• Alt, B. T., , and J. B. Maxwell, 1990: The Queen Elizabeth Islands: A case study for Arctic climate data availability and regional climate analysis. Canada's Missing Dimension: Science and History in the Canadian Arctic Islands, C. R. Harrington, Ed., Canada Museum of Nature, 294–326.

    • Search Google Scholar
    • Export Citation
  • Arft, A. M., and Coauthors. 1999: Responses of tundra plants to experimental warming: Meta-analysis of the International Tundra Experiment. Ecol. Monogr., 69 , 491511.

    • Search Google Scholar
    • Export Citation
  • Atkinson, D. E., 2000: Modeling July mean temperatures on the Fosheim Peninsula, Ellesmere Island, Nunavut. Environmental Response to Climate Change in the Canadian High Arctic,Geological Survey of Canada Bull., No. 529, Geological Survey of Canada, 99–111.

    • Search Google Scholar
    • Export Citation
  • Atkinson, D. E., , B. T. Alt, , and K. Gajewski, 2000: A new database of high arctic climate data from the Polar Continental Shelf Project archives. Bull. Amer. Meteor. Soc., 81 , 26212629.

    • Search Google Scholar
    • Export Citation
  • Bonham-Carter, G. F., 1994: Geographic Information Systems for Geoscientists: Modelling with GIS. Pergamon Press, 398 pp.

  • Bradley, R. S., , F. T. Keimig, , and H. F. Diaz, 1992: Climatology of surface-based inversions in the North American Arctic. J. Geophys. Res., 97 (D14) 1569915712.

    • Search Google Scholar
    • Export Citation
  • Daly, C., , and G. L. Johnson, cited. . 1998: PRISM spatial climate layers: Their development and use. [Available online at http://www.ocs.edu/prism/prisguid.pdf.].

    • Search Google Scholar
    • Export Citation
  • Daly, C., , R. P. Neilson, , and D. L. Phillips, 1994: A statistical–topographic model for mapping climatological precipitation over mountainous terrain. J. Appl. Meteor., 33 , 140158.

    • Search Google Scholar
    • Export Citation
  • Daly, C., , G. H. Taylor, , and W. P. Gibson, 1997: The PRISM approach to mapping precipitation and temperature. Preprints, 10th Conf. on Applied Climatology, Reno, NV, Amer. Meteor. Soc., 10–12.

    • Search Google Scholar
    • Export Citation
  • Dodson, R., , and D. Marks, 1997: Daily air temperature interpolated at high spatial resolution over a large mountainous region. Climate Res., 8 , 120.

    • Search Google Scholar
    • Export Citation
  • Edlund, S. A., , and B. T. Alt, 1989: Regional congruence of vegetation and summer climate patterns in the Queen Elizabeth Islands, Northwest Territories, Canada. Arctic, 42 , 323.

    • Search Google Scholar
    • Export Citation
  • Goodale, C. L., , J. D. Aber, , and S. J. Ollinger, 1998: Mapping monthly precipitation, temperature, and solar radiation for Ireland with polynomial regression and a digital elevation model. Climate Res., 10 , 3549.

    • Search Google Scholar
    • Export Citation
  • Havens, J. M., , F. Müller, , and G. C. Wilmot, 1965: Comparative meteorological survey and a short term heat balance study of the White Glacier, summer 1962. McGill University, 68 pp.

    • Search Google Scholar
    • Export Citation
  • Jacobs, J. D., 1990: Integration of automated station data into objective mapping of temperatures for an Arctic region. Climatol. Bull., 24 , 8496.

    • Search Google Scholar
    • Export Citation
  • Johnson, G. L., , C. Daly, , G. H. Taylor, , and C. L. Hanson, 2000: Spatial variability and interpolation of stochastic weather simulation model parameters. J. Appl. Meteor., 39 , 778796.

    • Search Google Scholar
    • Export Citation
  • Kahl, J. D., , M. C. Serreze, , and R. C. Schnell, 1992: Tropospheric low-level temperature inversions in the Canadian Arctic. Atmos.–Ocean, 30 , 511529.

    • Search Google Scholar
    • Export Citation
  • Maxwell, B., 1980: The Climate of the Canadian Arctic Islands and Adjacent Waters: Climatological Studies No. 30. Vol. 1. Atmospheric Environment Service, Environment Canada, 532 pp.

    • Search Google Scholar
    • Export Citation
  • Maxwell, B., 1982: The Climate of the Canadian Arctic Islands and Adjacent Waters: Climatological Studies No. 30. Vol. 2. Atmospheric Environment Service, Environment Canada, 589 pp.

    • Search Google Scholar
    • Export Citation
  • McCracken, D., , and W. Dorn, 1964: Numerical Methods and Fortran Programming. John Wiley and Sons, 457 pp.

  • McCullagh, M. J., 1981: Creation of smooth contours over irregularly distributed data using local surface patches. Geogr. Anal., 13 , 5163.

    • Search Google Scholar
    • Export Citation
  • Santibanez, F., , L. Morales, , J. Delafuente, , P. Cellier, , and A. Huete, 1997: Topoclimatic modeling for minimum temperature prediction at a regional scale in the central valley of Chile. Agronomie, 17 , 307314.

    • Search Google Scholar
    • Export Citation
  • Shepard, D., 1968: A two-dimensional interpolation function for irregularly-spaced data. Proc. 1968 Association for Computing Machinery (ACM) National Conf., Princeton, NJ, ACM, 517–524.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., Ed.,. 1992: Climate System Modeling. Cambridge University Press, 788 pp.

  • Watson, R. T., , M. C. Zinyowera, , and R. H. Moss, Eds.,. . 1995: Climate Change 1995: Impacts, Adaptations and Mitigation of Climate Change: Scientific–Technical Analyses. Cambridge University Press, 879 pp.

