable a decision to be made over which object will be least detrimental to approach too closely. The siting of screens and sensors on the eaves of the house is totally unacceptable as it will give largely meaningless results, except in windy weather; as is fixing it to the post of a wooden-paneled garden fence, although all of these transgressions have been observed.
Grass minimum and soil temperature
Grass minimum temperatures should be measured above a grass surface, not soil or other ground cover, unless this is more representative of the locality. Sensors used to measure grass minimum temperature should be of an appropriate design with minimum thermal bulk and good weather-resistance, as these sensors will experience a very wide range of temperatures over a full year and will also be exposed to rainfall or snowfall; not many battery-powered wireless sensors will operate reliably for long under such conditions.
Soil temperatures (down to 20-25cm depth) should be measured in an area of open soil representative of that naturally occurring at the site. Earth temperatures (at 30cm or below) should be measured beneath a ground cover of short grass.
Ideally, the rim of the raingauge should be sited at 30cm above ground level. Precipitation catch varies with height due to turbulent eddies and there is considerable reduction in catch as wind speed increases above the ground. Many raingauges are integral with the temperature screen and in this case you will have no option but to site it at 1.25 metres above ground level due to the positioning of the screen. In deciding the best position you will have to take into account the requirements of both temperature/humidity exposure and precipitation, possibly performing several compromises.
You should consider using a manual copper raingauge for accurate rainfall measurements in conjunction with the AWS. Daily totals can be adjusted where it cannot be emptied at 0900UTC by establishing the proportion of the total which has fallen either side of 0900UTC from the AWS data and allocating the manual gauge catch accordingly. Wherever the gauge is mounted, make sure it is absolutely level. ‘About right’ isn’t good enough. A slight error in leveling will cause significant recording errors.
Siting your anemometer and wind value
The anemometer and wind vane should be capable of being mounted separately from the screen and raingauge. Some professional AWS do have the anemometer/wind vane/screen/raingauge mounted on the same mast assembly around the same height, but it is to be noted that it is intended that this equipment will be used at a well exposed site where correction can be made. Within the average amateur site any wind readings from such an assembly are likely to be inaccurate and it cannot be recommended.
The standard height for meteorological wind measurements is ten metres above ground level, but some agricultural or environmental applications may require the use of lower heights. In a suburban site, even were it physically possible to accommodate a ten+ metre high mast with associated guying, it would likely raise a storm of protest from neighbours demanding a Planning Application be processed to which they would object. Reason must prevail and it must be left to individual judgement as to what will be acceptable to surrounding residents. Mounting to house eaves and chimneys is, however, to be avoided as the building will interfere detrimentally with the readings, notwithstanding the increase in height.
As with other instruments, it is very important to ensure the anemometer and wind vane are truly level once fitted. This applies equally to ultrasonic anemometers, notwithstanding their lack of moving parts.
Sunshine, solar and UV radiation
These sensors should be positioned so that they are not overshadowed by surrounding objects at any time of the day, at any season of the year. They should be positioned absolutely level, using either the spirit bubble provided with some makes of sensor or a separate spirit-level. In practice, in a suburban location with surrounding trees and houses, it will be difficult to avoid overshadowing, especially during the winter months when the sun’s declination is low, unless the sensor is mounted high up. This solution may itself bring problems of levelling and maintenance.
Calibration is the practice of checking an instrument against another of known accuracy, or with a process of known properties. Calibration is as essential for the amateur as for the professional meteorologist but is one of the most commonly neglected tasks. For the amateur, calibration can be difficult to achieve without access to standard, calibrated check-instruments and an approach in accuracy to that achieved in the professional environment can be difficult to achieve without incurring considerable costs in doing so.
Within the constraints of the amateur scene, however, much worthwhile calibration can be achieved at reasonable costs in time and effort; such calibration checks, even if only performed annually, serve to eliminate gross errors and can alert the observer as soon as possible to a faulty or unreliable sensor. Calibration is only possible up to a point. Only very expensive instruments are accurate to one decimal place although many are precise to this degree. The handbook with the AWS should describe the resolution and nominal accuracy of each sensor, or this may be available from the manufacturer.
Where this information is not available the sensor is likely to be not worth calibrating; indeed, the equipment is not likely to be of any serious use. As for detailed calibration, barometers are so easy to check (at least to within 1 hPa or so) that a monthly calibration with a METAR or BBC Weather Report should be done. All sensors should be subject to a detailed calibration at least annually. Of course, all sensors should be calibrated as soon as possible after purchase.
