56th Annual Eastern Snow Conference
Fredericton, N.B., Canada
June 2-4, 1999
Program and Abstracts

The structure of snowflakes as revealed from low temperature scanning electron microscopy (refer to W.P. Wergin, A. Rango and E.F. Erbe, 1996:The strcture and metamorphism of snow crystals as revealed by low temperature scanning electron microscopy. Proc. 53rd Eastern Snow Conference, May 2-3, 1996, Williamsburg, Virginia, 195-204.)


Wednesday, June 2

18:30 Ice Breaker and Registration
19:30 ESC Executive Pre-Conference Meeting

Thursday, June 3

07:00 – 08:00 ESC Breakfast
08:15 Welcome, Announcements, Business Meeting

08:40 1. Keynote Address: Dr. Monique Bernier, Pre-Operational Determination of Snow Water Equivalent (SWE) using RADARSAT Data. M. Bernier, J.-P. Fortin, Y. Gauthier, R. Gauthier, R. Roy and P. Vincent.

09:10 2. C.R. Duguay, T.J. Pulz, D. Drai and P.M. Lafleur, Spatial and Temporal Variations in RADARSAT Backscatter from Subarctic Lake Ice (Churchill, Manitoba).

09:30 3. Jonathan S. Barton, Dorothy K. Hall, Oddur Sigurdsson, Richard S. Williams, Jr., Laurence C. Smith and James B. Garvin, Calculation and Analysis of an interferometric digital elevation model of Hofsjökull, Iceland.

09:50 4. S.R. Fassnacht, E.D. Soulis and N. Kouwen, Modelling the Hydrology of a Southern Ontario Snowpack from Weather Radar Precipitation.

10:10 Coffee break

10:30 5. Peter B. Roohr and Thomas H. Vonder Haar, An Analysis of the Incorporation of Lightning into the Nowcasting of Enhanced Frozen Precipitation.

10:50 6. Richard R. Heim, Jr. and Robert J. Leffler, Cooperative Station Snow Climatologies.

11:10 7. Richard R. Heim and James R. Angel, Urbana, Illinois: 20th Century Snowfall Variations and Non-Climatic Influences.

11:30 8. Eva Mekis and Zuebin Zhang, Computation of Canadian Gridded Snowfall Based on Rehabilitated Station Data.

11:50 2-Minute Poster Presentations

12:20 Lunch

13:40 9. J. Smyth and K. Goita, Statistical Analysis of the Impact of Temperature and Vegetation Cover on Snow Water Equivalent using SSM/I data over New Brunswick.

14:00 10. Anne E. Walker and John R. Metcalfe, Investigation of Seasonal Snow Cover Variations in Eastern Ontario using Airborne Microwave Radiometry.

14:20 11. Andrew G. Klein and Dorothy K. Hall, Snow albedo determination using the NASA MODIS Instrument.

14:40 12. Peter Romanov, Ivan Csiszar and Garik Gutman, Towards Automated Multispectral Snow Mapping.

15:00 13. S. Brettschneider, M. Farzaneh, K.D. Srivastava and S.Y. Li, Study of Ice Surface Breakdown using Ultra-High Speed Photography.

15:20 Coffee Break - Posters

15:40 14. David N. Collins, Solute Fluxes in Meltwaters Draining from Mountain Glaciers.

16:00 15. L. Kitaev, A. Krenke and E. Barabanova, Interaction between Snow Cover, Climate and Hydrological Processes in the North of Russia – Information Support Peculiarities.

16:20 16. Patrick Tang, Roger Price and Mike Howe, The Saint John River Forecast System – an Integrated Approach.

16:40 17. William Richards and Christian Martin, Snow Cover Climate of the Maritime Provinces of Canada.

17:00 Adjourn

18:00 Happy Hour - Hotel

19:00 Banquet - Hotel

Friday, June 4

07:30 Executive Meeting

08:10 Student paper contest winner presentation

08:30 18. Robert A. Hellstrom, Representation of Forest Cover in a Physically-Based Snowmelt Model.

08:50 19. Mary Albert, Gary Koh and Frank Perron, Investigations of Preferential Melt Pathways in a Natural Snow Pack.

09:10 20. Susan Taylor, X. Feng, R. Osterhuber, J.W. Kirchner, B. Klaue and C. Renshaw, The Isotopic Evolution of Snow and its Melt.

09:30 21. F. Yusuf, N. Kouwen and S.R. Fassnacht, Flood forecasting the 1999 Snowmelt in the Grand River Basin using WATFLOOD.

09:50 22. Carl Egede Bøggild and Mogens Brems Knudsen, Tracer-Tests of Water Drainage from Cold Snow.

10:10 Coffee Break - Posters 10:30 23 H.G. Jones, J.W. Pomeroy, T.D. Davies, M. Tranter and P. Marsh, CO2 in Arctic Snow Covers: Gaseous Concentrations and Landscape Form.

10:50 24. C. Derksen, E. LeDrew and B. Goodison, Associations between the Principal Spatial Modes of North American Prairie Snow Water Equivalent and Low-Frequency Atmospheric Teleconnection Patterns.

11:10 25. Robert E. Davis, Aspects of Snow Extent and Water Equivalent Pertaining to Hydrologic Modeling in Complex Landscapes.

11:30 Adjourn 13:30 Leave for field trip to Mactaquac Dam (2 hours maximum).

Poster Session

David G. Barber and J. Hanesiak, Integration of Remote Sensing Measurements and Numerical Modelling of Snow on Sea Ice.

Neftali S. Cajina, Kaye L. Brubaker and Al Rango, Implementing the Martinec/Rango Snowmelt Runoff Model in the USGS Modular Modeling System.

J.G. Cogley, W.P. Adams and M.A. Ecclestone, Glaciological Indices of Climate Change, Axel Heiberg Island, Canadian High Arctic.

Jeffrey A. Cole, Wind Shield Evaluation at NCAR/Marshall Test Site in Boulder, CO during 1998-1999 Winter.

Danielle de Sève, M.Bernier, J-P. Fortin, A. Walker, Analysis of Microwave Radiometry of Snow Cover with SSM/I Data in a Taïga Area : The Case of James Bay Area.

Claude R. Duguay and Peter M. Lafleur, Determining Depth and Ice Thickness of Shallow Subarctic Lakes and Ponds using Spaceborne Optical and SAR.

Claude R. Duguay, Yannick Ernou and Jim Hawkings, SAR and Optical Satellite Observations of Ice-Covered Thermokarst Lakes, Old Crow Flats, Yukon Territory.

Deborah Drai, Claude R. Duguay and Frederique Pivot, Use of Georadar in Support of Satellite Remote Sensing Investigations of Lake Ice.

S.R. Fassnacht, F.R. Seglenieks, E.D. Soulis and N. Kouwen, A Shape Function Estimate for Fresh Dendritic Snowflakes.

Guillaume Fortin, Hardy Granberg and Jean-Marie Dubois, Snow Monitoring at the Kouchibouguac National Park, New Brunswick (1974-1998).

Dorothy K. Hall, Milan Allen and Andrew B. Tait, Snow Mapping through Dense Forests using Multiple Satellite Data Sets.

Janet P. Hardy, Rachel Jordan, Peter Groffman, Scott Nolin, Timothy Fahey, Charles Driscoll and Ross Fitzhugh, Snow Depth and Soil Frost Modelling in a Northern Hardwood Forest.

Suzanne Hartley, Atlantic Sea Surface Temperatures and Winter Snowfall along the Southern Margin of the Eastern United States Snowbelt.

Jane A. Hollingsworth and Julie L. Adolphson, The Integration of Digital GOES-8 Satellite Imagery with Observational and Prognostic Datasets during the Winter Storm of January 1999.

L. Kitaev and V. Zakharov, Climate, Snow and Glaciers Temporal Variability in High Latitudes of the North and South Hemispheres (Eruasian Sector and the East Antarctic Seashore).

George A. Riggs and Dorothy K. Hall, What’s in the MODIS Snow Data Products?


Pre-Operational Determination of Snow Water Equivalent (SWE) using RADARSAT Data

Monique Bernier 1, Jean-Pierre Fortin 1, Yves Gauthier 1, Raymond Gauthier 2, René Roy 2 et Pierre Vincent 3

1 INRS-Eau
2800 rue Einstein, CP 7500, St. Foy (Québec), Canada G1V 4C7
Tel: (418) 654-2585 Fax: (418) 654-2600
E-Mail: Monique_Bernier@inrs-eau.uquebec.ca

2 Hydro-Québec, Groupe Production
75 ouest boul. René-Levesque, 9 étage, Montréal, Québec H2Z 1A4
Tel: (514) 289-2211 poste 4062
E-Mail: Gauthier.Raymond@hydro.qc.ca

3 VIASAT Géotechnologies Inc.
419, boul. Rosemont, bureau 301,Montréal, Québec H2S 1Z2
Tel: (514) 495-6500 Fax : (418) 495-4191
E-Mail: pvincent@viasat.qc.ca


Within the ADRO program of the Canadian Space Agency (1996-1998), we have studied the opportunity to use RADARSAT-1 data (C band-HH) for snow monitoring in the La Grande River watershed, a large hydroelectric complex in northern Quebec (184 000 km2). An empirical algorithm has been developed to estimate the spatial distribution of the Snow Water Equivalent (SWE) using data from Standard beams and some field measurements. Calibrated RADARSAT images have been acquired in Standard beam modes S1 (20-27° ) and S7 (45-49° ) as well as in ScanSAR mode (not calibrated) in conjunction with field campaigns (snow lines, snow pits, air and soil temperature, etc.) in November, February, and March (1996-1998).

Considering that our approach for the SWE estimation is based on the temporal changes of the dielectric characteristics of the soil related to the snow thermal resistivity, the S1 mode has revealed to be more appropriated. The RADARSAT algorithm for the SWE estimation is similar in form to the one developed previously for ERS-1 data, but the parameters (m,b) of the regression between the thermal resistivity of the snow pack and the ratio of the backscattering coefficient of a winter scene over a fall scene have been adjusted. The RADARSAT estimation of the mean SWE on the experimental sites are comparable to the values measured on site. The standard deviation on the estimated values (+/- 17 to 26 mm) are also comparable to the field measurements standard deviation (+/- 15 to 24 mm).

The end objective of this study being the operational use of RADARSAT data within the Hydro-Quebec hydrological forecasting system, the algorithms and procedures have been implemented within a MapInfoTM application. This EQeau prototype, jointly developed by Viasat Géo-technologie Inc. and INRS-Eau, allows the mapping of the spatial distribution of the estimated SWE using Standard (1) or Wide (1) beams. It can also resample your map to a wider pixel or calculate the mean SWE on a sub-watershed. The demonstration of the operational feasibility and economical advantages of this RADARSAT approach will start in winter 1999, as part of the Earth Observation Pilot Project Program of the Canadian Space Agency, which is administrated by the Canadian Center for Remote Sensing.