    • Search Google Scholar
    • Export Citation
  • Williams, P. J., , and M. W. Smith, 1989: The Frozen Earth—Fundamentals of Geocryology. Cambridge University Press, 306 pp.

  • Willmott, C. J., , and K. Matsuura, 1995: Smart interpolation of annually averaged air temperature in the United States. J. Appl. Meteor., 34 , 25772586.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    The CAA. Upper-air stations operated by the MSC are indicated

  • View in gallery

    Typical vertical temperature profiles for individual ascents showing the inversion (thin line with black dots) and the fitted polynomial curve (heavier black line): (a) 2 Jul 1987 0000 UTC; (b) 23 Jul 1987 1200 UTC. (c) Mean rawinsonde ascent profiles for Jul generated by averaging all polynomial estimates for each ascent over the month of Jul for a given year (solid line). Inversion is removed by extrapolating the straight portion of the original curve (dashed line) to the surface (“high-slope” inversion removal algorithm). All profiles obtained from the Eureka upper-air station

  • View in gallery

    Daily temperature data observed from an automatic weather station (dashed lines) and estimated by the model for the same location (solid lines) for the years and periods indicated. All plotted data series have been filtered using a five-point Gaussian kernel. Automatic weather station was located at the PCSP Hot Weather Creek research camp, 30-km inland from the upper-air station at Eureka on Ellesmere Island

  • View in gallery

    Model results for 9–22 Jul 1974. Results are estimates of mean surface air temperature (°C), integrating the specified time period, arranged on a 1 km × 1 km grid. Locations of all verification sites have been plotted as black dots

  • View in gallery

    Model results for 4–17 Jul 1976. Results are estimates of mean surface air temperature (°C), integrating the specified time period, arranged on a 1 km × 1 km grid. Locations of all verification sites have been plotted as black dots

  • View in gallery

    Model results for 19 Jul–1 Aug 1977. Results are estimates of mean surface air temperature (°C), integrating the specified time period, arranged on a 1 km × 1 km grid. Locations of all verification sites have been plotted as black dots

  • View in gallery

    (a) Residual values for all model runs (n = 386) plotted against distance from coast of the verification station. Distance axis is in km plotted on a log scale. (b) Residual values for all model runs (n = 386) plotted against elevation of the verification station. Elevation axis is in meters plotted on a log scale

  • View in gallery

    Plots of surface air temperature (°C) residuals comparing different sensitivity runs for the 1974 run period (9–22 Jul 1974): Sources of the results are as follows (a) original (high-slope algorithm) model; (b) low-slope algorithm; (c) peak-point algorithm; (d) no curve modification; (e) low-slope algorithm with variable SST (XVD); (f) peak-point algorithm with variable SST (DVD); (g) low-slope algorithm, constant SST, and maximum wind (XFX); (h) low-slope algorithm, variable SST, and maximum wind (XVX). Shading indicates negative residuals (observed less than model estimates); horizontal line pattern indicates positive residuals (observed greater than model estimates)

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 135 135 5
PDF Downloads 22 22 5

High-Resolution Estimation of Summer Surface Air Temperature in the Canadian Arctic Archipelago

View More View Less
  • 1 Laboratory of Paleoclimatology and Climatology, Department of Geography, University of Ottawa, Ottawa, Ontario, Canada
© Get Permissions
Full access

Abstract

In the Canadian high Arctic patterns of temperature are poorly resolved at the mesoscale. This issue is addressed using a model to estimate mean summer surface air temperature at high spatial resolution. The effects on temperature of site elevation and coastal proximity were selected for parameterization. The spatial basis is a 1-km resolution digital elevation model of the region. Lapse rates and resultant wind estimates were obtained from upper-air ascents. These were used to estimate the change in temperature with elevation based on the digital elevation model. Advection effects are handled using resultant winds, air temperature above the ocean, and distance to coast. Model results for 14-day runs were compared to observed data. The two effects captured much of the mesoscale variability of the Arctic climate, as shown by verification with point observational data. Sensitivity analyses were performed on the model to determine response to alterations in lapse rate calculation, sea surface temperature, and wind field generation. The model was most sensitive to the lapse rate calculation. The best results were obtained using a moderate lapse rate calculation, moderate wind field, and variable sea surface temperature.

Corresponding author address: Dr. David E. Atkinson, Geological Survey of Canada (Atlantic), Bedford Institute of Oceanography, 1 Challenger Dr., P.O. Box 1006, Dartmouth NS B2Y 4A2, Canada. Email: datkinso@nrcan.gc.ca

Abstract

In the Canadian high Arctic patterns of temperature are poorly resolved at the mesoscale. This issue is addressed using a model to estimate mean summer surface air temperature at high spatial resolution. The effects on temperature of site elevation and coastal proximity were selected for parameterization. The spatial basis is a 1-km resolution digital elevation model of the region. Lapse rates and resultant wind estimates were obtained from upper-air ascents. These were used to estimate the change in temperature with elevation based on the digital elevation model. Advection effects are handled using resultant winds, air temperature above the ocean, and distance to coast. Model results for 14-day runs were compared to observed data. The two effects captured much of the mesoscale variability of the Arctic climate, as shown by verification with point observational data. Sensitivity analyses were performed on the model to determine response to alterations in lapse rate calculation, sea surface temperature, and wind field generation. The model was most sensitive to the lapse rate calculation. The best results were obtained using a moderate lapse rate calculation, moderate wind field, and variable sea surface temperature.

