Geoscience Reference
In-Depth Information
2.4. Geographic information used in statistical interpolations: advantages and
disadvantages
In addition to the problems dealt with in the previous section, other problems
also arise and these problems limit or prevent the use of spatial statistics in the
interpolation process. Some of them will be dealt with by studying the following
four points:
- studying the problems linked to the density of climatological stations in certain
areas;
- studying the information source and the quality of interpolation. In this case
the following issue is raised: how is it possible to convert data that provides
information on land cover into quantitative data that can then be used within the
regression models;
- studying the problems that are linked to the use of certain estimators (for
example, distance from a geographic object). These estimators are supposed to play
a role in the spatial distribution of certain climate factors;
- spatial variables and scales.
2.4.1.
Density of climatological stations
The density of climate data available influences the way in which the data itself
is interpolated. If we look at Figure 2.1 again it is clear that the density of
climatological stations, from where rainfall amounts and duration of sunshine was
recorded, is completely different.
2.4.1.1.
Rainfall amounts
Rainfall amounts are recorded by 3,165 rain gauges in France (see Figure 2.5).
For recordings taken in the month of August over a 30 year period from 1971 to
2000, the average rainfall was 590 mm, with a series that ranges from 9.9 mm (in
Corbara, Corsica) to 145.8mm (in Lamoura, Jura). The large number of
measurement points means that it is possible to analyze rainfall amounts with a high
resolution because all of the spatial configurations are sampled.
Figure 2.6 compares the frequencies of 13 different altitude classes, and deals
with two distinct populations:
- the first one is a corpus of 8,833,676 pixels with a resolution of 250 m and
which covers the whole of France;
- the second one is a corpus of the 3,165 pixels with a rain station.
Only two of the different altitude classes produce slightly different results from
one another. The altitude class, which includes altitudes of 50 m is over-
represented (11% of all rain stations come from this class, whereas this altitude class
makes up only 5.4% of the total altitude range in France). Conversely, the higher-
level altitudes are under-represented (only 0.1% of all rain gauges come from the