Spatial and temporal variations in RADARSAT backscatter from subartic lake Ice

(Churchill, Manitoba)


Claude R. Duguay1, Terry J. Pultz2, Déborah Drai1, and Peter M. Lafleur3


1Centre d’etudes nordiques, Universite Laval

Sainte-Foy, Quebec G1K 7P4



2Canada Centre for Remote Sensing

588 Booth Street

Ottawa, Ontario K1A 0Y7



3Department of Geography, Trent University

Peterborough, Ontario K9J 7B8





Results from an investigation on the use of RADARSAT (C-HH) Standard beam mode imagery for monitoring ice growth and decay, and related processes of shallow subarctic (tundra and forest) lakes in northern Manitoba, Canada, will be presented. Field observations on the structural and stratigraphic characteristics of snow and ice from four lakes acquired during three field campaigns (February, March, and May 1998), which coincided with RADARSAT overflight dates, were used in the interpretation of the temporal evolution of backscatter, as well as the variations in backscatter intensity with incidence angle (S1 to S7).


Results of this study show that: 1) bubble inclusions within the ice volume, most ofwhich are tubular and oriented in the direction of growth, significantly influence backscatter intensity in RADARSAT images, but can vary considerably as a function of incidence angle; 2) differences of as much as 6.5 dB can be observed for the same ice cover when observed at steeper (20o-35o) compared to shallower (35o-49o) incidence angles; 3) during the early stages of ice growth and when the ice volume contains a small amount of tubular bubbles, backscatter intensity from the floating ice measured at shallower incidence angles (S4-S7 beam positions) is similar to that observed from the grounded ice (-12 to -16 dB), and; 4) during spring thaw, the strong decrease in backscatter can be explained by the microwave signal being absorbed by the wet snow cover and by specular reflection from the standing water (ponds) on the lake ice surface.


Calculation and Analysis of an Interferometeric Digital Elevation Model of Hofsjökull, Iceland


Jonathan S. Barton1,2

Dorothy K. Hall2

Oddur Sigurðsson3

Richard S. Williams, Jr.4

Laurence C. Smith5

James B. Garvin6


1 General Sciences Corporation, 6100 Chevy Chase Drive, Laurel, MD 20707-2929

ph: (301) 286-4738, fx: (301) 286-1758, jbarton@glacier.gsfc.nasa.gov

2 NASA/Goddard Space Flight Center, Hydrological Sciences Branch, Code 974,

Greenbelt, MD 20771

3 National Energy Authority, Reykjavík, Iceland

4 USGS/Woods Hole Field Center, Woods Hole, MA 02543-1598

5 Department of Geography, University of California at Los Angeles, Los Angeles, CA 90095-1524

6 NASA/Goddard Space Flight Center, Geodynamics Branch, Code 921,

Greenbelt, MD 20771



Two ascending tandem pairs of SAR (C-band, VV polarized) images of Hofsjökull, Iceland, taken from the European Space Agency’s Earth Resources Satellites (ERS) 1 and 2 during the winter of 1995–1996, are used to create a digital elevation model (DEM) of the ice cap. The two pairs were taken on 2 and 3 January, 1995 ("January pair"), and on 6 and 7 February, 1995 ("February pair"), respectively. The January pair has a perpendicular baseline of 88 meters, and the February pair has a baseline of 208.8 meters.

The January pair is used to calculate a motion field for the ice cap, using the U.S. Geological Survey’s GTOPO30 30 arc-second DEM. First, a simulated SAR amplitude image is calculated from a subsection of the GTOPO30 DEM. This simulated amplitude image allows the image to be automatically registered and projected into the SAR-coordinate space. The phase is then simulated from the topography, and a simulated topography-only interferogram is produced, with the same baseline as the January pair. This topographic phase is then subtracted from the topography+motion contained in the January pair, yielding a motion-only interferogram.

The February pair is then used to calculate a refined topography for the ice cap. Because the phase effects of motion are independent of baseline, and assuming that the daily motion over the glacier does not change significantly between January and February, the phase of the motion-only interferogram from the previous step can be subtracted from the February interferogram to produce a topography-only interferogram. This interferogram is then unwrapped, its baseline refined using ground control points from 1:50,000-scale (Series C761) maps of the area, and the elevations are calculated from the phase information. In order to produce a full-coverage map of the surface, in areas where decorrelation is severe, the GTOPO30 DEM is used as a first-order approximation of the glacier surface. A transect across this composite DEM is then compared with a west-to-east geodetic airborne laser-altimetry profile across the glacier. This profile, acquired on May 7, 1997, has a root-mean-square vertical accuracy of 10 cm. In order to eliminate any east-west trending ramps and residual global errors resulting from orbital inaccuracies, the DEM is then warped to the laser track.

Modelling the hydrology of a southern Ontario snowpack from weather radar precipitation


S.R. Fassnacht, E.D. Soulis & N. Kouwen

Department of Civil Engineering, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1




While snowfall in southern Ontario can occur as early as October and as late May, snowcover may not persist over the entire winter. Winter air temperature often fluctuate around freezing and melting temperatures may continue for a week or longer. As a result snowcover can develop and ablate more than once during a winter in southern Ontario. To model the snow hydrology of this environment, especially for continuous simulation of streamflow, requires adequate meteorological data. These data are recorded at climate stations that are distributed throughout the area, however the variability in precipitation may not be represented by such gauges. Fortunately central southern Ontario is within the coverage of the King City radar, operated by the Atmospheric Environment Service.

The objective of this paper is to illustrate the usefulness of weather radar to estimate snowfall accumulation for the modelling of the snowpack, runoff, and streamflow using a distributed hydrologic model. The model is a linkage of the WATFLOOD hydrologic model, to perform the horizontal water routing, and the Canadian Land Surface Scheme for the vertical water budgetting. To meet the objectives, algorithms have been developed to improve the accuracy of radar winter precipitation measurement, various datasets have been amalgamated, the importance of various snowpack processes has been assessed, and several additional processes have been incorporated into the model. The additions to the model include redistribution, the partitioning of snowcovered and bare areas to improve energy and mass transfer between the two areas, variable initial snowpack densities, canopy interception for snowfall, and variable snowpack densification routines to consider landcover.

The results are presented in the form of computed versus observed SWE, snow depth, soil parameters and streamflow for five winter season (1993 through 1997) across the four basins in southern Ontario (Grand, Maitland, Saugeen and Thames River). The computed parameters match the observed data significantly better as the radar precipitation estimates are improved.

An Analysis of the Incorporation of Lightning into the Nowcasting of Enhanced Frozen Precipitation


Peter B. Roohr* and Thomas H. Vonder Haar

Department of Atmospheric Science, Colorado State University

Cooperative Institute for Research in the Atmosphere

Fort Collins CO 80523-5001

Phone: (402) 294-1690, E-mail: peter.roohr@afwa.af.mil


The major goal of this research is to investigate the potential of using cloud-to-ground (CG) lightning data in the nowcasting of winter storms. Incorporation in this context means comparing lightning data with other data sets just like a forecaster would do on an interactive workstation at an NWS Forecast Office, at the Storm Prediction Center (SPC), at an AF Weather Regional Hub such as the one at Sembach Airbase in Germany, or at some other location.

The first three objectives focus on the relationship of lightning data to satellite imagery, radar data, and model output parameters (respectively) in nowcasting (within 500 km and 6 hours) of heavy frozen precipitation. Hypotheses regarding these first three objectives are: (1) Lightning will occur in areas of satellite detected clouds where vertical motions perceived by satellite-derived radiation can be visualized; (2) lightning can detect areas of high upward vertical motion related to large radar reflectivity returns in winter storms, and lightning should exist slightly upstream of the radar echoes; and (3) model output would help locate features such as moisture flow, the advection of warm theta-e air, areas where elevated or slantwise convection may develop, etc. The final objective involves the compilation of all the results of the case study analyses and a determination of if and when CG lightning data can be used in conjunction with other data sources to nowcast enhanced frozen precipitation.

Major findings of this research are: (1) Lightning activity (as defined by 10x10 km bins) often occurs on the most intense gradient of cloud top temperature fields, 100-200 km equatorward of the coldest cloud area; (2) Bursts/progression of lightning activity identify oncoming heavy snow and ice more efficiently and faster (i.e., with 1-2 hours more lead time) than NEXRAD data; (3) Lightning activity (observed via the progression of lightning packets within a high qE tongue) (a) improves the identification of the enhanced moisture flux into winter storms, and (b) enhances the forecaster interpretation of model-derived strong vertical motion fields (more than 7 mbars/sec) within 500 km and 6 hours upstream of heavy snow/ice; (4) Lightning activity not only identifies the strength and location of strong low-level moisture flow (as defined by a tongue of enhanced low-level qE and southerly winds exceeding 35-40 kt) into a winter storm, but also depicts strong lifting along tight low-level qE gradients where lifting may be in doubt; and (5) Lightning provides a more valuable nowcasting tool for winter type precipitation as one proceeds from the East Coast to the Western Plains.


* Currently located at Air Force Weather Agency, Offutt AFB NE 68113-4039

Cooperative Station Snow Climatologies

Richard R. Heim Jr.

USDOC/NOAA/NESDIS/National Climatic Data Center

Asheville, NC

Robert J. Leffler

USDOC/NOAA/National Weather Service

Silver Spring, MD


The introduction, by the National Oceanic and Atmospheric Administration's (NOAA) National Weather Service (NWS) in the early 1990's, of the Automated Surface Observing System (ASOS) has improved meteorological observations by standardizing observational methodologies across the airport observing network and by creating the potential for increasing the number and geographical coverage of aviation stations.

However, ASOS instrumentation does not measure snowfall and snow depth amounts. The nation will therefore now have to rely more heavily on the daily snowfall and snow depth observations of the voluntary Cooperative Network stations.

Comprehensive snow climatologies were generated for 5525 stations in the Cooperative Network (in the contiguous United States and Alaska) to support NWS operations in the ASOS era and to enable NOAA to better respond to user requests for snow information for use in economic and engineering decision-making. This paper summarizes the data base, quality assurance, and methodology used to create the climatologies. The statistics produced here represent the most comprehensive snow climatology ever computed for the United States.

Daily snowfall and snow depth data from the National Climatic Data Center's (NCDC) TD-3200 data base were analyzed over the digital period of record through 1996. Three levels of quality control were applied. Several statistics were computed for several climatic elements using several snowfall and snow depth thresholds. The statistics include mean, median, first and third quartiles, extremes, and probabilities. The elements include number of days with snow (snowfall or snow depth) beyond various thresholds, monthly and seasonal total snowfall, number of consecutive days with snow, dates of the first and last occurrence of snowfall, daily and multiple-day extreme snowfall amounts, and daily snow depth amount.

