Stephen Déry and Marc Stieglitz

Lamont-Doherty Earth Observatory, Columbia University, Palisades, New York, U.S.A.

ABSTRACT

In the cold Canadian environment, humidity measurements can be very difficult to conduct. In this brief communication, humidity observations taken by two different sensors at six remote Canadian Arctic locations are compared The observations collected by Vaisala HMP35CF sensors display a strong tendency toward the ice saturation point whereas dewcell instruments exhibit significantly lower values of relative humidity with respect to ice (RH_i). Humidity data collected by HMP35CF hygrometers are the refore unreliable since they are subject to persistent icing that lead them to record values of RH_i near 100%, irrespective of air temperatures. Thus, great care must be taken in utilizing humidity data recorded by HMP35CF sensors across the network of climate autostations in Canada.

The Canadian Arctic, subject to long, frigid winters, remains the scene of frequent adverse wintertime weather (Déry and Yau, 1999). In this harsh environment, humidity measurements become rather difficult to conduct. Unheated and unventilated hygrometers often become coated with ice rendering their measurements questionable, if not utterly useless (Makkonen, 1996). Canadian climatic records also contain significant gaps during the winter months where humidity measurements are simply not available or so small (in an absolute sense) that they are reported as zero (consult, for example, Environment Canada, 1993).

With a growing interest in evaluating the water and energy fluxes at high latitudes, there is a renewed demand for accurate humidity values in the low temperature environment of the Arctic. Of notable concern are latent heat fluxes associated with the continuous transfer of water between the atmosphere (in the vapour phase) to the snowpack (in the solid phase). Additional perturbations to these surface latent heat fluxes may be associated with blowing snow sublimation (Déry et al., 1998; Pomeroy and Essery, 1999). Since the surface latent heat flux depends critically on the amount of moisture present in the atmospheric boundary layer (ABL), precise measurements of this quantity become necessary.

In the past, some authors have examined in detail humidity measurements conducted in subfreezing conditions at a single site in Antarctica (Anderson, 1994; King and Anderson, 1999). Others have also examined several aspects of humidity measurements taken by radiosondes without considering, however, the accuracy of surface humidity observations (e.g., Elliot and Gaffen, 1991; Garand et al., 1992). The purpose of this brief communication, therefore, is to document the effectiveness of two types of hygrometers commonly employed in the cold Canadian environment.

Two principal types of hygrometers are utilized by the Meteorological Service of Canada (MSC) to record surface humidity measurements. At principal meteorological stations and within the Automated Weather Observation System (AWOS) network in Canada, the humidity instrument usually consists of a ``dewcell'' type hygrometer (see De Felice (1998) for a full description of the instruments discussed herein). In the MSC's network of 400 or so climate autostations, on the other hand, the Vaisala HMP35CF remains the Canadian standard instrument for humidity (B. Funk and P. Kociuba, personal communications, 2001).

As a cost-cutting measure, the network of attended climatological stations in Canada has steadily declined during the past two decades at the expense of automation. To make matters worse, the HMP35CF hygrometers at automatic stations are generally unheated, unventilated and uncalibrated to remove instrument biases. Therefore, the quality of the humidity measurements produced by these instruments may be unreliable. For instance, Anderson (1994) proposes a recalibration technique to improve the relative humidity (RH) measurements at subfreezing temperatures. In addition, Anderson (1996) and Makkonen (1996) entertain a discussion about the inadequacy of these hygrometers to record humidity values at all temperatures below freezing. The distribution of RH_i measurements displayed in Anderson's (1996) reply to the comments by Makkonen (1996) on this issue demonstrate, in fact, a strong tendency for the HMP35A hygrometer (an instrument very similar to the HMP35CF hygrometer) to record values of RH_i near the ice saturation point. By using a heated hygrometer, King and Anderson (1999) effectively conclude that the HMP35A sensors can both overestimate and underestimate humidity values due to icing on the instrument.

In their study of blowing snow and surface sublimation at Trail Valley Creek in the Northwest Territories of Canada, Déry and Yau (2001) encounter similar trends in their humidity data set. Values of RH_i, collected by Essery et al. (1999) using an HMP35CF hygrometer, display a mean RH_i of 97% during the winter of 1996/97. According to J. W. Pomeroy (personal communication, 2001), the meteorological instruments were often coated with ice since this site was not regularly maintained and was visited only a few times during the winter of 1996/97.

Considering that most Canadian climate autostations employ the HMP35CF sensors, there are concerns that their humidity measurements are susceptible to the same systematic errors. By using hourly observations of temperature and humidity recorded during the month of December 2000 at several remote Arctic sites, we now investigate whether this phenomenon is an isolated event or is more widespread across the Canadian network. Thus, data from six meteorological stations are examined in closer detail (Table I). Meteorological stations at Gillam, Hanbury River, and Robertson Lake make use of the HMP35CF hygrometers whereas Baker Lake, Inuvik and Norman Wells have sheltered and ventilated dewcell hygrometers. All six sites are located in or near the Canadian Arctic and, as demonstrated by their mean monthly temperatures for December 2000, endure frigid conditions during wintertime.

The relative frequency distribution of 744 RH_i measurements at these six locations is shown in Figure 1. The histograms clearly illustrate the strong tendency of the HMP35CF sensors to record values near the ice saturation point. By contrast, the three stations mounted with dewcell hygrometers exhibit significantly lower values of RH_i. Even at Gillam where the HMP35CF sensor is ventilated, the mean RH_i is considerably higher than at the three sites mounted with dewcell hygrometers. This strongly suggests that the HMP35CF hygrometers remain deficient at subfreezing temperatures due to icing that leads them to record values of RH_i at the ice saturation point.

