Ž .
Ž .
al. 1988 and Dore et al. 1992a observed and quantified the enhancement. These experiments lead to the adoption of a simple scheme to predict and map orographic
Ž .
enhancement nationally Dore et al., 1992b; RGAR, 1990 . The central assumption was that the concentration of ions in the scavenged water is on average twice that in the
Ž unperturbed rain. More recent experiments have observed single rain events Inglis et
. Ž
al., 1995a,b and monitored rain and orographic cloud over long time scales Fowler et .
al., 1995 . This work has confirmed the dominance of the scavenging process in determining deposition to upland terrain and verified the simple mapping scheme
Ž .
RGAR, 1997 . However, the parameterisation only predicts average deposition to large grid squares
Ž .
of terrain 20 = 20 km . Sub-grid scale variation is important in terms of damage to soils and ecosystems which typically have characteristics which vary on length scales which
are considerably smaller than 20 km. Ultimately, the aim is to improve the resolution of wet deposition predictions to length scales close to 1 km. They can then be directly
compared to critical load calculations that are carried out on a length scale similar to that
Ž .
of the current soil type and land use maps CLAG, 1994 . Ž
. Raper and Lee 1996 made weekly collections of rain from a network of 10 sites in
the south Pennines for 1 year. Sub-grid scale variations were observed in the average concentrations of calcium, hydrogen, and sulphate.
Here, we report results from an experiment specifically designed to address the mechanisms that cause sub-grid scale variation in wet deposition to upland terrain. The
terrain and site separation are selected to allow the detailed evolution of wet deposition processes to be observed as the boundary layer encounters an extended region of
elevated terrain. The monitoring equipment deployed is simple, but full consideration of associated uncertainties is made and carried through to comparison to computer simula-
tions allowing firm conclusions to be reached.
2. The field experiment: Saddleworth Moor, 1993
The experiment was carried out in October and November 1993. The location of Saddleworth Moor in the south Pennines is shown in Fig. 1.
The bulk rain collectors are simple funnels mounted in boxes 1.5 m above the Ž
. surface, following the original design of May 1961 . The cloud collectors also support
Ž .
an array of vertically mounted polytetrofloromethane PTFE strings following the Ž
. Ž
. design of Dollard et al. 1983 and May 1961 . Cloud droplets impact onto the strings
and run down to be collected by the funnel in the same manner as rain. Samples are Ž
. refrigerated until analysis in duplicate for all the major ions using an ion chromatogra-
phy technique. Fig. 2 shows the configuration of the collector sites and the distribution of monitoring equipment.
The collection efficiency of a rain collector is expected to depend on wind speed. Ž
. However, data from a 2-year period 1996–1998 at site 9 indicates that the effect is
small. Weekly rain amount as measured by a tipping bucket rain gauge mounted 0.2 m above the surface was compared to rain amounts from a collector of the design used
Fig. 1. Map of northern England showing the position of Saddleworth Moor, regional boundaries, major cities and the radio-sonde launch site at Aughton.
here. Average wind speeds at 0.2 m are less than 40 of that at 1.4 m, yet, the total rain amount agreed to within 2. Despite this, on occasions when the wind speed is high or
the average drop size is low, the collection efficiency may be reduced. However, all moorland sites are at a similar altitude and were selected to be far from small surface
features on the moor. Differences in the wind speed between moorland sites are therefore expected to be small. Computer modelling of the surface speed during the two
events considered confirmed this. It follows that observed differences in deposition between these sites are unlikely to be a result of differences in collection efficiency.
The main function of the cloud collectors is to provide information on the ionic material dissolved in the orographic feeder cloud, i.e. the mass of ions dissolved per unit
volume of cloudy air. To do this, some considerations of the sampling characteristics of the cloud collectors are necessary. The simple design of the collectors means that they
sample both rain and cloud. The collection efficiency for rain was assessed over the same 2-year period, again at site 9, by isolating periods when rain but not cloud
occurred. During these periods, the volume sampled by the cloud collector averaged 97 of that sampled by an adjacent tipping bucket rain collector. According to this
result, all data presented as representing cloud in this work, have been arithmetically adjusted to remove the fraction of the sample resulting from rain. This process
introduces the dominant uncertainty in the calculation for cloud loading. The uncertainty depends on the observed fractions of rain and cloud.
For both of the rain events presented here, the collectors were washed with deionised water shortly before the onset of rain and so dry deposition to the collectors is not
Fig. 2. Contour map showing position of collector sites on Saddleworth Moor. B, Rain collector only; l, rain and cloud collector; , rain and cloud collector and AWS; and v, rain collector and AWS.
important. In addition, the volume of rain in each event was large enough for liquid left on the funnel and not collected to be insignificant.
3. Computer simulations of wet deposition