Several indicators were computed, based on the data and metadata, to enable the user to assess the quality of the stations. These include frequencies of station moves and ob time changes, number of missing values and breaks in the record, number of values failing the QC checks, and percentages indicating how complete the data record is.

Urbana, Illinois: 20th Century Snowfall Variations

and Non-Climatic Influences


Richard R. Heim Jr.

USDOC/NOAA/NESDIS/National Climatic Data Center

Asheville, NC

James R. Angel

Illinois State Climatologist, Illinois State Water Survey

Champaign, IL


Snowfall is a phenomenon whose characteristics and impact vary considerably across the United States. Snowstorms can disrupt transportation, cause extensive damage and loss of life, and snow removal costs can break city budgets. Lack of snow in normally snowy climes can bring economic hardship to the recreation and water management industries. Snow is also an important player in the climate picture, both reflecting climatic changes and fluctuations as well as exerting an influence on climate. It is clear that a better understanding of the history of snowfall will benefit this nation, for both climatic and economic reasons.

The 20th Century snowfall record for Urbana, Illinois, is examined in this paper as a case study to illustrate both the historical variations of snowfall and the importance of non-climatic influences. Daily snowfall data from the National Climatic Data Center’s (NCDC) TD-3200 data base formed the basis for several snowfall parameters, including seasonal total snowfall amount, daily extreme snowfall amount, length of the snow season, and number of days with snowfall. On a broad canvas, the data suggest that the first two decades had more snowfall when compared to the 1920's-40's. Snowfall amount increased in the 1950's-60's, with record amounts falling in the late 1970's. The 1980's-90's have had less snowfall, which is consistent with the 1972-present satellite snow cover record.

Non-climatic influences, such as time of observation, local topography (buildings and vegetation obstructing wind flow), and urbanization (urban heat island), can impact the amount of snowfall measured. A more robust parameter is the number of days with a trace or more of snowfall. This parameter suggests an increasing trend from the early 1900's to the late 1970's, with a marked drop off beginning in the 1980's. A related parameter, the length of the snowfall season, exhibits large interannual variability at the beginning and end of the century when compared to the years in between.

This paper will evaluate the Urbana record in light of other snow studies and with an eye towards station history.


Computation of Canadian Gridded Snowfall Based on Rehabilitated Station Data


Eva Mekis and Xuebin Zhang

Atmospheric Environment Service, Climate Research Branch

4905 Dufferin Street, Downsview, ON M3H 5T4

Eva.Mekis@ec.gc.ca and Xuebin.Zhang@ec.gc.ca




Snowfall variability maps can be used for several purposes (climate change detection, hydrological, regional climate model validation purposes, etc.), but gridded monthly or seasonal snow derivation is very difficult because of the uneven distribution of the snow within the year. The indirect approach presented here for creation of gridded snowfall is based on the new rehabilitated station precipitation data, the gridded precipitation normals computed at University of Waterloo (Soulis et al., 1994) are also used.

The first part of the presentation will summarize the derivation of rehabilitated daily snow and rain data. A total of 491 carefully selected stations, most covering the period 1900-present were used. Data availability in much of the Canadian Arctic is restricted to 1948-present. By using a daily time-interval, improved corrections to rain and snow data could be implemented. For each of the three rain gauge types, corrections to account for wind undercatch, wetting loss and evaporation were applied. For snowfall, ruler measurements were used throughout the time series, to minimize potential discontinuities introduced by the adoption of Nipher shielded snow gauge measurements in the mid-1960’s. Density corrections based upon coincident ruler and Nipher measurements were applied to all ruler measurements. Where necessary, records from neighboring stations were joined employing a technique based on a simple ratio of observations. Annual and seasonal graphs of national and north of 55oN time series will be presented.

The second part of the presentation will summarize the snow grid map derivation and evaluate the efficiency and applicability of the method. Spring as a shoulder season was selected for the first snow grid map. The direct statistical computation of the snow grid network is nearly impossible due to the occasional occurrence of zero events during the year. However seasonal precipitation anomalies can easily be calculated and gridded using Gandin statistical optimal interpolation based on the rehabilitated station information (Milewska and Hogg, 1998). Gridded seasonal precipitation normals are also available, the normals are computed on the same set of rehabilitated precipitation data including additional important factors (such as latitude, elevation, slope, etc.). The actual gridded seasonal total precipitation values can be derived by joining the two maps together. The next step is the station and gridded seasonal snow to total precipitation ratio computation. The actual gridded seasonal snow field is the multiplication of the actual seasonal total precipitation and the seasonal snow to total precipitation ratio grid values. The quality of gridded snowfall is cross-validated on a selected subset of evenly distributed stations.

Milewska, E., and H. D. Hogg, 1998: Spatial representativeness of a climate rain gauge network. Atmospheric-Ocean, (submitted).

Soulis, E.D., S.I. Solomon, M. Lee, and N. Kouwen, 1994: Changes to the distribution of monthly and annual runoff in the Mackenzie Basin under climate change using a modified square grid approach. Proceedings of the sixth biennial AES/DIAND meeting on northern climate and mid study workshop of the Mackenzie basin impact study, Yellowknow, Northwest Territories, April 10-14, 1994, pp 197-209. Environment Canada.

Statistical analysis of the impact of temperature and vegetation cover on snow water equivalent estimation using SSM/I data over New Brunswick

J. Smyth and K. Goita*

École de sciences forestières, Université de Moncton, Campus d’Edmundston, 165 Boulevard Hébert
Edmundston, Nouveau-Brunswick, Canada, E3V 2S8
Phone : 506 737 5246, Fax : 506 737 5373, Email : kgoita@cuslm.ca


The objective of this study is to understand the influence of vegetation cover and temperature variations on the estimation of snow water equivalent using passive microwave data in deep snow and high vegetation density conditions. The study region considered is New Brunswick over which 18 meteorological stations with snow depth and temperature measurements were selected. SSM/I data from 1988 to 1991 in Ease-grid format are used in the study. Forest cover information for the selected grid cells were derived from AVHRR and Landsat classification maps and from digital forest cover maps in ARC/INFO format. A statistical analysis was conducted to understand the spatial and temporal variability of snow depth using the selected meteorological stations data. To study the combined effects of forest cover and temperature, stable periods of snow depth conditions with varying temperature were considered over the selected grid cells. The statistical analysis is aimed to understand if significant differences exist in the spatial and temporal variability of SSM/I brightness temperature difference index (37V-19V and 37H-19H) in function of vegetation type, crown cover and height and temperature conditions.

(* for correspondence)

Investigation of Seasonal Snow Cover Variations in Eastern Ontario using Airborne Microwave Radiometry


Anne E. Walker and John R. Metcalfe

Climate Research Branch, Atmospheric Environment Service

4905 Dufferin Street, Downsview, Ontario, Canada, M3H 5T4

Ph: (416) 739-4357, Fax: (416) 739-5700, E-mail: anne.walker@ec.gc.ca



One of the research programs in the Climate Research Branch of the Canadian Atmospheric Environment Service (AES) focusses on development and use of passive microwave data to derive snow cover information (extent, water equivalent, wet/dry state) for regions of Canada in support of cold climate processes research and environmental monitoring and prediction activities. After many years of testing snow water equivalent (SWE) algorithms in the prairie region of western Canada, it became obvious that the variation in physical snowpack properties throughout a season and from season to season have a significant impact on the representativeness of SWE retrievals derived from a passive microwave algorithm. In eastern Canada, especially southern and eastern Ontario, seasonal snow cover is highly variable and the effect of physical snowpack variations must somehow be taken into account when developing a passive microwave satellite SWE algorithm for this region.

Since 1996, airborne microwave radiometers have been flown several times each winter along established agricultural and forest lines east of Ottawa, Ontario for the purpose of investigating the microwave emission behavior related to seasonal snowpack variations and forest cover. The airborne measurement system consists of three dual-polarization radiometers operating at 19, 37 and 85 GHz frequencies and viewing the ground at a 53° incidence angle, which are characteristics that match the current SSM/I (Special Sensor Microwave Imager) satellite radiometer. Additional airborne measurements include surface temperature and video of ground characteristics (land use, spatial snow cover variations) along the radiometer flight path. Coincident with each flight, ground survey teams acquire measurements along the flight lines to document snow cover and underlying soil characteristics, including snow depth, water equivalent, grain size, stratification, soil moisture, and temperature.

This paper will present results from this investigation based on more than 20 microwave radiometer flights that have occurred over 4 winter seasons and captured a wide variety of snow cover conditions, including the impact of the January 1998 ice storm. Comparisons between forested and non-vegetated agricultural lines will also be presented to demonstrate the influence of vegetation on the retrieval of snow cover information for the underlying snowpack. Finally, the potential contribution of these airborne data sets to the development of a robust satellite SWE algorithm for this region will be discussed.


Snow albedo determination using the NASA MODIS instrument


Andrew G. Klein, Department of Geography, Texas A & M University, College Station, TX 77843-3147, tel: 409-845-5219 fax: 409-862-4487 email: klein@geog.tamu.edu


Dorothy K. Hall, Hydrological Sciences Branch, NASA/Goddard Space Flight Center, Greenbelt, MD 20771, tel: 301-614-5771 fax: 301-614-5808 email: dhall@glacier.gsfc.nasa.gov




In 1999, the Moderate Resolution Imaging Spectroradiometer (MODIS) is scheduled for launch aboard the first NASA Earth Observing System platform (EOS AM-1). This instrument will provide a global daily view of the Earth's seasonally snow-covered areas. Algorithms have already been developed to produce daily maps snow-cover extent with a spatial resolution of 500m from MODIS data acquired in the visible to shortwave-infrared regions of the electromagnetic spectra. Development efforts are now underway to use this spectral information to determine the albedo of these snow-covered surfaces.

Current MODIS algorithms determine albedo by constructing bidirectional reflectance distribution functions (BRDF) from multiple observations taken over a 16-day period by both MODIS and MISR, the Multi-Angle Imaging Spectroradiometer that also will be flown on EOS AM -1. However, for snow-covered surfaces more frequent albedo measurements are desired. Improved temporal sampling will be accomplished via a simpler algorithm that assumes a BDRF to convert MODIS at-satellite radiances into surface albedo. Over forests the algorithm will utilize a simple isotropic scattering model. For surfaces with no or low-stature vegetation, such as cropland, tundra and ice sheets, the algorithm will account for the strongly forward scattering nature of snow reflectance. Topographic effects will also be considered. MODIS Airborne Simulator (MAS) images collected in February 1997 over the Midwestern United States during the Winter Cloud Experiment (WINCE) field campaign are being used to develop the prototype MODIS snow albedo algorithm.

