Abstract/Summary

The World Meteorological Organization (WMO), National Oceanic and Atmospheric Administration, and other organizations provide guidance on expected response time for sensors used to measure air temperatures intended for meteorological applications. Quantified as the sensor time constant (the time it takes for a sensor to reflect some percentage of a step change), recommendations differ somewhat depending on the organization. For instance, the WMO specifies the 63% time constant should be ≤20 s, although, crucially, the organization does not state the air flow velocity at which this time constant should be achieved. Recent independent tests at two laboratory facilities (initially the University of Reading, United Kingdom, and subsequently at Campbell Scientific, Logan, Utah, United States) were undertaken to determine time constants of a range of commercially available platinum resistance thermometer sensors. Results showed that many sensors fell far short of the WMO specification at airflow rates typical of naturally ventilated thermometer screens or radiation shields (1 m·s−1 or lower). In contrast, a recently released platinum resistance thermometer sensor from Campbell Scientific was shown to meet both specifications, even at airflow rates within a laboratory wind tunnel as low as 0.2 m·s−1, which is more typical of naturally ventilated thermometer screens or radiation shields. Across multiple sensors and repeated test runs, the new sensor's 63% response time averaged 10.7 s (standard deviation 0.5 s) at an airflow of 1 m·s−1 and 17.1 s (standard deviation 0.9 s) at 0.2 m·s−1. To our knowledge, this is the first commercially available sensor to attain this WMO specification. However, using or switching to faster-response sensors has important implications for long-term data records, the measurement of extreme temperatures (specifically daily maximum and minimum data), and intersite comparisons. This is compounded by seemingly conflicting recommendations from the WMO regarding sensor time constant versus data processing methods.