If a check-instrument of known calibration is available it should be compared with the AWS barometer and any correction necessary made. Pressure varies little over a wide area during periods of anticyclonic activity and light winds. If a check-barometer is not available calibration can be done by comparison during one of these periods with a METAR or BBC Weather report, or a neighbouring station of known calibration.
Indeed, this check can be done at most times except during severe cyclonic activity or windy weather and provided the station being used for the comparison is no more than 10 km away there is likely to be little difference in pressure, certainly less than the nominal accuracy of the instrument. Mean Sea Level Pressure (MSLP) should always be used. If comparing with a check-instrument you will need to know the height of the station where the calibration of this instrument took place and your own station height (available from the contours of Ordnance Survey 1:25,000 scale sheets to a sufficient accuracy for our purposes).
Pressure increases/decreases by roughly 1 hPa per 10 m decrease/increase in height respectively, although the actual value of the correction to mean sea level will vary with both air temperature and barometric pressure. Depending on any height differential between your station and the calibrating station you will need to apply a correction to the indicated pressure on the check-instrument to obtain the MSLP.
The check-instrument should be mounted at the same height as the temperature sensor to be checked. If the sensor is in a Stevenson Screen the check-instrument should also be positioned within it, if not and it is within its own AWS screen, then the check-sensor should be positioned within another adjacent screen.
To eliminate differences in readings caused by differing screen exposures, calibration can be performed on cloudy days with good airflow, although ideally it should be carried out under the same conditions at night. Where it is impossible to obtain a screen in which to house the check-sensor, calibration is still possible on a heavily overcast, breezy but preferably dry day. If a check-thermometer is not available one can be prepared using a traditional thermometer or a digital thermometer with temperature probe. If check-instruments cannot be obtained other methods of rough calibration can be used.
If a nearby station has calibrated instruments arrange to compare data from a cloudy, windy night. A correction must be applied for differing elevations of ~0.5° C per 100 metres increase in height, but if stations are truly comparable there is not likely to be a great height differential.
This is a difficult parameter to check without another instrument of known calibration as a reference. If one is available it should be positioned in accordance with the notes above for temperature and a similar calibration check performed. A good check – instrument to obtain, and one which is reasonably priced, is a whirling hygrometer. Without another check-instrument a test for serious faults can be carried out. This should be performed on a day of fog at which time the indicated RH should approach 100% when the stated accuracy has been allowed for.
It should be noted that most sensors do not read RH at all well above 90% so this check is of only limited value, to identify a grossly faulty sensor. A better check can be made around the middle of a breezy day by comparing readings with a neighbouring calibrated station, METAR or BBC Weather report. Where dew point is given instead of RH this relies on the same sensor and, assuming the temperature sensor is well calibrated, can be used to calibrate the humidity sensor instead. Dew point, being representative of the airmass at this time of day, should be similar throughout the locality. Where dew point is given, humidity tables or a calculator will be needed to convert this to RH to check if the reading is within the stated accuracy.
Most AWS do not allow a direct correction of dew point, the correction must be applied to RH which then corrects the dew point. Checks should be done over a period of time as humidity sensors are one of the least accurate and conclusions should not be drawn from one suspect reading. As with temperature, it is best to perform intercomparison between instruments when the humidity is not changing rapidly, as different instruments may have very different response and lag characteristics which could obscure conclusions drawn on calibration accuracy.
The notes below refer to the procedures to be adopted with tipping bucket raingauges (TBRs), which are most commonly used in amateur level and professional AWS systems. There are two ways of adjusting an AWS raingauge: by comparison of catch with a national standard manual raingauge or by pouring a measured amount of water into the gauge and comparing the recorded amount (a true calibration).
Of the two methods, comparing with a standard gauge is often more useful for the following reasons: • A standard gauge (in a standard exposure) is by definition correct • There will always be a difference in catch even with a correctly adjusted AWS raingauge if it is sited above ground level due to the necessities of equipment design. Comparison against a standard raingauge provides for an adjustment of some of this error • Even when the mechanism of an AWS raingauge is correctly calibrated against water poured into it, it will not necessarily record exactly the same precipitation as a standard raingauge, siting and exposure being identical, due to the fundamental differences in design affecting precipitation catch. This will mean that the rainfall data from the AWS will need to be adjusted by a correction factor to obtain close agreement with a standard gauge and thus provide data intercomparable with other sites. It may be far simpler for these applications to simply adjust the AWS raingauge to agree closely with a standard raingauge catch
The standard raingauge in the UK is a 5 inch diameter copper cylinder, part-sunk into the ground, with a deep funnel (to prevent out-splash in heavy rainfall and to minimise the loss of precipitation in snow or hail), with its rim mounted at 30 cm above short grass. To adjust against a standard raingauge it must be sited close to the AWS raingauge, taking care that it is not affected by the shelter of the AWS itself. Comparison should take place over a period of time against rainfall events on several occasions. It is preferable to use daily totals for the comparisons.