Corresponding author address: Dr. David E. Atkinson, Geological Survey of Canada (Atlantic), Bedford Institute of Oceanography, 1 Challenger Dr., P.O. Box 1006, Dartmouth NS B2Y 4A2, Canada. Email: datkinso@nrcan.gc.ca

1. Introduction

Interactions of the earth's surface with the atmosphere are particularly evident in the Arctic. Plant survival and growth is closely tied to the climate (Arft et al. 1999) and the presence of permafrost is a major influence on landscape dynamics (Williams and Smith 1989). Understanding such climate–surface interactions is important, as future climate changes are predicted to be greater here than in most areas of the world with a potentially large impact on the landscape (Watson et al. 1995). However, at the present time, environmental and paleoenvironmental research in the Arctic is hampered by a lack of mesoscale climate, and most importantly temperature, data.

The Canadian Arctic Archipelago (CAA; Fig. 1) is served by few meteorological stations. The mean separation between stations of the Meteorological Service of Canada (MSC) is 500 km and the representativeness of all stations suffers due to local coastal bias and highly varied topography and surface types. Physiography contributes to temperature pattern variability at the mesoscale for various reasons. The archipelago is heavily fiorded, exposing land areas to an ocean that can be ice covered, contain isolated floating ice floes, or be ice free, which varies significantly on an annual basis. Land surfaces near the coast experience a typical pattern of maritime attenuation, whereas the interiors of larger islands exhibit continental conditions. Exceptions are areas near snowpacks or extended ice fields, which are cooled in the summer. Topographic complexity also contributes to mesoscale variability in temperature, precipitation, and cloudiness. These factors serve to render questionable results taken from surface air temperature plots that are based on interpolation from the few available meteorological stations.

Improving the spatial resolution of surface air temperature estimates is thus an important contribution to an understanding of surface climate in this region, and one that is not forthcoming from the existing observational network. Two mechanisms exist to better understand mesoscale temperature: integrating alternate data or information into an analysis (Atkinson et al. 2000; Atkinson 2000; Kahl et al. 1992; Alt and Maxwell 1990) or using empirical (Willmott and Matsuura 1995; Daly et al. 1994) or physical models (Trenberth 1992) of the atmosphere to augment traditional analyses and/or data sources.

Maxwell (1980, 1982) used information from historical short-term stations and his own experience to subjectively modify isotherms to depict cooler ice field/upland regions. Alt and Maxwell (1990) employed nonstandard, short-term weather observation data from several, more recent sources (e.g., Atkinson et al. 2000) to increase spatial detail of a July temperature normal plot for the Queen Elizabeth Islands. Jacobs (1990) linked an automatic weather station to MSC weather stations using transfer functions allowing the generation of data at a “virtual” station. Other studies have used the approach of guided temperature estimation using a digital elevation model (DEM) in conjunction with a lapse rate for detailed climate work (Daly et al. 1994; Willmott and Matsuura 1995; Daly et al. 1997; Daly and Johston 1998; Johnson et al. 2000) or to support other types of research (Santibanez et al. 1997; Goodale et al. 1998; Dodson and Marks 1997).

In this paper, we describe a semiempirical model of the mesoscale summer temperature climate of the Arctic. The conceptual basis for the model is that much of the spatial variability of the Arctic surface temperature regime can be accounted for by several processes. Specifically, we hypothesized that the two most important contributors to the spatial variability of surface temperature patterns at the mesoscale (horizontal scale of tens to hundreds of kilometers) are 1) variation of temperature with elevation, and 2) location with respect to advective sources of air temperature modification, such as large bodies of water or ice fields.

Elevational effects were targeted because many of the islands consist of large central plateaus with a small coastal zone. In the northern and eastern parts of the archipelago, significant mountainous regions are found. Concurrent lapse rates applied to site elevations were felt to be the best way to improve estimates of temperature in these areas. Advective effects were also modeled because many of the islands in the CAA are large enough to possess a coast-to-interior heating gradient that ranges from unimpeded surface heating in the interior to coastal locations completely dominated by maritime air.

In general, the surface temperature climate at the mesoscale is formed by the interaction of the synoptic situation with the underlying surface. The model described below takes as input the synoptic-scale features of the temperature field, as estimated from the MSC upper-air stations, and modifies their signal using elevational and coastal proximity data derived from the DEM.

2. Data and model description

The model was implemented at a spatial resolution of 1 km × 1 km. Physical processes accommodated in this model areas follows:

  • The mean environmental lapse rate specific to the time period being modeled, derived using temperature data from rawinsonde ascents at MSC upper-air stations in the study region, is used to define the rate of temperature change with elevation.
  • The mean, low-level wind direction and velocity, derived from rawinsonde ascents, is used to determine the extent to which coastal zones are modified by onshore advective flow.
  • Surface temperatures for locations possessing major ice fields are stipulated using a linear modification of the base temperature estimate.

The spatial basis of the model is a DEM of the Canadian Arctic Archipelago, organized as a matrix of 1996 columns by 1833 rows, subset from the U.S. Geological Survey GTOPO30 DEM of the world (available online at http://edcdaac.usgs.gov/gtopo30/gtopo30.html). Each point represents approximately 1 km2.

The first step in estimating surface air temperature values for each point was to obtain mean environmental lapse rates for each station. These were generated using vertical profiles of dry-bulb temperature obtained from twice-daily rawinsonde ascents at stations throughout the region (Table 1). The mean ascent curve was described using a fifth-order polynomial. A high-order polynomial was used because it was felt important to model a shallow, surface inversion that was found to be present in many of the ascent profiles (Figs. 2a,b), which are discussed below.

The inversions (Table 1) were smaller in magnitude than those observed in winter (Bradley et al. 1992; Maxwell 1980). Their likely cause is advective, rather than radiative, given that the summer net surface radiation balance is positive. It was thus assumed that a summer surface inversion at a coastal location is a local-scale effect that must be removed before using the environmental lapse rate to represent interior sites.