Atmospheric corrections over snow remain a challenge and uncertainties in the effect of aerosols are expected to introduce an uncertainty in snow-albedo calculation of approximately 3%. Despite these limitations, a MODIS snow albedo product will represent a significant improvement over the current parameterization the snow albedo in Global Climate Models and Land-Atmosphere Transfer Schemes.


Towards Automated Multispectral Snow Mapping


Peter Romanov, Ivan Csiszar, and Garik Gutman

NOAA/NESDIS, Office of Research and Applications, Camp Springs, MD




Current daily and weekly snow products at NOAA are generated by analysts, who use visible imagery from geostationary and polar orbiting satellites. There is no automated satellite-derived product based on the visible and/or infrared data, however. The automated satellite-derived snow product based on SSMI and/or AMSU microwave observations is used by the analysts only for verification because of the coarse resolution and remaining ambiguities. The present study is based on GOES Imager, NOAA AVHRR and DMSP SSM/I data over North America during the 1998-1999 snow season. The data from all channels on two GOES imagers (East and West) were intercalibrated with NOAA-14 corresponding channels and updated on a bi-monthly basis. The synergy between SSM/I, GOES Imager, and NOAA AVHRR observations has been utilized by taking advantage of the ability to observe ground surface through clouds by SSM/I, frequent views from GOES throughout the day, and the high resolution full coverage of northern latitudes by AVHRR. The SSM/I daily 30-km spatial resolution automated snow maps are produced routinely at NESDIS. They have been used as the primary source for a daily snow map. The shortcomings of the SSM/I-derived maps, such as spatial resolution and snow detection over forests, have been alleviated by a combined use of visible, mid-IR and IR from GOES and AVHRR. Daily composites based on the maximum brightness temperature have been constructed from GOES and AVHRR. All pixels in a daily composite are then classified into snow, snow-free and cloudy. The visible (0.6 µm) reflectances are corrected for angular effects with a semi-empirical kernel driven model. The mid-infrared reflectances were derived from GOES (3.9 µm) and NOAA (3.7 µm) data by subtracting the thermal component from observed radiances. The mid-IR reflectances did not exhibit any significant angular variability. They are then combined with the visible reflectances to form a snow index that is resistant to data contamination by most cloud types. Daily composites of this index are produced over North America from the combined data of the two GOES Imagers and NOAA-14 AVHRR. To produce weekly snow maps, 7-day composites have been formed based on the maximum snow index, reducing cloud contamination even further and filling the data gaps in the daily composites. We conducted validation and comparison of the derived snow maps with reports from ground stations and the current operational snow analysis at NOAA.

Study of Ice Surface Breakdown Using Ultra-High Speed Photography


S. Brettschneider, M. Farzaneh

Industrial Chair on Atmospheric Icing of Power Network Equipment (CIGELE)

University of Québec in Chicoutimi, Canada


K.D. Srivastava, S.Y. Li

Department of Electrical and Computer Engineering

University of British Colombia, Canada


Insulator flashover caused by atmospheric ice accretion on power networks is one of the serious problems in cold regions of the world. Several power outages caused by this phenomenon have been reported from different countries. Although a relatively large number of investigations have been carried out on flashover performance of ice-covered insulators, very little has been reported on the fundamental aspects of arc propagation on ice surfaces.

Recently, with the creation of the Industrial Chair on Atmospheric Icing of Power Network Equipment (CIGELE) at the University of Quebec in Chicoutimi, fundamental studies on the subject have received considerable attention.

The main purpose of the present study is to further the understanding of ice surface breakdown. One of the most appropriate methods to study surface breakdown is ultra-high speed photographic observation. To the best knowledge of the authors, such a method has never been applied to ice surface discharges. A high-speed streak camera was used to record the discharge development when a lightning surge was applied. The effects of several parameters such as air temperature, freezing water conductivity, and voltage polarity on the 50 % lightning impulse breakdown voltage was investigated. An electrode configuration of two metallic hemispherical capped rods 12 mm in diameter, half submerged in ice, was used in this study.

Results showed that an increase in air temperature or in freezing water conductivity decreased significantly the breakdown voltage of the ice surface. The influence of voltage polarity on the breakdown voltage depended on the ambient air temperature and freezing water conductivity.

Analysis of the high-speed photographic observations revealed a significant decrease in the time to breakdown for freezing water conductivity and ambient air temperature increases. The discharge development was very similar for both cases. This suggests that increases in freezing water conductivity as well as ambient air temperature, creating a thin water film on the surface of the ice, have the same influence on the breakdown mechanism. In particular, the combination of both processes could cause critical situations for power network equipment leading to power failures.

Solute fluxes in meltwaters draining from mountain glaciers

David N. Collins

Alpine Glacier Project

Department of Geography

University of SalfordSalford Crescent

Manchester M5 4WTU.K.


Estimation of rates of contemporary denudation in glacierised mountain ranges is of considerable importance with respect to long term landscape development and to any impact chemical weathering of emerging orogenic belts may have had on the CO2 content of the atmosphere in the last 45 Myr. Stage and electrical conductivity (EC) of meltwaters draining from Batura Glacier, in the basin of the Hunza, a tributary of the Gilgit, the latter a confluent of the upper Indus in the Karakoram range, were continuously recorded between April and October 1990. Discharge was obtained by the velocity-area method. Samples of meltwater were collected throughout the range of EC, and the cations Ca2+, Mg2+, K+ and Na+ were determined by atomic absorption spectrophotometry. A similar measurement programme, in which discharge and EC were recorded and samples collected frequently, was undertaken between April and October in 1983 and 1991 on the Gornera which drains from Gornergletscher, Pennine Alps, Switzerland.

Meltwater cationic content through time was derived from a well-constrained calibration of EC with the sum of cations. Cationic flux was then obtained from the product of total cationic concentration and discharge. Since 90% of the discharge occurs between April and October, annual total sediment delivery and cationic denudation rates for the two glaciers can be estimated from this information. Batura glacier basin is 60% glacierised, and rests on some carbonate rocks, whereas the basin of Gornergletscher (83% glacierised) predominantly lies on metamorphic rocks. Annual cationic denudation rate for the Batura basin was 947 Meq, a rate of 2.59 Meq km-2 yr-1 for the glacierised area alone, five times that of the Gornera (0.478 Meqm-2 yr-1). Both are considerably higher than the average for the continents, and suggest that dissolution of sediment in mountain glacier meltwaters has a potentially important effect on drawdown of CO2 from the atmosphere.

The interaction between snow cover, climate and hydrological processes in the north of Russia - information support peculiarities

L.Kitaev, A.Krenke, E.Barabanova

Institute of Geography, Russian Academy of Sciences,

Russia, Moscow, 109017, Staromonetny per. 29,

Tel 7(095)9590032 Fax 7(095)9590033 e-mail climat@ipcom.ru


In terms of initial information the surface-based snow cover, weather and hydrological observations are used. The observation period is of a hundred years' duration. The data, obtained on the recent 30 years, seem to be the most representative. To analyse conjugate changes of parameters efficiently and with high-quality a special geoinformational technique was worked out. The database with information retrieval system allowing searching according to quantitative and spatial identifiers was generated.

A mechanism to reveal the interaction between climatic and hydrological parameters, and temporal and spatial snow cover changes was found. The regularities of spatial interaction between snow cover, climate characteristics and spring flood peculiarities were obtained.

The Saint John River Forecast System an Integrated Approach

Patrick Tang Roger Price and Mike Howe2



Since 1973, hydrologic models have been used to forecast floods along the Saint John River. The Saint John River Basin covers an area of 55,000 sq. km. of which 51% is in the Province of New Brunswick, 13% is in the Province of Quebec and 36% is in the State of Maine, U.S.A. There are three hydro-power generation plants on the main Saint John River, namely Grand Falls, Beechwood and Mactaquac. This paper describes the current operation and future development of an integrated flood forecast system for the Saint John River Basin. The basic component of the system is the U.S. Army Corps of Engineers’ Streamflow Synthesis and Reservoir Regulation (SSARR) model. The Simulated Open Channel Hydraulics (SOCH) model of the Tennessee Valley Authority and the Dynamic Wave Operational (DWOPER) model of the National Weather Service are also used.

The input data required for the SSARR model are snow pack information, climate data, streamflow data and weather forecasts. The snow pack data include snow water equivalent information obtained from snow survey and gamma ray snow surveys as well as the areal extent of snow-covered area from satellite imageries. The climate data consists of temperatures and precipitation amounts. Both the snow pack data and the climate data are pre-processed with a grid square analysis program before input into the SSARR model. The streamflow data are collected via satellite and landline transmission. Flows along the river system are analyzed using a spreadsheet for data validity and for estimating initial streamflow conditions of ungaged sub-basins. If necessary, estimates for streamflow under ice-covered conditions are made. Fictional reservoirs are set up to handle ice jams and releases. The spreadsheet is also used to facilitate the updating of the model parameters such as soil moisture index snow-covered area and snow water equivalent. Weather forecasts of temperatures and precipitation are required and are provided by the New Brunswick Weather Centre. An objective analysis program to generate the temperature forecast for each sub-basin uses the maximum and minimum temperature forecasts obtained from the weather bulletins. The Canadian Meteorological Centre (CMC) provides the six hourly quantitative precipitation forecast (qpf) for each sub-basin transmitted over the circuit as a special weather bulletin. The staff in the New Brunswick Weather Centre checks the forecast and modifies it if necessary before forwarding the data for input to the SSARR model.

The output of the SSARR model is a five-day streamflow forecast for Saint John River and its tributaries. The streamflow forecast is used in the SOCH model to determine the discharge from the hydro-power generating facility at Mactaquac. The forecast discharge from Mactaquac, the local runoff forecast from the SSARR model and tide predictions are used in the DWOPER model to forecast the water levels along the Lower Saint John River.

Future developments for the Saint John River forecasting system include the modelling of ice jams to determine the water profiles upstream of equilibrium jams and the modelling of a surge following release of an ice jam. The coupling of atmospheric and hydrologic models using radar data is also planned for the Saint John River Basin.


Snow Cover Climate of the Maritime Provinces of Canada

William Richards

Environment Canada, New Brunswick Weather Centre


Christian Martin

Memorial University of Newfoundland


The climate of snow cover is of great interest to a number of sectors such as agriculture, winter recreation and hydrology. Snow depth measurements are taken as part of the routine daily measurements at ordinary climatological stations in the Environment Canada network. However the quality of the data base is degraded by a large number of missing values. We developed a regression model to simulate snow depth using temperature, rainfall and snowfall. The model accounts for accumulation, melt and ablation. The procedure was applied to about 270 stations over a 30 year period to complete the data base. From an analysis of the resulting data we created maps of the probability of snow cover exceeding certain depths in two week intervals from November to April.


