Locating outside sensors for optimum accuracy
Weather data is scientific data and, if you're serious about wanting to collect weather data of good quality, then your AWS system and procedures need to conform to the principles of making good scientific measurements. This isn't necessarily difficult, but does mean that you shouldn't just accept the readings of a new weather station at face value.
There are perhaps two guiding principles for collecting good data: firstly, ensuring that the quality of your readings is as good as possible, mainly by siting your weather sensors with due care; secondly, by being aware of the limitations of your data (and everyone else's data come to that).
Weather data of high precision and accuracy can only be collected if the sensors for the various weather variables are sited and installed with care. Mounting the sensors of a new AWS simply at the most convenient location at your observing site will almost certainly lead to significant errors in the data. The sensor locations really do need to be chosen with due scientific consideration for best results.
Siting sensors is a key topic in its own right, which is covered in some detail in a PDF document downloadable from the Royal Meteorological Society website and in the faq pages for the uk.sci.weather newsgroup (see the 'Instrument Siting' section under the 'Observing and Reporting' topic). A useful section on the UK Met Office website also has information on weather observations in the form of downloadable PDF files.
All serious commercially available AWS systems have a published specification which will show an estimated degree of accuracy for each sensor. (If this information is difficult to find for a particular make/model or only talks about display resolution and not accuracy, it's often an indication that the station is not suitable for anything other than 'gadget' use). It's worth taking a look at this specification table for your AWS to get an idea of how reliable your data is likely to be. It's also often possible to increase the accuracy of values of some parameters by calibrating the sensors - a topic which is covered elsewhere.
Remember also, that just because you can display or measure a value of, for example, rainfall to two decimal places, does not mean that it is accurate to that degree. It should be possible to set up most weather stations to measure weather parameters to 5-10% accuracy (or in the case of temperature-related values to ±1°C accuracy), but achieving higher accuracy, becomes increasingly difficult for nonprofessional observers.
Comparing data - an introduction
One tempting way to assess the quality of your observations is to compare your own AWS data with that from an established station in your own locality. But this is a process with many pitfalls. All too often, new AWS owners reach the conclusion that their instrumentation is faulty because the two sets of observations are significantly different, when in fact either the differences are genuine (the two locations having different microclimates) or the fault lies in the siting of the sensors, not the instrumentation itself. Again, see the comparing data page for more details.
Guidelines for optimal siting of sensors
The international standard height for measurements of air temperature is at 1250mm or about four feet above ground level. Temperature can vary by a surprisingly large amount with height above the ground under certain weather conditions, and it is essential that the 1250mm sensor height is used for comparability with official figures. Ground level or soil temperatures should be measured with separate sensors if required. The sensor must also be protected from both direct sunlight and rainfall if wildly erroneous temperature values are to be avoided, which is usually achieved by placing the sensor inside the familiar white louvred housing of the Stephenson Screen. The sensor should also be positioned away from any nearby potential sources of heat such as buildings and brick walls, in a position where free circulation of air can occur, and over a natural surface - grass is recommended (other surfaces such as concrete can cause significant error). As with all sensors, the more accurate your observations are required to be, the more care must be taken over exact siting of the sensor.
The humidity sensor is usually mounted alongside the temperature sensor in most commercial AWS systems and does not therefore have its own independent siting criteria. Where a separate sensor is used, the same guidelines as for temperature sensor siting are recommended. NB Humidity is not measurable to high accuracy (typically ± 3-5%) by standard electronic sensors and may not always quite reach 100% as a maximum reading because of sensor limitations.
Wind speed and direction
Wind speed and direction are dramatically affected near to ground level by all physical obstructions. Even in a flat, unobstructed location, wind speed is markedly reduced close to ground level simply by the frictional effect of the ground surface. For these reasons, the official sensor height for recording wind speed is 10m above ground level in a clear unobstructed location. In practice, it is often impossible to achieve anything like an optimum exposure of the wind sensor, unless you live in a very rural location . The recommendation is therefore simply to place the anemometer as high as possible given the local circumstances and to accept that the readings will almost certainly be significantly lower than would be measured by 'official' observations at the same location. The sensor readings will still be valuable as part of the continuing record of weather at that site, but will not be directly comparable with official records of for example maximum wind speed and gusts during a gale. Often the exposure will be worse from a certain direction and the obvious advice is to aim for maximum exposure of the anemometer to the commonest wind directions. Note that height above obstructions rather than height per se is the criterion. An anemometer placed just at roof level on a house will often misread because of wind swirling around the roof structure. The sensor should be placed as high above the roof structure as it safely and economically can be (preferably 2-3m above the highest point), to avoid potentially turbulent air below.
Rain rarely falls vertically, but is usually blown to a greater or lesser degree in the wind. For consistent and accurate measurements it's therefore important that the rain gauge is located in an open area, for example a large lawn, where nearby objects such as buildings, walls and trees won't deflect the entry of wind-blown rain into the gauge. Rain shadow effects can be surprisingly large and the standard recommendation is that the gauge be positioned at a distance corresponding to two to four times the height of any nearby obstruction. Where this is difficult to achieve, providing good exposure to the most common directions of wind/rainfall should be the priority. Gauges should be positioned in a flat area, away from any obstructions such as fences which might cause air turbulence and consequent non-uniform deposition of rain droplets. (NB For Davis Vantage Pro stations the rain gauge rim will automatically be at ~1400mm height if the temperature sensor is mounted at its official height of 1250mm.) Most rain gauges for automatic weather stations are of the swinging bucket type, which must be installed in an accurately horizontal plane for correct operation.
Since barometric pressure does not vary across a local area at uniform altitude, pressure is generally measured by a sensor inside the AWS console and not by an external sensor. Consequently, there are no major concerns about siting the pressure sensor, other than to be aware that its accuracy is only specified over a limited temperature range. Note that pressure reduces by about 1mb for every additional 32 feet of elevation and it is therefore essential to know the altitude accurately of the AWS base location. Pressure is also relatively easy to check and to calibrate.