Removal of the inversion involved first detecting the inflection point on the curve above the inversion using a global-maximum detection algorithm (McCracken and Dorn 1964). Next, data were extrapolated from this point to the surface using the rate of change that existed in the curve above the inversion. The new ascent series than had the polynomial equation refit to it (Fig. 2c). This procedure was verified by comparing estimates of surface temperature made by the refit polynomial to observations from summer research camps at inland sites (Atkinson 2000).

Next, the polynomial coefficients representing the lapse rate at each station were interpolated throughout the DEM grid. Each coefficient was interpolated individually onto the grid using an inverse distance weight procedure with decay set to a factor of 2; this was selected to provide a balance between local weighting and range of influence. The paucity of observing sites and a lack of spatial structure (e.g., no point clustering) did not warrant use of more specialized interpolation techniques (e.g., McCullagh 1981; Shepard 1968). Temperature values were then obtained by solving the equation at each grid point using elevation data as the independent value. This gave a regionwide estimate of surface temperature that reflected the environmental lapse rate without a coastal signal. A concern when using upper-air temperature data to estimate near-surface air temperatures is the potential for underestimation of near-surface temperatures; however, consistent bias or large departures from verification data were not observed (Figs. 3a–d).

Temperatures for ice fields were then estimated. Havens et al. (1965) demonstrated an average “ice field cooling” factor of about 3°C using data from two stations, one on top of an ice field and the other nearby on a nonice surface. In the model, ice field locations were assigned a new value that consisted of the initial temperature estimate minus this cooling factor.

Next the coastal effect was parameterized. The influence of wind for this model is expressed as a mixing of the base land estimates, obtained as described above, with the temperature over the ocean. Wind velocity from the 90-kPa level was extracted from upper-air ascents. The 90-kPa level was selected because it is high enough (∼900 m) to be above most topography and to possess the steady characteristics of winds at higher levels, yet low enough to reasonably represent the direction and speed of winds felt at the surface. Based on these values the image was classified into four direction and speed classes, giving a total of 13 categories (12 categories when speeds were >0, and 1 category for 0 wind speeds) (Table 2). Velocity classification was based on a breakdown of observed wind speeds such that the majority of wind events fell into the “low” category and progressively fewer into the “medium” and “high” categories. These wind categories formed the basis for the selection of a “matrix filter” (Bonham-Carter 1994) that was applied to a binary representation of the DEM in which land pixels are assigned a value of 1 and ocean pixels are assigned 0. A matrix filter is a small, square matrix composed of values that are symmetric and opposite. This filter is placed over a given pixel on the binary DEM. The neighborhood around the pixel that matches the filter in size is extracted from the binary DEM and multiplied, pixel by pixel, with the filter. All the values in the resulting matrix are then summed to arrive at a single value; this value represents the potential wind influence on a pixel, which is used in Eq. (1) below. The filter is arranged such that the largest values are near the middle, representing close proximity to the ocean, with a steady decay to the edge of the filter. Thus, pixels near the ocean will feel the greatest potential influence of an onshore flow, decreasing with distance from the coast. The effects of a stronger flow, which are greater potential impact on the near-shore environment and farther potential inland penetration, is represented by a filter that has both larger values, to capture greater impact, and larger physical size, to represent a greater inland penetration. Greater physical size of the filter is used to represent the increased range of effect of a stronger flow because in the DEM the wind filter cannot be applied to an area that it does not physically reach.

The result of application of the wind filter was a “resultant wind effect” parameter that represented a potential modification of the base temperature estimate at a site. The maximum value for a resultant wind effect is 100, indicating that 100% of the temperature at that pixel is a result of ocean influence, and the minimum is 0, for no modification due to wind. The wind effect parameter and the values from the temperature estimates image were combined using
TrWTsWTL
where Tr = resultant temperature value at a given point, W = wind modification value (%), Ts = air temperature over the ocean surface, and TL = air temperature over the land surface obtained from the polynomial-based estimate. Values for Ts were set at 2°C or ranged between 0° and 4°C depending on the type of model run being conducted. Topographic modification of wind was not explicitly parameterized.

The final output of a model run was a 1 km × 1 km grid of estimates of the mean surface air temperature for the period of the model run. Values were estimated for all land surfaces over a region encompassing all the islands in the Canadian Arctic Archipelago, Boothia Peninsula, and some of the north coast of the mainland.

Temperatures were estimated for 2-week averaging periods using an initial set of parameterizations (identified as “original”), one run for each of 17 years (Table 3), although the model can be run for any averaging period. Dates of application varied from year to year and were chosen to maximize the availability of observational data (Atkinson 2000) for verification purposes. Based on these results various permutations of model parameterization changes were run on subsets of 5 years. Three parameterizations were targeted for sensitivity analysis: the inversion removal algorithm, the sea surface temperature (SST), and the wind effect (Table 4). Each sensitivity combination was tested on 5 separate years; the same 5 years were used in each case to permit comparison (Table 5). Three additional approaches to dealing with the inversion were considered: a “peak-point” removal, in which the slope removal line was drawn vertically down from the point of maximum warming to the surface; a “low-slope” removal, in which the slope removal line was drawn from a point roughly halfway between the original model and the peak-point removal; and “none,” in which no alteration to the observed lapse rate was performed. These represent a gradation in the magnitude of inversion removal, from a maximum in the original model (“high-slope” removal) to no alteration (none). For SST, the constant value was replaced by values derived from a map of mean observed SST (Maxwell 1982). The existing wind effect was increased in strength, such that its influence could be felt twice as far inland as in the original model.