Representation of Forest Cover in a Physically Based Snowmelt Model

Robert A. Hellstrom

Department of Geography, The Ohio State University,

Columbus, OH 43210. Internet: hellstrom.1@osu.edu


Snowcover experiences the largest spatial and temporal variability of all natural surface conditions on Earth. Modern numerical snowmelt models are producing realistic simulations for regions with insignificant forest cover, but most lack a physically based representation of vegetation and its effect on the distribution of snow. This project evelops a pragmatic procedure for measuring and modeling the effects of forest cover on the distribution and evolution of snow on the ground, thus facilitating improvement of existing an state-of-the-art snowmelt model. Modifications include independent and combined integration of four numerical sub-models that modify the amount of radiation, precipitation and wind speed beneath a forest cover. The University of Michigan Biological Station, which builds on several decades of successful climate and biological research within various environments, served as the site for the field observations. The snowmelt model ingests standard hourly meteorological observations, as derived from the above-canopy measurements, recorded at the near-by AmeriFlux tower. Two automatic weather stations, located beneath a deciduous and pine forest canopy, provide data to for verification of the model's output during the snow accumulation and melt season of 1998-99. Statistical comparisons between the output of the modified and original models assess the ability to simulate snow accumulation and melt at the forest floor. Results from this project will improve simulations of snow depth in forested areas, thereby providing specialists such as hydrologists and meteorologists with a practical tool for snowcover assessment in forested areas. Some applications include mitigation strategies for flood control, improved agricultural practices, and improved interpretation of remote sensing images.

Investigations of Preferential Melt Pathways in a Natural Snow Pack

Mary Albert, Gary Koh, Frank Perron

Cold Regions Research and Engineering Lab

Hanover, N.H. 03755


The formation of preferential flow pathways in porous media has long been recognized, and its importance in snow has been acknowledged. Models of snow melt that attempt to include these effects have been formulated, but unfortunately leave as 'user input' the unknown number and size of flow channels through snow. This fundamental omission has severely limited application of the models. It has been shown (Albert and Hardy, 1993) that homogeneous, or "ripe" snow exists only for the last several days of the snow melt season; thus preferential flow pathways are likely to be important to snow melt predictions through most of the season.

We used an FMCW radar can be used to examine the size and distribution of preferential flow pathways (e.g. flow fingers) formed in natural, undisturbed snow. The use of radar for detecting melt pathways in snow has several advantages over dye tests: it does not change the surface albedo of the snow, and there is no water sprayed on the snow, and thus will be detecting flow caused by ambient meteorological conditions only.

Field measurements made on a natural snowcover at the Sleepers River Research Watershed in northern Vermont. Three field measurements were made that include old, dry, refrozen snow, fresh snow, and ripening snow during the active melt period. The radar was able to detect melt fingers when they existed, and excavations of the refrozen fingers confirmed their existence. Results are compared with snow melt lysimeter measurements throughout the season in an attempt to estimate the amount of water transmitted by the melt fingers through the pack.


Albert, M. and J.P. Hardy, 1993, Snowpack Stratigraphy Evolution in Open and Forested Sites, Proceedings of the 1993 Eastern Snow Conference.

The Isotopic Evolution of Snow and its Melt

S. Taylor1,2, X. Feng2, R. Osterhuber4, J.W. Kirchner3, B. Klaue 2 and C. Renshaw2

1 U.S. Army Cold Regions Research and Engineering Laboratory, Hanover, NH. 03755

2 Dept. Earth Sciences, Dartmouth College, Hanover, NH 03755.

3 Dept. of Geology and Geophysics,University of California, Berkeley, CA 94720.

4 Central Sierra Snow Laboratory, Soda Springs, CA 95728


In northern latitudes the isotopic variability of snowmelt is a key component for separating the spring hydrograph into its component water sources. Knowing the provenance of the water is important for estimating storm runoff and understanding thermal and chemical inputs to a river. However, because of the complex and variable nature of processes within a snowpack a quantitative link between the isotopic composition of the original snow, the snowpack and its melt has not been established. By keeping track of the snow stratigraphy we can compare the isotopic composition of freshly fallen snow with its composition in both the snowpack and meltwater samples. Our results are from a field study conducted at the Central Sierra Snow Laboratory (CSSL), California during the winter of 1997-98.

At CSSL, we collected fresh snow from all major storms and collected daily meltwater samples from a 3x6 m meltpan. After each large storm, the snow surface over the meltpan and an adjacent area were sprayed with a dilute solution of a rare earth element (REE) chloride. Before the onset of spring melt we sampled snow from snow pits dug in the adjacent area. REE analyses of the samples taken at 0.1 m increments from a 3.1 m-deep pit, let us determine the depth interval corresponding to each fresh snow sample. We analyzed oxygen isotopic compositions of 15 new snow samples, all 31 samples from a snow pit and 15 melt water samples (a sample about every five days).

Our results show that the isotopic redistribution within the snowpack occurs both during snow metamorphism and when meltwater interacts with snow as water percolates through the pack. Snow metamorphism significantly reduced the isotopic variability within the snowpack. The isotopic exchange between liquid water and ice caused the snowmelt to have low d18O values early in the season and become heavier as melting progressed.

Flood forecasting the 1999 Snowmelt in the Grand River Basin using WATFLOOD

F. Yusuf, N. Kouwen, & S.R. Fassnacht

Department of Civil Engineering, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1


In early January of 1999, a low pressure storm system originated in the Gulf of Mexico and traveled north, through the midwest United States. This storm resulted in near-record snowfalls in numerous areas including southern Ontario and Quebec, with snow depths up to 40 cm being recorded. This large snowfall was followed by several smaller, yet significant winter storms. By the middle of January, approximately 100 cm of snow had fallen in many parts of southern Ontario, including the Grand River basin. This accumulation of snow represents one of the highest in the Grand River basin in several years. Above freezing temperatures occurred in mid-January that partially melted the pack and produced significant streamflow. Prior to this partial melt, the spring snowmelt was expected to cause major flooding along the Grand River and its tributaries. However, a substantial quantity of water remained in the pack and flooding may still be expected when the spring snowmelt occurs.

This paper is intended to summarize the results of a flood forecasting study for the Grand River, which will focus on the 1999 snowmelt. The distributed hydrologic model, WATFLOOD, was used in conjunction with meteorological data for the Grand River basin to obtain real-time flow estimates. Precipitation data obtained from the King City radar station, temperature data from local weather stations and snowpack measurements from the Grand River Conservation Authority (GRCA) were used with WATFLOOD to track the snowmelt and subsequent runoff. The results of the flood forecasting for the Grand River are presented in hydrographs for the entire winter, with an emphasis on periods when the snowmelt runoff was significant.

A statistical analysis is performed on snowpack data using all historical measurements as well as the 1999 snowcourse data. Both snow depths and SWE are examined using frequency analysis methods to estimate the magnitude of the January 1999 early season snowfall with respect to previous years.

Tracer-Tests of Water Drainage from Cold Snow

Bøggild, Carl .Egede 1, Knudsen, Mogens Brems 2

1The Geological Survey of Denmark and Greenland (GEUS), Thoravej 8, DK-2400 Copenhagen NV, Denmark

2ASIAQ, Greenland Field Investigations, P.O.Box 1003, DK-3900 Nuuk, Greenland



In arctic regions such as Greenland, the access to fresh water is closely connected with the climate. Insight into the hydrological processes and snowmelt is essential for assessment of the water resource, since the major discharge from Greenland catchments is confined to a few summer months with snowmelt.

The vicinity of the Greenland hydrological basins to the ice sheet means that they almost entirely constitute of exposed bedrock, since most sediments have been removed by glacial erosion during several glacial cycles. This bedrock surface result in a fast hydrological response to precipitation and snowmelt events. Under such conditions timing and magnitude of snowmelt are important for evaluating the resulting hydrograph.

In a basin "Tasersuaq" in Westgreenland different models have been tested to gain insight to the key processes which control snowmelt. During this study it was evident that the discharge from the basin started almost simoultanously after onset of melting, despite the occurance of a cold snowpack which in principle would refreeze the meltwater. In light of this, tracer tests have been carried out to study the process of run-off from a cold snowpack.

When injecting the dissolved tracer on top of the wind packed surface, the water was trapped below surface in ice lenses due to snow temperatures some 10 K below the freezing point. However, when repeating the same experiment at an excavated surface one meter below, the water concentrated into vertical pipes and resulted in drainage out of the snowpack, at snow temperatures 8 K below the freezing point. At the conference observations will be presented which documents this run-off from cold snowpacks.

CO2 in Arctic Snow Covers: Gaseous Concentrations and Landscape Form

H.G. Jones1, J.W. Pomeroy2, T.D. Davies3, M. Tranter4 and P. Marsh2

1 INRS-Eau, Ste-Foy, Quebec, G1V 4C7, Canada

2 National Hydrology Research Centre, Saskatoon, Saskatchewan, S7N 3H5, Canada

3 Climatic Research Unit, University of East Anglia, Norwich, NR4 7TJ, United Kingdom

4 Department of Geography, University of Bristol, Bristol, BS8 !SS, United Kingdom


The physical characteristics and CO2 concentrations of snow cover in the Canadian arctic (Inuvik) were examined at two sites with different landscape forms (valley floor, hillslope, plateau). The greater exposure of plateau snow cover to wind results in differences in the structure of the snow cover and the permeability of the different snow strata compared with snow covers on the other landscape forms. Both higher in-pack concentrations of CO2 and the largest vertical CO2 concentration gradients were found in plateau snow cover, the smallest in the deeper hillslope and valley snows. CO2 gradients in all landscape snow covers followed two patterns i.e. where concentrations at the soil-snow interface are higher than those just below (5 cm) the snow-atmosphere interface and vice-versa. It is suggested that the latter pattern may be due to the transport of the gas from the soil surface to the upper levels of the snowpack by wind-induced advection (windpumping) - the phenomenon being of greater importance in the case of the plateau snows than those of the other landscape forms.

