Snow Depth and Snow Water Equivalent
There are very few measurements of snow characteristics on Arctic sea ice. The most comprehensive data set of these measurements, in terms of area and time covered, is the data set from the Russian North Pole (NP) drifting stations. Warren et al. interpreted these NP data in order to represent the spatial and temporal variation of snow depth on sea ice. We use the interpretation of Warren et al. here because the NP measurements are the only source of snow depth data complete enough to function as a climatology.
Warren et al. [1999] have represented the geographical and seasonal variation in snow depth by fitting a two-dimensional quadratic function to available data by month. The surface given by this function can be treated as a field. This representation is the best possible, in the absence of dense station coverage. Over land, coverage is more extensive. Station data from data sets available from NSIDC and NOAA, as well as station data from the Russian coastal stations on this Atlas, are used to create grids of mean snow depth.
Snow depth fields for land and for ocean are presented separately, because they represent different ranges in time and vastly different sampling densities.
Ocean
Snow depth and snow water equivalent were obtained at NP stations in accordance with methods described in the Arctic and Antarctic Research Institute (AARI) manual [Arctic and Antarctic Research Institute, 1985]. Daily measurements were made at the weather station at fixed stakes, and once or thrice monthly at 10 m intervals on a survey line distant from the weather station, and 500 m or 1000 m long. The survey line was offset from the previous line, so that measurements were always of undisturbed snow. Snow density was measured along the survey line every 100 m by weighing a cylinder of snow that sampled down to the ice surface. Snow water equivalent for the snow line was computed by multiplying snow depth by the ratio of the density of the snow to the density of water. See Radionov et al. [1996] and Colony et al. [1998] for more information on snow measurements at NP stations. The NP snow data are available from NSIDC (contact NSIDC User Services for information).
Measurements of snow depth from the snow lines total 499 for the years 1954 through 1991. Warren et al. [1999] use snow depth measurements from the snow lines rather than those from the stakes, when snow line measurements are available. They estimate the uncertainty in mean snow depth for a single snow line to be between about 0.5 cm and 2.0 cm, depending on the length of the line, the time of year, and the method used to estimate uncertainty.
An estimate of snow depth every five days for the operating life of each station was constructed with the snow line and stake data by interpolating if necessary. If snow water equivalent was not available, it was estimated using a density measurement, or if density was not available, the mean density for that month was used. These five-day values were averaged to a monthly value for each station. A two-dimensional quadratic function (six coefficients) was fitted by least squares for each month to all station data over the entire 37 year period, to produce 12 fits. There are about 70 independent data values for each month (an average of two stations reporting during each of 37 years), from which six (but no more) coefficients can be determined. See Warren et al. [1999] for further details.
Land
Five data sets were used to compute long-term monthly means for the snow fields over land.
Snowfall and Snow Depth for Canada, 1943-1982 has snow depth values at the end of the month. These end-of-month values were used for the Atlas fields because monthly average values were not available. Long-term monthly means were computed for each station where a given month is represented in the data record in 70 percent or more of the years from 1966 through 1982 (the period of overlap for the land data sources). The data are originally from the Atmospheric Environment Service, Canada, and from the NOAA National Climatic Data Center. The data set was compiled by J. Walsh, University of Illinois. It is available from NSIDC (http://nsidc.org/NSIDC/CATALOG/ENTRIES/G00922.html)
Former Soviet Union Hydrological Snow Surveys has 10-day averages of daily snow depth at WMO stations. These data are nominally averages of the first 10 days of the month, the second ten days, and the remaining days. However, observations were not made on every day, and the distribution and number of the days on which observations were made in any one month varies significantly. To reduce bias that may result from uneven sampling, the average for a month was constructed by weighting the average snow depth for the three periods in the month by the number of days of data for that month. Thus, the average snow depth for a month with only 15 days of observations would get only half the weight of a month with 30 days of observations when the long-term monthly means were calculated. Long-term monthly means were computed for each station where a given month is represented in the data record in 70 percent or more of the years from 1966 through 1982.
Former Soviet Union Hydrological Snow Surveys snow data are from the Institute of Geography, Russian Academy of Sciences, Moscow, and were compiled by A. Krenke. The data set is available from NSIDC, (http://nsidc.org/NSIDC/CATALOG/ENTRIES/G01170.html) where the data underwent quality control procedures and questionable observations were flagged. For the Atlas snow fields, only "good" (non-flagged) data were used.
Mean monthly snow-depth data from Russian coastal stations were provided by AARI. These are included in data files on this Atlas. Fifteen of these 65 coastal stations are represented in Former Soviet Union Hydrological Snow Surveys. Long-term monthly means were computed for each station where a given month is represented in the data record in 70 percent or more of the years from 1966 through 1982. (Data from the 50 stations not represented in Former Soviet Union Hydrological Snow Surveys were used).