For verification, mean temperature values were calculated using available surface stations present during the period of the run. This included both MSC and nonstandard data from the Polar Continental Shelf Project archives (Atkinson et al. 2000). Model estimates at grid points coinciding with the station locations were extracted and residuals were calculated. Residuals within the range ±1.4°C were considered acceptable, as this is the minimum standard deviation of two-week means obtained from the observational data. Values outside of this range were mapped to gauge the performance of the model by revealing regions of systematic over- or underestimation.

3. Results

Output from selected years is presented in Figs. 4–6 and Table 6. Temperature values have been rounded to the nearest whole degree Celsius. As expected, cooler temperatures were found at higher elevations in the eastern Arctic. A general north–south temperature gradient was also captured, as was a “northwest cool bulge,” which is a typical temperature pattern caused by the persistent penetration of cool air southeast from the Arctic Ocean into the central CAA.

The primary diagnostic tool for assessing model performance was a set of residuals obtained by subtracting model estimates from observed station temperature data. Model estimates were obtained from the grid points closest to a given station, and the observed station data were averaged for the time period coincident with the model run. These residuals were processed as a complete set (i.e., and not by year) to obtain a mean absolute error and were both plotted against station distance from coast and elevation to look for situational biases and in a mapped form to identify spatial zones of model inconsistency. An overall mean absolute error of 1.5°C was obtained on the 386 residual values available for the 17 two-week model runs. Considering the residuals separately for each run, the residuals ranged from having a mean that was close to zero with low variation (e.g., 1976) to a mean that deviated significantly from zero with large variation (e.g., 1974). Overall, negative residuals (model overestimation) were more common. All available residuals were plotted against station elevation and distance from coast (Fig. 7). Biases in the model estimates were not apparent.

A more detailed assessment of model performance was determined by considering the size of zones formed by residuals of a given sign. A large residual zone of a given sign is more likely a result of a systematic shortcoming in the model, whereas small, discontinuous residual zones of both signs indicate local forcing agents or random fluctuations. The 129 residuals that fell outside the acceptable range formed a total of 57 zones; those zones, which possessed three or more stations, represented only 24% of all observed zones (Table 7). Seven residual zones possessed five or more stations; of these six were negative and in all cases the residual zones were situated largely in the northern part of the archipelago. Several persistent features were noted in the residuals plots. Large and small zones of positive residual were frequent in the north, on Ellesmere and Axel Heiberg Islands (1977, 1978, 1982, 1983, 1984, 1988, 1990), and often in the eastern parts of Ellesmere (1977, 1988, 1990).

In general, none of the sensitivity combinations investigated yielded a clearly superior result (Table 8); however, they all yielded results that were superior to the original inversion removal algorithm (high slope). Applying different inversion removal algorithms while maintaining the constant sea surface temperature resulted in skewed residual groupings: skewed negative using the high-slope and low-slope removals, and skewed positive using the peak-point and none residuals, with the total number of residuals in the acceptable category changing little each time. Similar skewed results were observed when using the stronger wind field with the low-slope removal. The most even distribution of residuals was obtained using the low-slope inversion removal with a variable sea surface temperature; however, it must be noted that this method also yielded one of the lowest numbers of residuals in the acceptable range. The largest number of residuals in the acceptable range was obtained using no inversion removal; however, it also generated the most highly skewed residuals set. An examination of the residuals of the sensitivity analyses shows how the different alterations affected specific years (Table 9).

Altering the inversion removal to reduce the magnitude of lapse rate correction resulted in the large negative residual zones being reduced in size and/or broken up into smaller zones (e.g., Figs. 8a–h). Changes introduced by alteration of inversion removal were more significant that those resulting from altering the SST or wind fields, which tended to result more in sporadic, low-magnitude changes.

4. Discussion

Areas well represented in the original model included the central, south-central, and west-southwest regions. When the model was in error it tended to overestimate temperatures. In several years large zones of model overestimation were observed in the northwest and along the eastern edge of the archipelago. Model overestimation occurred in the presence of unusually deep and widespread inversions or when a shallow, surface inversion has undergone a slight surface warming. The large zones of systematic overestimation (zones of seven or more stations) accounted for 27% of all residual values and were confined to four specific years in the 17-yr run period. When these four years were excluded, the residuals showed no particular tendency toward positive or negative skewing.

In several cases the model overestimated when the resultant wind was zero. This occurred because, without an onshore wind component, the model did not apply any cooling to coastal areas. This can be remedied by allowing some cooling for areas close to the coast even during conditions of zero wind. A related problem is that the model overestimated temperatures on small islands, such as Prince Leopold Island or Seymour Island. In these cases the radiative heating capacity of the small land area of the island is insufficient to modify the cool lowest levels of the atmosphere. Application of a corrected vertical temperature profile, which has been designed for the interiors of large islands, thus overestimated at these locations. This can be remedied by allowing the model to take into account the size of the land area a given location is situated in, and adjusting the lapse rate accordingly. Finally, too few wind direction options in the wind filter was another cause of wind-related model overestimation at coastal locations. Using eight, instead of four, wind directions would improve this. Another problem that may contribute to a model overestimation in the “original” model runs in the northwest is the value used for the air temperature over the ocean. That is, 2°C is too high for an ocean that is usually ice covered. This was altered for some of the sensitivity runs in which a variable air temperature over the ocean was used and found to be an improvement.

Residuals along the eastern edge of the archipelago most likely occurred because none of the upper-air profiles are characteristic of this region. Alert, while on the coast near the eastern coastal region, is located at the extreme northern limit of this area, which limits the representativeness of its profile. Furthermore, the nature of the interpolation procedure is such that, for much of the central-east coast of Ellesmere Island, the influence exerted by Eureka's vertical profile, a station poorly suited to guide estimates in this area, will exceed that of Alert. Incorporation of vertical profiles from Thule Air Force Base in western Greenland could improve this situation.