Modes of North American Prairie Snow Water Equivalent and Low-Frequency Atmospheric Teleconnection Patterns

C. Derksen, E. LeDrew

Waterloo Laboratory for Earth Observations, Department of Geography
University of Waterloo, Waterloo, Ontario, Canada N2L 3G1
Phone: (519) 888-4567 Ext. 2689, Fax: (519) 888-6768
email: chris@watleo.uwaterloo.ca

B. Goodison

Climate Research Branch, Climate and Atmospheric Research Directorate
Atmospheric Environment Service, Downsview, Ontario, Canada M3H 5T4


Time series of passive microwave derived snow water equivalent (SWE) data are able to provide the synoptically sensitive data required to monitor associations between snow cover and atmospheric circulation. In this study, ten winter seasons (December, January, February 1988/89 to 1997/98) of five-day average (pentad) SWE imagery are utilized to examine the seasonal snow cover characteristics of a ground validated North American Prairie study area. The images are derived from Special Sensor Microwave/Imager (SSM/I) brightness temperatures, processed with the Canadian Atmospheric Environment Service dual channel SWE retrieval algorithm. The ten-season time sequence of pentad imagery were subjected to a rotated principal components analysis in order to isolate the dominant spatial modes of SWE within the study period. The component loading patterns, which provide a measure of the temporal persistence of a given component pattern, were then cross correlated (with time lags) to atmospheric teleconnection indices (East Pacific, North Pacific, Pacific/North America, Tropical/Northern Hemisphere, and North Atlantic Oscillation patterns) archived by the National Oceanic and Atmospheric Administration. This analysis allows an examination of relationships between regional SWE patterns and a quantitative measure of low-frequency atmospheric circulation. Results indicate that unique SWE regimes are present within the time series, which can be linked to preferential modes of atmospheric circulation. No clear trend in passive microwave derived North American Prairie SWE or snow covered area is evident, with both anomalous high and low snow cover occurring throughout the time series.

Aspects of Snow Extent and Water Equivalent Pertaining to Hydrologic

Modeling in Complex Landscapes

Robert E. Davis

U.S. Army Engineer Research and Development Center

Cold Regions Research and Engineering Laboratory

72 Lyme Road, Hanover NH 03755


It is a common observation that snow extent and snow depth correlate over many, if not most, land cover types. During snow accumulation, very little precipitation can cover all land cover features under conditions of low wind speed and sub-freezing temperatures. On the other hand, during ablation the relationship between snow covered area (SCA) and total snow water equivalent (SWE) over an area depends on the distribution of snow depth and components of surface energy exchange. Specifically, the interaction between the spatial statistics of SWE, the distribution of SWE at scales finer than the area of interest, and the spatial distribution of net energy flux to the snow cover control the functional relation between SCA and SWE. The area-depth relationship for snow cover has important implications for modeling snow processes and modeling the spatial distribution of snow from spatially sparse measurements of depth or SWE. This literature review and synthesis presents and discusses the relationship between snow extent and water equivalent in the context of snow modeling and remote sensing over complex landscapes, such as the temperate forests of North America.

To be credible, spatially distributed models of snow must be validated or updated with spatial data. Many tests of snow model predictions distributed over large areas have suffered from the difficulty in quantitative evaluation due to variability in snow extent patterns and snow physical properties. Some studies have made the effort to measure snow properties over time at a wide variety of sites in the modeling domain. However, this type of validation becomes increasingly cost prohibitive when testing snow model predictions over larger and larger areas. Snow extent is mapped operationally, and many areas have sufficient cloud free periods to compile time series of snow cover patterns, but suffer from the masking effects of forest canopy. How snow extent maps are used for validating snow models depends on the spatial resolution of the map and model, and the spatial and temporal variance of snow accumulation and ablation. Regardless of the method chosen to segment land cover and terrain for modeling, as model domain size increases, so does the likelihood that snow cover on an element becomes discontinuous while significant SWE remains. Runoff from snow covers in temperate forests of northeastern North America, can be ephemeral, occuring intermittantly through winter and spring periods. This adds another dimension of complexity to using snow extent and water equivalent map products for hydrologic forecasting.

Abstracts of Poster Presentations
(in alphabetical order by last name of first author)



Integration of Remote Sensing Measurements and Numerical Modelling of Snow on Sea Ice

D. G. Barber and J. Hanesiak

Centre for Earth Observation Science

Department of Geography

University of Manitoba




Our understanding of snow distributions in the polar regions is severely restricted due to the heterogeneity, both in space and time, of this solid precipitate. Processes such as vapour, mass and energy fluxes across the interface are, to a large extent, controlled by the presence and geophysical state of the snow cover on sea ice. A significant portion of the uncertainty in cryospheric processes can be linked to the role which snow play in moderating these fluxes across the ocean-sea ice-atmosphere interface.

In this paper, we present several years of research results which show how both passive and active microwave remote sensing can be used to estimate geophysical characteristics of snow on sea ice (grain size, water in liquid phase and SWE). We then present our approach to integrating these remote sensing estimates directly within a numerical snow model (SNTHERM) and one dimensional sea ice model (Flato). We conclude with a description of how this integration of remote sensing and modelling can be used to understand physical and thermodynamic processes operating across the ocean-sea ice-atmosphere interface.


Implementing the Martinec/Rango Snowmelt Runoff Model in the USGS Modular Modeling System

Neftali S. Cajina and Kaye L. Brubaker

University of Maryland, College Park


Al Rango

Hydrology Laboratory, U.S.D.A. Beltsville Agricultural Research Center,



Current practice in modeling snow hydrology uses a variety of self-contained, stand-alone models of varying complexity. Models differ in their scale and resolution, in their data requirements, and in their ability to simulate different phenomena (for example, point or profile water and energy balance versus streamflow hydrographs). In the case of off-the-shelf or packaged models, it is often difficult to examine the inner workings of the model, let alone adjust some part of the model procedure for a slightly different application. The snow modeling community could benefit from a systems approach, in which the user may exercise judgement and control in selecting and combining model components that best represent the physical processes of interest, in light of available data and desired results. The Modular Modeling System (MMS) developed by USGS provides a framework for this approach; however, the MMS has not been widely applied in snow hydrology. In this paper, the widely used Martinec/Rango Snowmelt Runoff Model (MRSRM) is re-programmed into the MMS framework. The goal of this work is to determine the investment required and the benefits gained by incorporating a stand-alone model into MMS. Eventually, our goal is to identify the strongest components of the MRSRM and similar existing models and judge whether they can be blended in the systems framework to build more accurate and useful models. The process of MMS implementation is presented, including conceptual modularization, variable accounting, incorporation of MMS syntax to connect the modules, debugging, and reproducing the stand-alone model results. Conversion requires basic programming skills and some knowledge of physical hydrology. Advantages of the MMS framework include easy display of intermediate results, the ability to concentrate on process representation and process connectivity (rather than data input/output or the graphical user interface), automatic sensitivity analysis, and deeper insight into both the model and the physical system.

Indices of Global Warming in the high Arctic - Contradictions and



J.G. Cogley, W.P. Adams and M. A. Ecclestone

Dept. of Geography, Trent University

Peterborough ON



The mass balances of White and Baby Glaciers, at 79° N on Axel Heiberg Island in the high Arctic of Canada have been monitored more or less consistently since 1960. Photogrammetric estimates of terminus fluctuations are available for White Glacier and its larger neighbour, Thompson Glacier, from relatively frequent photographs dating back to the first recorded image in 1948. In addition, an ice-off (breakup) record of Colour Lake, located between White and Baby Glaciers, is available to 1959. These data series, up to 1997, are presented. While the glacier records show an excess of ablation to accumulation and so suggest warming, the lake ice record and the advance of Thompson Glacier seem to indicate cooling. These records are of special significance because they are indices of conditions at nearly 80° N, where global climate models suggest that warming should be most pronounced. However, the large-scale significance of such local records should, of course, be assessed with caution.


Wind shield evaluation at NCAR/Marshall test site in Boulder, CO during 1998-99 winter

Jeffrey A. Cole

National Center for Atmospheric Research

Applied Science Group

Research Applications Program

P.O. Box 3000 Boulder, CO 80307-3000

Phone: 303-497-8421 2048Fax: 303-497-8401



Measurement of real-time solid precipitation is complicated by many factors including snow density, wind effects and gauge design charateristics. NCAR has been testing various precip gauges and wind shield configurations over the past several winter seasons and has collected an extensive data set for evaluation of these devices. the focus of the 1998-99 winter season was an extensive evaluation of various wind shields and their effectiveness towards producing an accurate real-time measurment of solid precipitation. The NCAR/Marshall test site was used for this evaluation, which gets approximately 80+ inches of snow and strong downslope wind events that occur several times per month. Data from the gauges and various wind shield configurations will be presented for an overall comparison of the instrumentation.

Analysis of Microwave Radiometry of Snow Cover with SSM/I Data in a Taïga Area : The Case of James Bay Area

Danielle de Sève1, M.Bernier1, J-P. Fortin1, A. Walker2

1 Université du Québec, Institut National de la Recherche Scientifique (INRS-Eau)

2800 rue Einstein, C.P 7500, Sainte-Foy, Québec GIV 4C7, Canada


2 Climate Research Branch, Atmospheric Environment Service

Downsview, Ontario M3H 5T4, Canada


Since September 1995, the Institut National de la Recherche Scientifique (INRS-Eau) has teamed up with the Atmospheric Environment Service (AES) as part of a multidisciplinary project on changes to the cryosphere named CRYSYS (Use of the CRYospheric SYStem to Monitor Global Change in Canada). The research has focused on the development of algorithms to derive snow cover information passive microwave remote sensing data.

The focus of this study is a spatio-temporal analysis of brightness temperatures (Tb) over a taiga landscape from 2 winters seasons (1997-1998). A second objective is to evaluate the impact of the land cover on Tb variations.

A first analysis conducted in 1996 on Tb variations at 37 GHZ has shown that TB decreased when SWE or snow thickness increased and that the relationship was inverted when the SWE was over 200 mm. Analyses conducted for the winters 1997 and 1998 have shown similar trends, however, with an inversion of the relationship between SWE and Tb at 150 mm. Analysis of Tb variations at 19 GHz for series 1996 to 1998 indicates generally a slight increase in Tb when SWE increase. The variations in Tb at this frequency are mainly associated with variations in air temperature and, to a lesser extent, to ground temperature. So, over the entire year, the increase in Tb values at 19 GHz is thus related to gradually rising air temperatures.

Land cover variability within a pixel is an important factor to keep in mind because it strongly influences the snow signature. Within the study area, frozen lakes and coniferous forest have a large effect on the Tb value of a pixel since they cover over 20% of the territory.

SAR and optical satellite observations of ice-covered thermokarst lakes, old crow flats, Yukon Territory

Claude R. Duguay1, Yannick Ernou1, and Jim Hawkings2

1 Centre d’etudes nordiques, Universite Laval
Sainte-Foy, Quebec G1K 7P4, Canada

2Canadian Wildlife Service, Environment Canada
Whitehorse, Yukon Territory Y1A 5B7, Canada


Recently, some investigators have shown the potential of ERS SAR data for monitoring ice formation, the thickening of ice cover, and freezing to the bottom, of shallow Arctic and Subarctic lakes. Shallow lakes represent promising sites for long-term monitoring and the detection of changes related to global climate change and its effects on the Polar Regions. However, monitoring of lake ice during the spring break-up period is more difficult because surface water (i.e. ponding and free liquid water in snow) causes the SAR signal to be absorbed, thus masking the ice below.

