Review of Air Temperature Measurement: Passive Versus Active Shields, Error with Humidity Probes, and Significance of Averaging Interval.

Air temperature measurements on automated weather stations are made with electronic sensors that typically provide high accuracy, but accurate measurement of air temperature requires more than an accurate sensor. The sensor must be in thermal equilibrium with air, which requires ventilation and shielding from solar radiant heating. Two main types of radiation shields have been used: passive (static) and active (fan-aspirated). Passive shields are typically louvered enclosures that rely on natural aspiration from wind to equilibrate the sensor with air. Passive shields do not require power, but they are less accurate due to solar heating in low wind. Field measurements indicate air temperature errors for passive shields can be 2 C in low wind over grass, and as high as 4 C when used in low wind over snow. Active radiation shields maintain better equilibrium with air through fan-driven aspiration of the sensor, and are significantly more accurate under high solar loads in low wind. However, active shields require power to operate the fan. As a result, passive shields are far more common than active shields. The most common configuration for air temperature measurement is a humidity probe mounted in a passive shield, where temperature is determined from the humidity probe. Due to large thermal mass of humidity probes, errors with this configuration can be 0.5 C greater than errors with a stand-alone temperature sensor in a passive shield. Measurement errors from shield selection and sensor type are not the only issue. The method used for averaging air temperature measurements can have a major impact on data interpretation. There isn’t an accepted standard for processing temperature data. Instantaneous measurements, block averages, running averages, and different measurement time frames have all been used, complicating comparisons among data sets and comparison to historical records.