End-of-month snow depth data for locations in Alaska were obtained from the United States Department of Agriculture (USDA) National Resource Conservation Service (NRCS) with the assistance of R. McClure, NRCS Anchorage Field Office. The NRCS Alaska home page (http://www.ak.nrcs.usda.gov/) has information on the NRCS Alaska snow survey program. Snow course data are taken with the Standard Federal Snow Sampler according to the method described in USDA Agriculture Handbook Number 169. Most snow courses consist of five to 10 sample points. Individual measurements are averaged to give one value for the course. Measurements are generally taken at the end of the month from January through April, although measurements may be taken mid-month at some locations. Long-term monthly means were computed for each station where a given month is represented in the data record in 70 percent or more of the years from 1966 through 1982.
End-of-month snow depth data for 463 NOAA National Weather Service stations in Alaska were obtained from the Alaska Climate Research Center of the Geophysical Institute, University of Alaska, Fairbanks (UAF), with the assistance of G. Wendler and C. Rowher at UAF, and T. Zhang at NSIDC. Long-term monthly means were computed for each station where a given month is represented in the data record in 70 percent or more of the years from 1966 through 1982.
In all, the gridded snow depth fields are drawn from a total of 782 stations: the 197 stations north of 50° North in Former Soviet Union Hydrological Snow Surveys; the 35 stations north of 50° North from Snowfall and Snow Depth for Canada, 1943-1982, from 37 snow course locations from NRCS, from 463 stations from NWS, and from 50 Russian coastal stations. The years of overlap for these data sources are 1966 through 1982. Station locations are shown in Figure 5.
Fig. 5. Snow course locations used in the land snow depth fields. Green crosses mark stations from Former Soviet Union Hydrological Snow Surveys, black squares mark locations from Snowfall and Snow Depth for Canada, 1943-1982, yellow crosses mark NRCS snow courses, red crosses mark NWS data locations, and blue squares mark the 50 Russian coastal stations. All stations north of 55° North are shown, although not all lie within the Atlas EASE-Grid.
Ocean
Warren et al. [1999] fit a two-dimensional quadratic function to all NP snow data for a particular month, irrespective of year. (Usually, only two stations were operating in any given year.) Tables 1 and 2 in Warren et al. Warren et al. give the coefficients for this fit to a rectangular grid for snow depth and snow water equivalent. To obtain snow depth and snow water equivalent for the Atlas EASE-Grid, the coefficients were used to find the snow depth at the EASE-Grid centers. The RMS error of the quadratic fit is shown in Warren et al. [1999].
Because NP stations were established on multiyear ice, the fields are only valid for the area of the Arctic Ocean covered by the perennial ice pack. The average ice extent for the month of September, which is generally the month of minimum ice extent and thus a reasonable proxy for mean multiyear extent, was used to mask the area outside the mean perennial ice pack. Average ice extent for September was calculated using data from 1979 through 1995 from the NSIDC data set Northern Hemisphere EASE- Grid Weekly Snow Cover and Sea Ice Extent (Figure 6).
Fig. 6. "Average" ice extent for September calculated using ice concentration data from satellite passive microwave brightness temperatures. The image is constructed by averaging ice concentration for 1979 through 1995, and then counting any grid cell with an average of more than 15 percent concentration as "ice-covered". This method yields a larger extent than would a method that calculates an average ice edge position.
Land
Cressman interpolation [Cressman, 1959] was used to grid irregularly spaced land station data to the Atlas EASE-Grid. In Cressman interpolation the value at a point (the EASE-Grid cell center) is determined by the value of the nearby surrounding observations weighted by the distance of the observation station from the cell center:
where xi is the observation at station i within a radius R of the grid cell center, and n is the number of stations found within radius R. Weights wi are given by
,
where d is the distance from the grid cell center to the station. If less than three stations were found within radius R, the grid cell value was set to missing.
Long-term monthly means from the land data sources described above were interpolated. Data from stations north of 50° North, and for 1966 through 1982 (the period of overlap for the four data sources) were used. Figure 5 shows the location of the stations. A radius of 375 km around each EASE-Grid cell center was used. (In comparison, the average spacing between Canadian stations north of 60° North is 286 km, and the average spacing between Russian stations north of 60° North is 127 km). The radius used for Cressman interpolation determines the extent to which the station data are spatially smoothed.
Snow field products are:
The browse versions of these files are shown as .gif format images with a color bar and contours. These browse files are for the purpose of quickly visualizing the content of the corresponding ASCII data files. The IDL routine used to color-map the images gives a smooth and visually pleasing result, but keep in mind that the gridded ASCII files have a value only for every grid cell. The grid cell centers are shown as red dots. For information on the structure of the gridded files, see the section on "EASE-Grid."