In this paper, results from the analysis of backscatter and reflectance signatures of ice and snow on shallow thermokarst lakes will be presented. The study area is located in the Old Crow Flats, northern Yukon (68o N, 140o W). A time series of ERS-1 images (winter 1994-1995) is used to monitor the evolution of ice cover from 30 lakes of known depth. Emphasis will be placed on the analysis of spring imagery. In particular, ERS-1 backscatter and Landsat TM reflectance signatures from late May acquisitions will be compared to show: 1) the limitations of SAR in differentiating various ice and snow types during the break-up period, and 2) the benefits of using SAR and optical satellite data in concert.

Determining Depth and Ice Thickness of Shallow Subarctic Lakes and Ponds using Spaceborne Optical and SAR

Claude R. Duguay1 and Peter M. Lafleur2

1Centre d’etudes nordiques, Universite Laval

Sainte-Foy, Quebec G1K 7P4


2Department of Geography, Trent University

Peterborough, Ontario K9J 7B8


Shallow lakes and ponds are conspicuous features in northern Canada, Alaska, and eastern Siberia. During summer they play a significant role in the regional energy and water balance, and their importance in the global methane budget is now well recognised. There is evidence that conductive heat losses from ice-covered lakes on the Alaskan North Slope also make a significant contribution to the regional energy budget during winter. For residents of northern settlements, lakes are an important source of fresh water. However, during the winter period, ice freezes completely to the bottom of many of the lakes since a large number of them are shallower than the maximum ice thickness. The determination of lake depth, the timing of bottom freezing, and ice thickness of shallow lakes is therefore necessary for a better management of water resources in high latitude environments.

In this paper, we present an approach to determine depth and ice thickness of shallow lakes and ponds using Landsat Thematic Mapper (TM) and ERS-1 SAR data. A summer time Landsat TM image is used to map lake bathymetry and multi-date ERS-1 images acquired during winter are utilised to determine when and which lakes freeze to the bottom during winter. The two remotely sensed derived products are then combined to estimate ice thickness from lakes and ponds on a monthly basis. The approach has been developed and tested successfully in a Subarctic tundra-forest landscape in the Hudson Bay Lowland near Churchill, Manitoba. Results indicate that the approach is particularly well suited for estimating depth and ice thickness of oligotrophic and ultra-oligotrophic lakes that are ubiquitous in many regions above treeline.

Utilisation d’un Georadar en Support a l’Etude de la Glace Lacustre par Teledetection


Deborah Drai, Claude R. Duguay, and Frederique Pivot

Centre d’etudes nordiques, Universite Laval

Sainte-Foy, Quebec G1K 7P4




Les étendues lacustres représentent une constituante importante du paysage canadien, et sont, dans les régions subarctiques, caractérisées par des couverts de glace et de neige 8 à 10 mois par an. Or, la glace de lac se révèle importante non seulement en tant qu’indicateur des conditions météorologiques et climatiques d’une région donnée (par l’intermédiaire de son épaisseur ou encore de ses dates de gel/dégel), mais également dans le cadre de la productivité biologique ou encore de la gestion des ressources biologiques de certains habitats de proximité.

L’objet de cette recherche, menée dans le cadre du projet CRYSYS d‘étude de la cryosphère, s’intégrant lui-même au projet EOS d’observation de la terre commandité par la NASA, est donc d’étudier l’évolution spatio-temporelle de couverts de glace lacustre d’un milieu de toundra et de forêt subarctique par télédétection. Des deux méthodes de télédétection utilisées, à savoir l’imagerie satellitaire RADARSAT et le géoradar, seul la seconde sera ici abordée.

Des données de terrain concernant les caractéristiques structurales et stratigraphiques des couverts de glace lacustres de 4 lacs de la région de Churchill, située au nord du Manitoba, ont été acquises au cours de 3 campagnes de terrain (février, mars et mai 1998), parallèlement à des profils géoradar au niveau des mêmes sites d’étude (mars et mai 1998 seulement) sous 2 modes de relevés (réflexion et CMP) et 2 différentes fréquences (450 et 900 MHz). L’utilisation de cette technique a été motivée par la difficulté d’obtenir des carottes nettes ou entières.

Les résultats préliminaires à ce jour sont les suivants :

  1. plus la fréquence est basse, plus la résolution est basse mais meilleure est la profondeur de pénétration du signal radar; en effet, le fond du lac est toujours discernable sur les profils de fréquence 450 MHz, alors que ce n’est pas le cas pour ceux de fréquence 900 MHz où la stratigraphie des couverts lacustres est en revanche beaucoup plus détaillée, mais où la profondeur de pénétration n’excède pas les 1,70m.
  2. la vélocité du signal dans le couvert de glace semble être un bon indicateur de sa densité bullaire; en effet, on constate que les sites où la vélocité se révèle supérieure à la normale établie pour la glace d’eau douce (entre 0,15 et 0,16 m/ns) semblent être caractérisés par une densité bullaire plus importante que les autres sites d’étude.

Use of a Georadar in Support of Satellite Remote Sensing Investigations of Lake Ice


Deborah Drai, Claude R. Duguay, and Frederique Pivot


Centre d’etudes nordiques, Universite Laval

Sainte-Foy, Quebec G1K 7P4



Lakes are a conspicuous feature of the Canadian landscape and, in subarctic regions, they are covered by ice and snow 8 to 10 months of the year. Lake ice is important not only as an indicator of meteorological and climatological conditions of a region (through variables such as ice thickness, as well as freeze-up and break-up dates), but also of interest in the context of biological productivity and water resources management of northern settlements.

The objective of this research, which is being conducted as part of the Canadian EOS-CRYSYS program, is to study the spatial and temporal evolution of ice cover from subarctic tundra and forest lakes. Of the two approaches used to study lake ice (RADARSAT imagery and georadar), only the second one will be covered.

Field observations on the structural and stratigraphic characteristics of ice cover from four lakes located near Churchill, northern Manitoba, were acquired during three field campaigns (February, March, and May 1998) along with ground probing radar profile measurements made on the same lakes in March and May 1998. The georadar acquisitions were obtained in reflection and CMP modes at 450 MHz and 900 MHz frequencies. The use of this technology was motivated in part by the difficulties in obtaining complete ice samples at some sites.

Preliminary results obtained to date show that:

  1. There is a decrease in resolution (details), but an increase in penetration depth at the lower frequency. Indeed, lake bottoms are always visible on profiles acquired at 450 MHz. At the higher 900 MHz frequency, however, the ice stratigraphy is more detailed but with a loss in penetration to a depth of about 1.7 m.
  2. The signal velocity examined for the lake ice covers seems to be a good indicator of the density of bubble inclusions. Indeed, we observed greater velocities at sites where the bubble density was more important, compared to standard values of freshwater ice (0.15-0.16 m/ns).

A Shape Function Estimate for Fresh Dendritic Snowflakes

S.R. Fassnacht, F.R. Seglenieks, E.D. Soulis, and N. Kouwen

Department of Civil Engineering, University of Waterloo, Waterloo, Ontario,

Canada N2L 3G1


The shape of newly formed snowflakes is an important qualitative parameter as input to the development of a snowpack and for potential atmospheric scavenging, which ultimately influences the metamorphosis and transport of snow and contaminants. The relationship between formation temperature and type of particle, i.e., shape, is well known and other relationships have been developed that define length and width ratios for various crystal types. However, little data exists that quantifies the specific surface area (SSA) of different particle shapes.

In this paper, known spatial properties, i.e., geometric relationships, and particle size distributions are applied to a portion of the dendritic snowflakes observed by Bentley (Bentley and Humphries, 1931), in order to estimate their SSA. This type of snowflake is said to form in the temperature range of -13 to -17oC. Probability distribution functions are derived for SSA and are compared to measured estimates of SSA for fresh snow. The presence of rime is discussed with respect to the increase of SSA as a function of degree of riming.

Snow monitoring at the Kouchibouguac National Park, New Brunswick (1974-1998)

Guillaume Fortin1, Hardy Granberg, Jean-Marie Dubois

Département de géographie et de télédétection, Université de Sherbrooke,Sherbrooke,Qc, Canada

1. Present address : Institut National de la Recherche Scientifique, INRS-EAU, Université du Québec, Qc, Canada


The main goal of our work is to present the method used to verify the validity of the data from the Kouchibouguac National Park and we want to provide an example of the snow dynamics for a typical winter in the Park. Kouchibouguac National Park is located on the east coast of New Brunswick and the area covered is238,8 km2.

Snow is an important factor in the Kouchibouguac that influences many phenomena in the ecosystem. The annual mean snow precipitation is 208 cm. Kouchibouguac National Park decided to start a snow monitoring in 1974. The data collect is conduct by the Warden Service at every two weeks. Nine representatives sites were selected and every site has two parts : one site in an open area and another in an closed area (forested area). Since 1974 the data collection starts with the first snow and ends with melting of the snow. Generally this period extent on four months and starts at the end of december and finishes at mid-april.

The purpose of that monitoring is to provide information about the snow conditions (quantity and types of snow).

For this study we use data from three differents sources :

1- data of the snow type and snow depth from 1974 to 1998; those data come from the Warden Service;

2- temperatures (minimum and maximum), daily precipitations (solid and liquid) and the total accumulation of snow on the ground from 1974 to 1995; those data come from an automatic station (managed by Environment Canada) who was located in the park;

3- detailed data from two snow surveys in 1997 and in 1998.

All these data were uniformised to become conform to the International Classification for seasonal snow on the ground (1985).

A problem in these study is that we have subjective data, like the snow type from the Warden Service, and, in other hand, some objective data like the meteorological data from Environment Canada. The challenge is than to obtain a good degree of uniformity between the different sources. A good way to verify the validity and the conformity is to make comparaisons between the sources using graphics. Once the comparaisons were done we could describe the principal snow conditions for every winter and for a typical winter that include the dynamics of the snow accumulation.

The typical snow conditions that can be observed in the Park generally start with some snow precipitations in mid-November but the snow accumulation on the ground really starts in mid-December with colder temperatures. With the very cold temperatures of January there are slight snowfalls and then snow accumulation tends to be stable. The slighty higher temperatures of February advantage greater snowfalls and then the quantity of snow on the ground. The maximum thickness of snow on the ground occurs at the begin of March and could be explained by all the accumulation of the previous months. The final melt of the snowpack starts at the end of March but some snowfalls can occur in April. We can also say that the mean snow thickness and surveyed by the Warden Service and the temperatures that can be derived from its stratigraphy are generally conformed with the snow data and temperatures surveyed by Environment Canada.

