Clamp Calibration Notes - extracted from:

Renfrew, I. A. and P. S. Anderson, 2002: The surface climatology of

an ordinary katabatic wind regime in Coats Land, Antarctica,

Tellus, 54A, 463-484.

Halley is a meteorological observing station making 3-hourly synoptic observations and daily radiosonde launches. Here we make use of the operational cloud observations, plus hourly pressure and wind data. The winds are measured at 4 m by a cup-vane anenometer, which is checked daily for rime build up. The wind data tabulated here are corrected to 3 m using a neutral logarithmic wind profile, and assuming a roughness length of 1×10-4 m (King and Anderson, 1994). In addition, a number of research instruments have been sited at Halley for use in boundary-layer meteorology experiments. Here we use temperature and humidity measurements at 2 and 4 m, from Vaisala HMP35A sensors housed in R. M. Young force-ventilated radiation shields. King and Anderson (1994) discuss instrumentation at Halley in more detail.

At the Coats Land sites an AWS records hourly station pressure; air temperature and humidity at two heights (nominally 1 and 2.5 m); and winds at one height (nominally 3 m). The temperature and humidity data are from Vaisala HMP35D sensors housed in a modified R. M. Young naturally-ventilated shield, where an internal solar-powered fan assists ventilation during periods of strong insolation. The HMP35 range of instruments employed at Halley and the AWS sites use 1/30 DIN platinum resistance thermometers and solid state capacitive humidity sensors. The wind data are from an R. M. Young propeller-vane anenometer. It was found that this design is less susceptible to becoming frozen into position. The AWS pressure and wind data are laboratory calibrated prior to deployment and the pressure sensors are checked annually on site against a Vaisala PA11 digital barometer, which itself is calibrated annually by the UK Met Office calibration laboratory. The temperature data are calibrated in two stages: firstly using a series of precision resistors, which establish a linear correction to the temperatures registered by the AWS, and secondly by making a uniform adjustment of the lower (1 m) temperatures at each AWS, using an offset determined by averaging all the data with very high wind speeds (greater than 15 m s-1) and assuming the atmosphere is well-mixed by mechanical turbulence under such conditions. The second calibration step involves corrections of only ~0.1 oC and is carried out to remove discontinuities in the time series and to obtain more reliable surface heat flux estimates. The temperature calibrations are implemented separately for each AWS and each year. The relative humidity data are post-calibrated for each sensor and each year following the method of Anderson (1994). This makes use of the fact that over a snow-covered surface the atmosphere is saturated with respect to (w.r.t.) ice much of the time (e.g. King and Anderson, 1999). The capacitive sensor acts as a nucleation site when the atmosphere is supersaturated, and thus a well-defined relative humidity versus temperature upper bound can be obtained by curve-fitting to the data. This method is extremely robust and circumvents the on-site calibration problems inherent in using capacitive humidity devices to give reliable sub-saturated humidity measurements.

Surface sensible and latent heat fluxes have been calculated using the temperature and humidity measurements at two heights and following a profile bulk-flux methodology. In this case we use a surface roughness length of 1x10-4 m, and use the limited-value flux-profile relations of King et al. (1996). Although it is robust, the profile method is extremely sensitive to small changes in temperature or humidity difference, which means unphysically large fluxes can be calculated (e.g. Stearns and Weidner, 1993).To try and eliminate these erroneous fluxes the temperature and humidity difference data are neglected if t1 - t0 > 10 oC on the stable side, and t1 - t0 < -0.5 oC on the unstable side. The unstable side threshold is smaller as grossly unstable conditions do not occur at Halley or further inland (King and Anderson, 1994). To give an idea of the uncertainty in the calculated fluxes a sensitivity study was carried out, with the following varied: the threshold for 'bad' data on the unstable side was changed to -0.2 oC and -1.0 oC, the roughness length was changed to 0.5x10-4 and 1.2x10-4 m (King and Anderson, 1994; King et al., 1996), and the instrument heights were changed by -0.5 and +1.0 m (the instrument heights are only known exactly when the site is visited). The accumulated (i.e. worst case) differences for these changes are used to define a range of uncertainty, as shown in Table 5, for each season and site. The ranges are relatively large in the summer, and all year at C1 (where earlier instrumentation was used). To give a quick impression of the uncertainty, where the magnitude of the mean flux is larger than the range the mean is printed in bold. In other words, the bold estimates are more reliable. To concentrate just on these values: Qs is always negative, that is a flux of heat into the snow surface, with the largest magnitude at C2, the windiest site, followed by C4, C3 and then Halley. The mean latent heat fluxes are an order of magnitude smaller than the sensible heat fluxes and have greater uncertainty. To focus on Ql in JJA, there appears to be a change of sign from positive at C3 (e.g. due to sublimation) to negative at C2 and on the ice shelf (e.g. due to freezing).

The perturbation pressure (p') is the deviation from a mean pressure for that month at that site. Thus large differences in p' between the sites indicate that the pressure distribution is anomalous from the mean pressure distribution. Perturbations from a monthly mean are used to nullify the seasonal shifts of mass observed over the Antarctic, a by product of the polar location and elevation of the continent (Parish and Bromwich, 1998). It was found that interpreted carefully, using p' was a useful indicator of the mesoscale pressure gradient, better than calculating a mean sea-level pressure where the problems of reducing to sea level without a temperature profile are well known (e.g. King and Turner, 1997). However one should bear in mind that p' does not tell us anything about the 'background' pressure distribution that will exist due to the differential heating between the continent and the surrounding ocean for example.