Spatial and Temporal Dependence Characteristics
of Passive Microwave Derived Prairie Snow Cover:
A Comparison of Three Winter Seasons
C. Derksen1, M. Wulder2, E. LeDrew1, and B. Goodison3
The unique influence of snow cover on energy exchange processes means that varying patterns in
snow cover distribution have climatological implications at local, regional, and global scales. The
evolution of seasonal terrestrial snow cover also influences the hydrological cycle with respect to
water storage and melt release. A means of monitoring regional, continental, and hemispheric snow
cover at spatially and temporally sensitive resolutions is therefore desired. Satellite passive micro-
wave derived snow water equivalent (SWE) imagery shows potential in providing these data through
all weather imaging capabilities, wide swath width, and frequent scene revisits.
In this study, we use five-day averaged (pentad) SWE imagery derived from Special Sensor Micro-
wave/Imager (SSM/I) brightness temperatures to assess the spatial and temporal dependence charac-
teristics of snow cover for a ground validated Prairie study area. Three winter seasons of SSM/I data
(December, January, and February 1988/89, 1989/90, and 1990/91) are currently available in the
Equal Area SSM/I Earth (EASE) Grid projection required for investigation of terrestrial snow cover,
and are compared in this study. Temporal patterns within each season are isolated through principal
components analysis (PCA). Component loadings indicate the strength of association between each
component and original time series image, thereby indicating the temporal persistence of the snow
cover pattern isolated by each component. SWE clusters within pentads are isolated through compu-
tation of the Getis statistic ( G* ). The Getis statistic, a local indicator of spatial association, provides
i
a measure of the spatial dependence of each pixel to the surrounding pixels, and also indicates the
relative magnitude of the SWE values in a given pixel neighborhood. Variable window size in the
computation of G* allows for spatial association to be computed at a range of distances from the
i
center pixel.
The passive microwave imagery and two analysis techniques provide a means of quantitatively
monitoring snow cover evolution, while also identifying a high degree of interannual variability in
the dominant spatial and temporal modes of Prairie snow cover. DJF 1988/89 is characterized by
persistent snow cover with noncontiguous spatial clusters; DJF 1989/90 is composed of spatially
coherent snow cover with low temporal persistence, while DJF 1990/91 is characterized by gener-
ally sparse snow cover. The results of this study confirm that synoptically sensitive data, which
passive microwave sensor can provide, are essential for the remote monitoring of Prairie snow cover.
The analysis methods explored can be utilized to provide significant climatological and hydrological
information.
1Waterloo
Laboratory for Earth Observations, Department of Geography, University of Waterloo, Waterloo,
Ontario N2L 3G1 Canada
2Pacific Forestry Centre, 506 West Burnside Road, Victoria, British Columbia V8Z 1M5 Canada
3Climate Research Branch, Climate and Atmospheric Research Directorate, Atmospheric Environment Ser-
vice, Downsview, Ontario M3H 5T4 Canada
104
Previous Page