Overestimation occurred more frequently than underestimation, and the magnitude of most of the positive residuals was small. That fact, coupled with the spatial distribution of residuals, did not indicate systematic model underestimation. The most likely situation in which positive residuals would occur is during periods of low cloud cover and low wind speed when surface heating is greatest. Examination of specific situations revealed that this was not necessarily the case and that there were several instances of underestimation when reported cloud cover was high. More work is needed to investigate the relationship between clouds, wind, and temperature to ascertain whether the inclusion of cloud cover, perhaps based on wet bulb depression from the ascents, should be considered to better estimate in these situations.

There were few situations in which the modeled wind effect could be assessed due to lack of data for verification, although there were instances in which stations were located within an onshore wind zone that correctly modified their temperatures (e.g., 1981, 1988). In several years there were large areas in which the model apparently overestimated temperature (e.g., 1975, 1976, 1977, 1979), but these large areas consisted of just a few stations, all situated on the coast, that were joined to form a contiguous zone by contouring. These errors are confined to coastal regions. This suggests that the model is capable of generating accurate temperature estimates for large areas of the Arctic.

Sensitivity analyses indicated that modification to the inversion removal algorithm had the largest impacts in those years for which deep inversions were observed at multiple stations (1974 and 1988). A successive decrease in the magnitude of inversion removal resulted in corresponding decreases in the magnitude and occurrence of negative residuals. This suggests that the original model algorithm overcompensated when a deep inversion was present, resulting in temperature estimates that were too large.

Although the occurrence of overcompensation was reduced using inversion removals of lower magnitude, excessive overcompensation resulted in a problem with model underestimation. In the “peak-point” and “none” removal results for 1974, underestimation was almost as common as overestimation, but these residuals did not form a large, continuous zone similar to that observed using the original algorithm. In 1988 larger zones of positive residual appeared, notably in the central-north and the west. 1988 was a warm year in the region surrounding MSC Eureka (Edlund and Alt 1989). In general, underestimated sites were situated inland or in sheltered areas, such as at the head of a fiord. In 1988, however, sites that were exposed to a stronger maritime influence, such as that to the east of Ellesmere Island or to the northwest, were still overestimated by the model, even in the absence of an inversion correction, indicating that an erroneous inversion removal algorithm was not the only cause of model overestimation.

A large underestimation that occurred in north Ellesmere Island in 1988 (Biederbick Lake) provides a good example of the potential problem that can exist with inland sites. The main estimator for this site is the temperature profile from MSC Alert. A location like Alert, unlike a sheltered, inland site, is cooled by two mechanisms: cool air advection, and blocking of insolation by low-level cloud. Biederbick Lake, however, experienced many days of low wind and cloud, which most likely allowed the site to realize its maximum potential warming, and which made it different enough from Alert that even a corrected temperature profile was unable to reproduce its observed temperature. Although small, local variability will always be a problem in a topographically diverse region such as the High Arctic, the general mesoscale patterns were captured by the model.

Overall, the model was most sensitive to the manner in which inversions were handled. In terms of area of effect, modifications to this parameter also had the largest effect. For all combinations of parameterization, the majority of residuals were within the acceptable range. Where they exceeded the acceptable range, examination of their patterns suggested physical causes. The general accuracy and physical interpretability of the model results suggest that this is a promising avenue for Arctic climate research, both to generate high spatial resolution temperature data and to explore the physical processes that control the climate in this region.

Another consideration for future work is to expand the time periods over which the model can currently operate by using vertical temperature profiles and winds generated by general circulation models (GCMs) or the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis upper-air gridded dataset. A model of this nature also provides a means of rendering GCM output in high spatial resolution because the low-resolution, gridded vertical temperature fields generated by the GCM can be used as surrogate upper-air stations in the model. Using gridded profiles would require detailed work to determine how such profiles should be treated; that is, coastal or inland.

The processes selected for parameterization in a model of this nature are dependent on the spatial scale of the model. Use of a higher-resolution DEM would necessitate incorporation of smaller-scale processes, such as down-valley winds or slope-differential radiation effects. The current version of this model is designed for summer use in the high Arctic. For use in other seasons modifications to the handling of inversions would have to be introduced, because the nature of the inversion changes in the wintertime Arctic. Modification to handle the wintertime situation is readily possible, as would be modification to port the model to other regions of the world. Work is currently underway for use of the model in the Yukon.

The model in its present state is limited only by the temporal resolution of the input environmental lapse rate data, in this case, twice-daily upper-air series. It is theoretically possible to generate estimates on a daily or even twice-daily basis; however, estimation at high temporal resolution would possess increased error because of variability in the upper-air series. Estimates that integrate several days mitigate problems in the upper-air series. Estimation of daily maximum and minimum temperatures could only be performed in conjunction with some knowledge of the near-surface radiation regime. It would be possible to get a rough idea of cloud from the upper-air data because dewpoint temperature is recorded; however, clouds, and therefore maximum and minimum temperature, have greater spatial variability than the mean temperature field and the accuracy of estimates generated by the model for these parameters would be lower than that for mean temperature.

Future work will improve the model by better parameterization of the climate of the lower atmosphere and condition of the surface, and specifically the following:

  1. An assessment of the summer inversion regime to tailor model operation to the different types of inversion that are present;
  2. Improving the near-shore low-wind cooling effect, so that even under no wind, unless there is a strong wind from the land side, the model should apply some sort of coastal cooling to land grid points situated beside ocean points;
  3. Improving the spatial distribution of air temperature estimates above the ocean by incorporating a more detailed sea surface temperature or mean sea ice conditions map;
  4. Using the vertical profile of dewpoint temperature in conjunction with the dry-bulb temperature profile to get an estimate of clouds; and
  5. Incorporating a more detailed ice field/glacier map and model downslope katabatic effects in large glacial valleys.

