Whereas the stations established early in the WMO GAW program were situated either on mountain tops, such as, Mauna Loa or Tenerife, or along windward coastlines, the current trend of establishing continental stations makes it very important to determine what the influence of local and regional sources might have on the proposed measurements. The pathway taken by an air parcel arriving at any particular place is typically called the air parcel trajectory. (See the figure for an example.) As part of a collaborative project between the Atmospheric Environment Service of Environment Canada and the Finnish Meteorological Institute to evaluate the suitability of the Pallas station for ambient CO2 measurements, air parcel back trajectories were calculated 4 times per day using the Lagrangian isobaric model of Olsen et al. (1978). This model uses the objectively analysed wind fields out of the Canadian Weather Forecast Model. The latitude, longitude and height of the air parcel are calculated backwards in time every 6 hours for a 5 day period.
While individual air parcel trajectories have been used with considerable success to determine the influence of nearby and distant source regions on the "character" of the air arriving at the station for specific measurement events (see the next section), summarising these to determine typical trajectory pathways for the specific location is more problematic. To this end we have borrowed from hydrology the concept of the airshed (analogous to a watershed) for the station. Unlike a watershed, the airshed is not established by a fixed boundary determined by topography. Rather the boundaries are flexible and change with the passing weather systems.
The predominant source regions for the air parcels arriving at the station are the mid and north Atlantic, central to northern Canada and the Arctic Ocean basin. The frequency of trajectories coming out of Russia (east and south-east) are much fewer and the distance travelled is much shorter than those coming out of the Atlantic (west) and Europe (south-west). This is not unexpected given the predominant weather patterns and westerly flow in the northern hemisphere. Secondly, the general pattern is similar for the all three time steps plotted. Examination of individual trajectories indicates that the very long 5 day trajectories (>5000 km) may originate in the high Canadian Arctic, may extend well into the North American continent or pass through central Canada but originate in the Canadian Arctic. Even though the 5 day end points may originate in relatively clean source regions, the curvature of the trajectories often bring the air parcels through Europe and approach the Pallas area from the south.
The winter trajectories travel the longest distance and are predominantly from the north-west, slow down in spring and come predominantly westerly, swing to the south in the summer where they remain in the autumn but increase in speed (longer distances). Generally, the winter and spring air will be the cleanest, whereas the summer and autumn will be the most contaminated.
The final step in determining the airshed for the station involves calculating the mean distance (Mt) travelled and its standard deviation (sigmat) in each 1 degree sector for all the trajectory data. The 90% and 10% distances were then calculated from the Mt ± 1.273st which contains 80% of the trajectory distances and plotted on this figure. Ten percent of the trajectories extend well into the North American continent or across the Arctic Ocean to the Bering Sea area. The majority of the south-west to northerly trajectories originate in relatively clean source regions but the westerly ones may be influenced by the outflow from the eastern United States industrial area. The direction of approach to the station of the 1 day trajectory must also be taken into account.
In a earlier study (Rummukainen et al. 1996)1, a three-dimensional trajectory climatology was used for assessing tropospheric transport patterns with a data set of 292 trajectories for 1992-1993. Regions where air masses had spent their time during transport were identified with a form of seasonal transit time analysis. For this analysis, the horizontal source area distribution for air-mass advection was divided into eight regions (not sectors). It was found, that the Arctic was the most important source region at the 950 hPa level. In the summer, it contributed 40-45 % to the transport. With increasing altitude, however, the contributions of the European and Atlantic regions become comparable.