|
Is this map
OK? It shows that there is snow where I live, but I see through
the window that there is now snow! To answer those questions we
have to know a little bit more about how the map is developed.
The map is
produced within the The Interactive Multisensor Snow and Ice Mapping
System (IMS) project, based on data collected with various satellite
sensors. The data are interpreted manually. The sensors observe
Earth mostly daytime, when the Earth's surface is illuminated by
the solar light. With no clouds it is quite easily to see the ground,
and decide if it is covered with snow, or not. The problem starts
when clouds appear, as they cover the planet's surface, making impossible
to indicate where snow can be.
On IMS maps
information on snow cover extent is continuous - there are no gaps
in the data due to clouds. How is it possible? IMS' philosophy is
as follows. Suppose that one day a pixel is marked as 'snow'. Observation
on a second day confirms that status (pixel still classified as
'snow'). However on the third day there were clouds that prevented
the observation of land surface. In such case, the IMS project analyst
maintains the status of 'snow' as long as clouds occur. It can be
a day, or two, or week until next cloud-free observation will be
available. Then the analyst is able to reinvestigate the status
and maintain 'snow', or change to 'no snow' if appropriate.
In Poland (especially
winter-time), it happens that the whole country is covered by the
clouds for few consecutive days. If such situation takes place IMS
data cannot be updated on a regular basis. Thus, even if the map
was generated yesterday, this does not mean that yesterday a satellite
was able to observe the actual extent of snow cover in Poland. One
can note a consequence of that fact when looking at the plots of
snow cover extend - sometimes the line seems to be 'frozen', indicating
exactly this same fraction of country covered by snow during consecutive
days.
Sometimes the
discrepancy between map and reality is driven by the nature of snow
cover. Consider following situation. At some location there was
a sleet during night time. There was not enough now to make a snowman,
but enough to cover the surface with thin layer of wet snow. During
the day the temperature rises, and the snow melts. In the afternoon
there is not a trace of snow. This ephemeral snow cover could be
missed by the satellites at all: satellite could fly over the area
not in the morning but in the evening.
Another aspects
is the misclassification of 'snow' as 'no snow' (and vice versa)
by the analyst, who investigated the data. Since the decision is
made by specially trained personnel, the risk of misclassification
is significantly smaller, than for purely automatic algorithms,
however - it still exists.
Now, taking
into consideration all above-mentioned circumstances, you are ready
to draw conclusions from the map.
|