In general the model has performed satisfactorily and physical mechanisms can be suggested for most errors. This study has shown that a significant amount of the mesoscale climatic variability of the Arctic may be explained by a few factors operating on the complex topography of the Arctic.

Acknowledgments

This paper is a contribution to the Climate Systems History and Dynamics (CSHD) project funded by the Meteorological Service of Canada (MSC) and the Natural Science and Engineering Council of Canada (NSERC). The DEM data are distributed by the EROS Data Center Distributed Active Archive Center (EDC DAAC), located at the U.S. Geological Survey's EROS Data Center in Sioux Falls, South Dakota.

REFERENCES

  • Alt, B. T., , and J. B. Maxwell, 1990: The Queen Elizabeth Islands: A case study for Arctic climate data availability and regional climate analysis. Canada's Missing Dimension: Science and History in the Canadian Arctic Islands, C. R. Harrington, Ed., Canada Museum of Nature, 294–326.

    • Search Google Scholar
    • Export Citation
  • Arft, A. M., and Coauthors. 1999: Responses of tundra plants to experimental warming: Meta-analysis of the International Tundra Experiment. Ecol. Monogr., 69 , 491511.

    • Search Google Scholar
    • Export Citation
  • Atkinson, D. E., 2000: Modeling July mean temperatures on the Fosheim Peninsula, Ellesmere Island, Nunavut. Environmental Response to Climate Change in the Canadian High Arctic,Geological Survey of Canada Bull., No. 529, Geological Survey of Canada, 99–111.

    • Search Google Scholar
    • Export Citation
  • Atkinson, D. E., , B. T. Alt, , and K. Gajewski, 2000: A new database of high arctic climate data from the Polar Continental Shelf Project archives. Bull. Amer. Meteor. Soc., 81 , 26212629.

    • Search Google Scholar
    • Export Citation
  • Bonham-Carter, G. F., 1994: Geographic Information Systems for Geoscientists: Modelling with GIS. Pergamon Press, 398 pp.

  • Bradley, R. S., , F. T. Keimig, , and H. F. Diaz, 1992: Climatology of surface-based inversions in the North American Arctic. J. Geophys. Res., 97 (D14) 1569915712.

    • Search Google Scholar
    • Export Citation
  • Daly, C., , and G. L. Johnson, cited. . 1998: PRISM spatial climate layers: Their development and use. [Available online at http://www.ocs.edu/prism/prisguid.pdf.].

    • Search Google Scholar
    • Export Citation
  • Daly, C., , R. P. Neilson, , and D. L. Phillips, 1994: A statistical–topographic model for mapping climatological precipitation over mountainous terrain. J. Appl. Meteor., 33 , 140158.

    • Search Google Scholar
    • Export Citation
  • Daly, C., , G. H. Taylor, , and W. P. Gibson, 1997: The PRISM approach to mapping precipitation and temperature. Preprints, 10th Conf. on Applied Climatology, Reno, NV, Amer. Meteor. Soc., 10–12.

    • Search Google Scholar
    • Export Citation
  • Dodson, R., , and D. Marks, 1997: Daily air temperature interpolated at high spatial resolution over a large mountainous region. Climate Res., 8 , 120.

    • Search Google Scholar
    • Export Citation
  • Edlund, S. A., , and B. T. Alt, 1989: Regional congruence of vegetation and summer climate patterns in the Queen Elizabeth Islands, Northwest Territories, Canada. Arctic, 42 , 323.

    • Search Google Scholar
    • Export Citation
  • Goodale, C. L., , J. D. Aber, , and S. J. Ollinger, 1998: Mapping monthly precipitation, temperature, and solar radiation for Ireland with polynomial regression and a digital elevation model. Climate Res., 10 , 3549.

    • Search Google Scholar
    • Export Citation
  • Havens, J. M., , F. Müller, , and G. C. Wilmot, 1965: Comparative meteorological survey and a short term heat balance study of the White Glacier, summer 1962. McGill University, 68 pp.

    • Search Google Scholar
    • Export Citation
  • Jacobs, J. D., 1990: Integration of automated station data into objective mapping of temperatures for an Arctic region. Climatol. Bull., 24 , 8496.

    • Search Google Scholar
    • Export Citation
  • Johnson, G. L., , C. Daly, , G. H. Taylor, , and C. L. Hanson, 2000: Spatial variability and interpolation of stochastic weather simulation model parameters. J. Appl. Meteor., 39 , 778796.

    • Search Google Scholar
    • Export Citation
  • Kahl, J. D., , M. C. Serreze, , and R. C. Schnell, 1992: Tropospheric low-level temperature inversions in the Canadian Arctic. Atmos.–Ocean, 30 , 511529.

    • Search Google Scholar
    • Export Citation
  • Maxwell, B., 1980: The Climate of the Canadian Arctic Islands and Adjacent Waters: Climatological Studies No. 30. Vol. 1. Atmospheric Environment Service, Environment Canada, 532 pp.

    • Search Google Scholar
    • Export Citation
  • Maxwell, B., 1982: The Climate of the Canadian Arctic Islands and Adjacent Waters: Climatological Studies No. 30. Vol. 2. Atmospheric Environment Service, Environment Canada, 589 pp.

    • Search Google Scholar
    • Export Citation
  • McCracken, D., , and W. Dorn, 1964: Numerical Methods and Fortran Programming. John Wiley and Sons, 457 pp.

