Comparison of Two Closely Located Meteorological Measurement Sites and Consequences for Their Areal Representativity

We compare meteorological data collected at the Met Office Research Unit, Cardington, UK with similar data from a temporary meteorological station located approximately 8.5 km away. Data were examined for a period of 10 months to ascertain differences in mean quantities, and in heat and radiation budgets, at different heights between the two locations, one of which is located in a wide shallow valley, the other on a plateau at the valley edge. Results reveal that screen-level variables at the two sites show the greatest differences in mean quantities, for most conditions, but that at 50 m the differences are negligible, indicating that temperatures had become aggregated and homogeneous at that height. For flux measurements between 10 and 50 m, however, significant differences were observed at certain times of the year, which appear to be related to local vegetation and soil conditions, and these are discussed. The study also presents data for stable conditions that show that temperature differences at screen level are again the most significant difference between the two sites. On average these were not large, but on occasions discrepancies at low levels (up to approximately 10 m) as large as 5°C were observed. It is thought that these greater differences during stable conditions may be caused by cold air pooling at the valley site, despite the very shallow orography there. Data from the two sites have been compared with forecasts from two Met Office mesoscale models. Results show that, during daytime hours, model-predicted values lie outside the range of values observed at the two sites, indicating a possible model bias. However, the opposite was true for most nighttime hours, during which model values fell within the range observed at the two sites, indicating that at night the model predictions are representative of the region. In this case, however, comparison of the model prediction with either one of the available observations could lead to the false conclusion that the model temperatures were either too high or too low, depending on which observational site was used for the comparison.