Continuity of climatological observations with automation - temperature and precipitation amounts from AWOS (Automated Weather Observing System)

Abstract

Recent automation of meteorological observations affects homogeneity of the long-term climatological records, which are used to study climate change and variability. In order to avoid false conclusions regarding apparent climate trends, these records must be adjusted to account for biases caused by new instrumentation, computerized processing algorithms and relocation of the observing sites. This study of the effects of automation on two primary climatological elements, temperature and precipitation amounts, was conducted at five stations situated in various climatological regimes across Canada, where concurrent Automated Weather Observing System (AWOS) and manual observations were collected over the period of one year. The authors attempted to assess observations at higher temporal scales: average daytime and night-time temperature; daily maximum and minimum temperature; daily precipitation - in addition to the usual annual, seasonal or monthly precipitation. Individual hourly or daily observations of temperature were grouped according to meteorological conditions that either maximize or minimize instrumental and site differences, e.g., sky cover and wind speed. Similar electronic temperature sensors were used by both the observer and AWOS, which resulted in a rather small instrumental bias: AWOS reported temperatures that were warmer by up to 0.2°C. The sitting bias, caused by AWOS typically being installed in the middle of an airfield, was often much more pronounced due to the greater radiative cooling: on average AWOS reported temperature minima that were colder by up to 1.3°C. Differences between gauges, especially in resolution and height of the orifice above the ground, were identified as the main source of observed biases. It was not possible to quantify the sitting portion of the overall bias. Precipitation was categorized according to the amounts reported by AWOS. In the category of light daily amounts up to 5 mm d-1, no consistent reliable relationship between AWOS and the observer could be established, while in the moderate to heavy category of amounts higher than 5 mm d-1, AWOS underestimates precipitation by up to 13%. Cases, when either the observer or AWOS reported some precipitation, while the other reported null, were also examined in detail. Over time periods of one month or longer, undercatch by the AWOS automated weighing gauge, as compared to the Type B gauge for rain and the Nipher gauge for snow, is quite severe, on the order of tens of percentage points. This study emphasizes that the availability of at least one or two years of concurrent conventional and automated observations is crucial to the development of adjustment factors, especially for observations of sub-diurnal resolution. With the growing demand for good quality sub- and diurnal resolution data for construction of scenarios of impacts of global climate change on humans and the environment, it is expected that more research on adjusting high temporal resolution data will be required and conducted in the future.