  • McCullagh, M. J., 1981: Creation of smooth contours over irregularly distributed data using local surface patches. Geogr. Anal., 13 , 5163.

    • Search Google Scholar
    • Export Citation
  • Santibanez, F., , L. Morales, , J. Delafuente, , P. Cellier, , and A. Huete, 1997: Topoclimatic modeling for minimum temperature prediction at a regional scale in the central valley of Chile. Agronomie, 17 , 307314.

    • Search Google Scholar
    • Export Citation
  • Shepard, D., 1968: A two-dimensional interpolation function for irregularly-spaced data. Proc. 1968 Association for Computing Machinery (ACM) National Conf., Princeton, NJ, ACM, 517–524.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., Ed.,. 1992: Climate System Modeling. Cambridge University Press, 788 pp.

  • Watson, R. T., , M. C. Zinyowera, , and R. H. Moss, Eds.,. . 1995: Climate Change 1995: Impacts, Adaptations and Mitigation of Climate Change: Scientific–Technical Analyses. Cambridge University Press, 879 pp.

    • Search Google Scholar
    • Export Citation
  • Williams, P. J., , and M. W. Smith, 1989: The Frozen Earth—Fundamentals of Geocryology. Cambridge University Press, 306 pp.

  • Willmott, C. J., , and K. Matsuura, 1995: Smart interpolation of annually averaged air temperature in the United States. J. Appl. Meteor., 34 , 25772586.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

The CAA. Upper-air stations operated by the MSC are indicated

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3601:HREOSS>2.0.CO;2

Fig. 2.
Fig. 2.

Typical vertical temperature profiles for individual ascents showing the inversion (thin line with black dots) and the fitted polynomial curve (heavier black line): (a) 2 Jul 1987 0000 UTC; (b) 23 Jul 1987 1200 UTC. (c) Mean rawinsonde ascent profiles for Jul generated by averaging all polynomial estimates for each ascent over the month of Jul for a given year (solid line). Inversion is removed by extrapolating the straight portion of the original curve (dashed line) to the surface (“high-slope” inversion removal algorithm). All profiles obtained from the Eureka upper-air station

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3601:HREOSS>2.0.CO;2

Fig. 3.
Fig. 3.

Daily temperature data observed from an automatic weather station (dashed lines) and estimated by the model for the same location (solid lines) for the years and periods indicated. All plotted data series have been filtered using a five-point Gaussian kernel. Automatic weather station was located at the PCSP Hot Weather Creek research camp, 30-km inland from the upper-air station at Eureka on Ellesmere Island

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3601:HREOSS>2.0.CO;2

Fig. 4.
Fig. 4.

Model results for 9–22 Jul 1974. Results are estimates of mean surface air temperature (°C), integrating the specified time period, arranged on a 1 km × 1 km grid. Locations of all verification sites have been plotted as black dots

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3601:HREOSS>2.0.CO;2

Fig. 5.
Fig. 5.

Model results for 4–17 Jul 1976. Results are estimates of mean surface air temperature (°C), integrating the specified time period, arranged on a 1 km × 1 km grid. Locations of all verification sites have been plotted as black dots

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3601:HREOSS>2.0.CO;2

Fig. 6.
Fig. 6.

Model results for 19 Jul–1 Aug 1977. Results are estimates of mean surface air temperature (°C), integrating the specified time period, arranged on a 1 km × 1 km grid. Locations of all verification sites have been plotted as black dots

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3601:HREOSS>2.0.CO;2

Fig. 7.
Fig. 7.

(a) Residual values for all model runs (n = 386) plotted against distance from coast of the verification station. Distance axis is in km plotted on a log scale. (b) Residual values for all model runs (n = 386) plotted against elevation of the verification station. Elevation axis is in meters plotted on a log scale

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3601:HREOSS>2.0.CO;2

Fig. 8.
Fig. 8.

Plots of surface air temperature (°C) residuals comparing different sensitivity runs for the 1974 run period (9–22 Jul 1974): Sources of the results are as follows (a) original (high-slope algorithm) model; (b) low-slope algorithm; (c) peak-point algorithm; (d) no curve modification; (e) low-slope algorithm with variable SST (XVD); (f) peak-point algorithm with variable SST (DVD); (g) low-slope algorithm, constant SST, and maximum wind (XFX); (h) low-slope algorithm, variable SST, and maximum wind (XVX). Shading indicates negative residuals (observed less than model estimates); horizontal line pattern indicates positive residuals (observed greater than model estimates)

Citation: Journal of Climate 15, 24; 10.1175/1520-0442(2002)015<3601:HREOSS>2.0.CO;2

Table 1.

Upper-air stations used to generate regional estimates of environmental lapse rate. Frequency of inversions observed in mean ascent curves during model runs (1974–88, 1990) are listed. Value class is height of the inversion maximum in m above the ground. Here GT refers to “greater than 700 m” (observed only at the Alaska stations)

Table 1.
Table 2.

Wind direction and velocity classification categories

Table 2.
Table 3.

Model run periods selected for the “original” parameterization set

Table 3.
Table 4.

Sensitivity analyses

Table 4.
Table 5.

Periods for which sensitivity analyses were run

Table 5.
Table 6.

Surface air temperature (°C) mean and std dev, and sample size of observed values, model estimates, and residuals for each year of “original” model run and averaged for all years

Table 6.
Table 7.

Frequency of occurrence of residual zones possessing certain number of stations, by residual type

Table 7.
Table 8.

Residual totals by model factor parameterization set

Table 8.
Table 9.

Surface air temperature (°C) mean and std dev, and sample size of observed values, model estimates, and residuals for each year of model run and for all years

Table 9.
Save