|
sented by the 7 sensor measurements in order to obtain
consistent dataanda general viewof the snow profile along
the slope. This processing follows four steps: afirst smooth-
ing of extreme data, checkingof real presenceof at least 4
sensor's data, missing values are replaced by a moving
averagemethod (only if no more than one value is missing
after another) andafinal smoothing by theprevious method
for the whole period (figure 4).
In addition to this processing, some statistical parameters
can be analysed. For example, it is interesting to compare
each sensor data with the smoothing average in order to
|

display their particularity. On the other hand, the stand-
ard-deviation of the snow-depth profile can be calculated
at each time interval. It appears that when the standard
deviation is high, the behaviour of the sensors is obviously
different. The snow surface does not undergo the same
modifications all along the slope.

Focus on a blowing snow period

The major aim of this test is to get information on the cor-
|
 |