Showery days may show greater variations, particularly if one of the gauges has a shallow funnel which may allow heavy rainfall to splash in or out of the gauge orifice, as will windy days if the AWS raingauge is mounted above ground level. It should be noted that tipping bucket raingauges particularly are prone to increasing inaccuracy (undercatch) as the intensity of rainfall increases, unless they are calibrated to be accurate at that rate of rainfall. Intense rainfall events should not be used in the comparison.
AWS raingauges which only record in 1 mm increments have inherent drawbacks which can display themselves as apparent calibration errors. These are largely overcome when using the method of measured amount described on the Royal Meteorological Society’s website, details of which will follow.
Wind speed and direction
The difficulty in calibrating an anemometer and wind vane is that exposure is such a significant factor that typical amateur stations within the same district will see considerable variations in readings. The recorded wind is only likely to be representative of that particular station at the position of the instruments. This is likely to be more so with respect to wind speed and a reasonable correlation of wind direction should be seen.
The most accurate calibration will be achieved by the use of a check-instrument mounted alongside the sensors to be tested. Starting speeds can be a factor denoting differences in instrument performance rather than calibration issues, so it is better to calibrate on a breezy day. If the exposure of the site is good, an acceptable alternative to fixing the check-instrument alongside the AWS sensors would be to have it at ground level and apply a correction for the difference in height (adding approximately 10% to the speed of the lower instrument per 3 metres difference in height).
The disadvantage of having the check-instrument at ground level is that applying a correction for wind direction is very difficult and best not attempted. The calibration check should continue for several hours to be effective. Where a check-instrument is not available a rough check on accuracy can be undertaken by comparing the recorded gust speeds on a windy day with the effects described in the Beaufort Scale. Calibration of the wind vane can be done visually by ensuring hat it is pointing in the direction indicated.
It is important to note that wind directions are given from true north, not magnetic north, and when setting up the vane it is usually necessary to reference the mounting direction to true north. Calibration is usually only useful to correct incorrect alignment on installation.
Intricacies for sunshine
These sensors are expensive and complex and it is unlikely that the amateur will have access to a check – instrument on site. The spot readings being obtained from the solar/UV sensors of a neighbouring station of known calibration on a cloudless day, however, can be compared and any necessary adjustments made from this data.
Electronic sunshine recorders usually come with a trim adjustment to enable the setting of a recording threshold. This should be adjusted near local noon on a day with (for example) thick cirrostratus which is just allowing the formation of discernable shadows so the instrument just registers sunshine. Once ‘trimmed’ the adjustment should be left alone unless it appears to have drifted seriously out of adjustment. Frequent repeated adjustments during the course of a year should not be necessary and will adversely affect the quality of the record obtained.
All of these sensors should be subject to a basic check which ensures that their output is zero after dark. More information about data compilation and storage, plus other resources for meteorologists is available in Andrew Overton’s full guide, which you can see by logging on to: http://www.rmets.org/pdf/guidelines/aws-guide.pdf
Andrew Overton is a railway signalman who pursues meteorology as a hobby, operating a climate observing site in Doncaster, South Yorkshire, using an automatic weather station (AWS) and manual rain gauge, submitting data to the Climatological Observers Link. A member of the Royal Meteorological Society’s Special Interest Group for the History of Meteorology & Physical Oceanography, and the Meteorological Observing Systems Group, it was while serving on the committee of the latter group that he became aware of the need to encourage individuals using AWS to take greater care to obtain accurate and comparable data. The easy availability and lowering price of this equipment has seen increasing ownership, with data acquisition often being applied to commercial tasks. Existing guidelines were not easily available to the non-specialist, were difficult to understand and often not easily applied to AWS installations. His authorship of a web-based guide was undertaken to address this issue.
Published: 01st Mar 2012 in AWE International