Estimating the Mean Areal Snow Water Equivalent
by Using Satellite Images and Snow Pillows
Thomas Skaugen 1
For operational flood forecasting and flood warning in Norway, it is crucial to be aware of the
amount and coverage of snow at all times throughout late winter and spring. For considering flood
hazards related to snowmelt in spring, knowledge of the amount of snow, or the mean areal snow
water equivalent (SWE), is particularly important. Remote sensing sensors cannot, with current
methodology, provide estimates of the mean areal SWE except at a very coarse resolution (passive
microwave sensors with 25-km resolution). This paper presents a methodology that combines infor-
mation from satellite images with that of the frequency of precipitation events, and gives an expres-
sion for the estimate of the mean areal SWE.
By modeling the snow accumulation process in time and space as sums of random gamma distrib-
uted variables, the mean areal snow water equivalent (SWE) can be estimated. In the methodology
we make use of the fact that sums of gamma distributed variables with a certain set of parameters
also are gamma distributed variables with parameters being functions of the original and the number
of summations. The measured snow depth/SWE at a point and at a certain time t can thus be seen as
the accumulation, or the sum, of snowfall events from the beginning of the snowfall season to the
time t. The integration of these points over an area at time t is seen as another summation.
From snow pillows and precipitation gauges, the value of daily accumulated precipitation/snow has
been found to be well represented by a two-parameter gamma distribution. This distribution has been
found to be representative for large areas. The number of events where the precipitation was accu-
mulated can be estimated from snow pillows situated in the area. The mean snow coverage over an
area, which represents the summation of the individual points over an area, can be derived from
satellite images represented in a GIS. The methodology is tested for eight satellite scenes and for two
nested catchments (4,723 km ≤ and 19,832 km ≤) in a mountainous area in southern Norway. The
results are compared to simulated snow reservoirs using a rainfall-runoff model, and are found to
agree well. Large discrepancies in the computed snow reservoirs between the proposed method and
the rainfall runoff model are found in late spring and are probably due to errors in the estimated mean
snow coverage derived from the satellite images.
1
Norwegian Water Resources and Energy Administration, P.O. Box 5091, Maj., N-0301 Oslo, Norway
112
Previous Page