Snow Mapping through Dense Forests using Multiple Satellite Data Sets

Dorothy K. Hall*

Milan Allen**

Andrew B. Tait+

*Code 974, Hydrological Sciences Branch, NASA/Goddard Space Flight Center, Greenbelt, MD 20771


**National Operational Hydrologic Remote Sensing Center, Chanhassen, MN 55317


+Universities Space Research Association, NASA/Goddard Space Flight Center,

Greenbelt, MD 20771



It has long been known that mapping snow from space through dense forest cover is difficult. While there are many ways to circumvent this problem by subjective analysis to provide useful and accurate snow maps, it remains a serious problem for fully-automated snow mapping from space. In this study, we utilize multiple satellite data sets and snow-cover products to map snow cover through evergreen needleleaf forests in Idaho for a time period in January 1998. We also employ the International Geosphere-Biosphere Project (IGBP) land-cover map of North America, developed from 1-km Advanced Very High Resolution Radiometer (AVHRR) data, and available from the EROS Data Center in Sioux Falls, SD. The National Operational Hydrologic Remote Sensing Center (NOHRSC) 1-km snow-cover product which is produced weekly in 4,000-6,000 basins in the United States and Canada, is compared with data derived from 30-m resolution Landsat Thematic Mapper (TM) and 25-km resolution Special Sensor Microwave/Imager (SSM/I) data from which snow maps were produced. The prototype of an algorithm that was developed to map global snow cover in an automated way, called SNOWMAP, was run on a 28 January 1998 TM scene of Idaho. SNOWMAP was developed to map daily, global snow cover using future Earth Observing System (EOS) Moderate Resolution Imaging Spectroradiometer (MODIS) data. MODIS snow products will be available from the National Snow and Ice Data Center (NSIDC) following the July 1999 launch of the first EOS satellite. Snow cover is mapped from SSM/I data once the scattering signatures from precipitation, cold deserts, and frozen ground have been filtered out. The TM, NOHRSC and SSMI maps show that the area is 62%, 87% and 100% snow covered, respectively. Land-cover data show that the areas on the TM scene that are not snow covered are composed of evergreen needleleaf forests and croplands. In addition, snow is not mapped along the rivers. These lower-elevation areas are snow-free according to a digital elevation map and the SNOWMAP algorithm. Meteorological data show snow depths ranging from 0 to 117 cm in the study area. The difference in resolution among the data sets explains some of the difference in amount of snow mapped, but does not explain a discrepancy in amount of snow mapped in the southwestern part of the study area that exists between the NOHRSC and SNOWMAP data sets.

Snow Depth and Soil Frost Modelling in a Northern Hardwood Forest

Janet P. Hardy and Rachel Jordan

USA CRREL, 72 Lyme Road Hanover, NH 03755

Peter Groffman

Institute of Ecosystem Studies, Box AB, Millbrook, NY 12545

Scott Nolin

Institute of Ecosystem Studies, HBEF, Campton, NH 03223

Timothy Fahey

Cornell University, Dept. of Natural Resources, Ithaca, NY 14853

Charles Driscoll and Ross Fitzhugh

Syracuse University, Dept. of Civil and Environmental Engineering, Syracuse, NY 13244


The insulating properties of a continuous snow cover are known to protect soil from extensive freezing in a normal winter. In this study, we propose that one of the most dramatic effects of climate change in the temperate forest may be less snow cover and more extensive soil freezing resulting in changes to soil biogeochemical processes and to the delivery of solutes to streams. We conducted an interdisciplinary study to quantify the effects of reduced snow cover on soil freezing and biogeochemical processes in a northern hardwood forest. Our field investigation took place at the Hubbard Brook Experimental Forest near N. Woodstock, New Hampshire. The Hubbard Brook Experimental Forest was established in 1955 and became a Long-Term Ecological Research (LTER) site in 1988. Consequently, there is an extensive database of meteorological, stream, snow and soil frost data. Within the forest, we established four experimental sites in pure stands of sugar maple (Acer saccharum) and yellow birch (Betula lutea). At each site we established two 10 x 10-m plots; including a "control" plot and "freeze" plot. At each of the eight plots, thermistors and soil moisture probes continuously measured snow and soil temperatures and soil moisture. We also made periodic measurements of soil water chemistry, trace gas fluxes and fine root dynamics. At each site, the "freeze" plots were manually kept snow-free until early February to mimic a late arriving snow cover. During the first year of treatment, maximum snow depths in the control plots ranged from 0.75 to 1.0 m in mid-March, while snow accumulating on the freeze plots never exceeded 0.35 m. Snow removal successfully induced soil freezing in all "freeze" plots. Soil temperatures at 0.1 m depth reached -2° C in all but one anomalous "freeze" plot, but were above 0° C in all "control" plots. Physically based adjustments to the meteorological input for the one-dimensional mass and energy model, SNTHERM, allowed us to account for the forested environment and accurately predict snow depth, and the extent of soil freezing during the first winter of treatments (1997/98). Recent modifications to SNTHERM, improved its capability to predict soil properties and processes at forested sites. We used SNTHERM and Hubbard Brook’s long-term dataset to hindcast previous soil-freezing events. Snow depth and soil freezing were modelled for the winter 1989-90, as this year experienced extreme soil freezing and significant chemical signatures in the streamwater. Our modelling scenario is also used to predict the effect of future climate change scenarios.

Atlantic Sea Surface Temperatures and Winter Snowfall along the Southern Margin of the Eastern United States Snowbelt

Suzanne Hartley

Department of Geography, Government and History

Morehead State University

Morehead, KY 40351



A previous study identified significant correlations between winter sea surface temperature anomalies (SSTAs) off the northeast coast of the United States and winter snowfall in southern and coastal New England. A follow-up study of wider geographic scope has identified a similar snowfall-SSTA association for a region covering parts of Virginia, North Carolina and Tennessee. The association in this region is much stronger, despite its location far inland and distant from the area of the Atlantic Ocean in question. Furthermore, there are significant lag associations between winter snowfall and fall SSTAs which at some locations, are of greater significance than the contemporaneous associations.

Following falls with warm SSTs in the index region, the winter 700 mb height anomaly pattern suggests zonal flow and a more northerly storm track across the eastern US, while cold SST falls tend to precede winters with a more meridional flow pattern and more southerly storm track. From this study alone, it is not possible to determine to what extent, if at all, the fall SSTAs actually influence the subsequent winter atmospheric circulation. But it is possible that the coupled atmosphere-ocean system may have modes of fall-winter transition that are characterized by distinct patterns of SSTAs in the western Atlantic Ocean that are already evident in the fall.

Just how useful this lag association may be for seasonal snowfall forecasts is not clear though. At Roanoke, VA, the fall SSTA lag association explains roughly 25 percent of winter snowfall variance from 1951-1992. However, the snowiest winters of this period occurred in the 1960s which was a decade of unusually cold conditions off the east coast of the USA, and this may have biased the result. An examination of winters since 1992 suggests that fall SSTAs in themselves may not always be a reliable predictor of winter snowfall conditions.

The Integration of Digital GOES-8 Satellite Imagery with Observational

and Prognostic Datasets During the Winter Storm of January 1999

Jane A. Hollingsworth and Julie L. Adolphson

National Weather Service Office, Northern Indiana, USA


The winter storm of January 1-3, 1999 resulted in the first significant synoptic snowfall of the 1998-1999 winter season to the Midwest and Great Lakes regions. This paper will illustrate the significance of high resolution digital satellite imagery in combination with operational datasets in diagnosing and forecasting the mesoscale and synoptic features associated with this event. Special emphasis will be given to the recent advances in using the 3.9 micron (shortwave infrared) imagery (Menzel et al. 1994) in determining water phase and precipitation types. Additional satellite imager channel data is presented to support the operational forecasting challenge. Finally, surface and upper air data along with gridded model fields are shown to detail the synoptic and mesoscale dynamics associated with this case.

Climate, snow and glaciers temporal variability in high latitudes of the North and South hemispheres (Eurasian sector and the East Antarctic seashore)


Institute of Geography, Russian Academy of Sciences

Russia, Moscow, 109017, Staromonetny per., 29

tel. (095)9590034, fax (095)9500033, e-mail climat@ipcom.ru


The main aim of investigations is to ground remote interhemisphere relations between long-term dynamics of snow storage, climate elements and glaciers. These interrelations are determined by conjugated circulation processes in Eurasian sector of the North hemisphere and the East Antarctic seashore. In the capacity of initial information were taken the snow cover observations and climate characteristics for the territory of Russia (in the whole and for separate regions), and the data on the dynamics and regime of East Antarctic glaciers marginal parts (1955-1995).

A comparison between parameters was based on typical schemes of the hemisphere air circulation (the North hemisphere – the B.G.Dzerdzeevskyþs typisation, the South hemisphere – the P.D.Astapenko).

The results showed an agreement between temporal snow storage dynamics, recipitation, air temperature of high-latitude Russian territories and fluctuations and nourishment regime of East Antarctic glaciers. An antiphase of parameters investigated is clearly exhibited. When in the North hemisphere the magnitudes of snow storage and precipitation are maximum and air temperatures are extremely low East Antarctic glaciers retreat on the background of snow accumulation reduction. This can be related to the duration changeability of Arctic intrusions in the eastern parts of Russia and conjugated with them by time east Antarctic intrusions, and be proved by the drop in air temperatures in the regions examined during the blocking maxima of hemispheres. During the similar time periods at the drop in air temperatures on the territory of Russia the snow storage increases. On the East Antarctic seashore at the drop in air temperatures the rate of snow storage and glaciers movement velocity fall, while iceberg calving becomes more active.

What's in the MODIS Snow Data Products?

George A. Riggs

Research & Data Systems Corp.

Code 974

NASA, Goddard Space Flight Center

Greenbelt, MD 20771-0001



Dorothy K. Hall

Code 974

NASA, Goddard Space Flight Center

Greenbelt, MD 20771-0001




A preview of the snow data products to be produced from the Moderate Resolution Imaging Spectrometer (MODIS), an instrument of the NASA Earth Observing System (EOS), is given. MODIS snow data products are developed to meet the objective of mapping snow cover extent globally. Recommendations from an external ad-hoc advisory committee have resulted in modifications to the products. These snow data products are in hierarchical data format (HDF) HDF-EOS format, a format unfamiliar to many in the community as it is a format developed specifically for EOS data. The sequence of snow data products from a "scene" product to a global gridded product is described. A concise depiction and explanation of data content and structure of each

product through the sequence of products is presented using products produced from simulated MODIS instrument data. Launch of the EOS-AM platform is scheduled for 15 July 1999. If performance of the spacecraft and instruments is nominal, and the data production system is capable of processing MODIS data, snow products may be produced and become available in the October - November time frame.