Figure 2 reveals that the HMP35CF hygrometers tend toward 100% RH_i irrespective of the air temperature. A slightly curved profile in the approximate maximum RH_i at very cold temperatures arises due to calibration errors of the humidity sensors (Anderson, 1994). On the other hand, humidity data collected by the dewcell instruments exhibit less probability of reaching the ice saturation point as the air temperature rises. This behaviour is expected since supersaturation with respect to ice is more readily achieved as air temperatures decrease.

With its spacious, sparsely populated territories, Canada cannot escape the automation of many of its remote weather stations. This automation, however, does come at a price. Unattended instruments can easily become ineffective due to icing. Humidity observations recorded at three Canadian Arctic locations by HMP35CF sensors during the month of December 2000 clearly show a propensity toward the ice saturation point (irrespective of the air temperature), indicative of persistent icing on the instruments. These errors may not lead to significant problems for those interested in the absolute values of humidity which remain extremely low in the cold Canadian environment. However, small differences in the forcing data for relative humidity can lead to large variations in the evaluation of latent heat fluxes associated with surface and blowing snow sublimation (Déry and Yau, 2001). Consequently, the sublimation rates reported by Essery et al. (1999) and Déry and Yau (2001) remain somewhat inconclusive at this point in time. Thus, further investigation of the sublimation process is required to determine its importance in the surface mass balance of Canadian snowpacks.

Based on the preliminary evidence presented in this note, we recommend that users of meteorological data collected within the network of Canadian climate autostations employ great care in interpreting and applying their humidity measurements. A more thorough investigation of the deficient humidity measurements at cold temperatures and a complete list of affected stations would provide useful means to avoid this serious problem in the Canadian climate archive and mechanisms to improve the collection of future humidity observations in the cold Canadian environment.

We are indebted to Mrs. Monique Lapalme (MSC, Edmonton) who kindly provided the data used in this study. Dr. John Pomeroy (University of Wales Aberystwyth), Mr. Pete Kociuba (MSC, Edmonton), Mr. Gary Rink (MSC, Downsview), Mr. Barry Funk and Dr. John Hanesiak (MSC, Winnipeg) are sincerely thanked for their generous comments and insights into this work.

Anderson, P. S.: 1994, `A Method for Rescaling Humidity Sensors at Temperatures Well Below Freezing', J. Atmos. Oceanic Tech. 11, 1388-1391.

Anderson, P. S.: 1996, `Reply', J. Atmos. Oceanic Tech. 13, 913-914.

De Felice, T. P.: 1998, Meteorological Instrumentation and Measurement, Prentice Hall, 225 pp.

Déry, S. J., Taylor, P. A., and Xiao, J.: 1998, `The Thermodynamic Effects of Sublimating, Blowing Snow in the Atmospheric Boundary Layer', Boundary-Layer Meteorol. 89, 251-283.

Déry, S. J. and Yau, M. K.: 1999, `A Climatology of Adverse Winter-Type Weather Events', J. Geophys. Res. 104(D14), 15,657-16,672.

Déry, S. J. and Yau, M. K.: 2001, `Simulation of Blowing Snow in the Canadian Arctic Using a Double-Moment Model', Boundary-Layer Meteorol., in press.

Elliot, W. P. and Gaffen, D. J.: 1991, `On the Utility of Radiosonde Humidity Archived for Climate Studies', Bull. Amer. Meteorol. Soc. 72, 1507-1520.

Environment Canada, Canadian Climate Normals, 1961-1990, vol. 3, The North: Yukon and Northwest Territories, Atmospheric Environment Service, Toronto, Ont., 58 pp.

Essery, R., Li, L., and Pomeroy, J. W.: 1999, `A Distributed Model of Blowing Snow over Complex Terrain', Hydrol. Proc. 13, 2423-2438.

Garand, L., Grassotti, C., Halle, J., and Klein, G.: 1992, `On Differences in Radiosonde Humidity-Reporting Practices and their Implications for Numerical Weather Prediction and Remote Sensing', Bull. Amer. Meteorol. Soc. 73, 1417-1423.

King, J. C. and Anderson, P. S.: 1999, `A Humidity Climatology for Halley, Antarctica based on Hygrometer Measurements', Antarctic Sc. 11, 100-104.

Makkonen, L.: 1996, `Comments on ``A Method for Rescaling Humidity Sensors at Temperatures Well Below Freezing''', J. Atmos. Oceanic Tech. 13, 911-912.

Pomeroy, J. W. and Essery, R. L. H.: 1999, `Turbulent Fluxes During Blowing Snow: Field Tests of Model Sublimation Predictions', Hydrol. Proc. 13, 2963-2975.

Figure 1

Figure 2

Table 1: Details of the six sites investigated in this study. Mean monthly values of temperature (T) and relative humidity with respect to ice (RH_i) for December 2000 recorded at each station are also indicated.

Station | Code | Lat. | Lon. | Elevation | Hygrometer | T | RH_i
| (degrees N) | (degrees W) | (masl) | Type | (degrees C) | (%)
Baker Lake | 2300500 | 64.3 | 96.1 | 18 | dewcell | -29 | 80
Gillam | 5061001 | 56.4 | 94.7 | 145 | HMP35CF | -25 | 95
Hanbury River | 2202351 | 63.6 | 105.1 | 317 | HMP35CF | -30 | 99
Inuvik | 2202582 | 68.3 | 133.5 | 224 | dewcell | -26 | 77
Norman Wells | 2202800 | 65.3 | 126.8 | 241 | dewcell | -28 | 84
Robertson Lake | 2303610 | 65.1 | 102.4 | 244 | HMP35CF | -30 | 97