Geoscience Reference
In-Depth Information
catastrophic fire in the first place. We will never be sure if the absence of a fire,
which was initially given a high risk index value, is due to the overestimation of the
risk index and, therefore, due to the poor forecasting of the weather conditions for
an area. Conversely, the weather conditions could have been forecast correctly and
would, therefore, have led to the creation of a fire; however, no fire started because
there was no initial flame or spark present to ignite it.
For an index to be validated there needs to a situation in which the hazard
associated with lighting the fire is removed, and the only conditions that vary from
one fire to another are the meteorological conditions that influence the fires and the
way in which they develop. The solution to this issue is to ignite pieces of land that
have been chosen for testing, with each plot having similar slopes and vegetation
cover. The tests would be carried out during the most varied weather conditions.
The idea is to correlate the fire behavior (speed of progression, height of flames,
heat emitted, etc) with a particular index value. From there it would be clear that a
violent fire cannot correspond to a low-risk index value. This approach of using
experimental fires is carried out by our lab at the end of winter as part of a burning
program developed by the French National Forestry Office in the department of the
Alpes-Maritimes in the southeast of France. However, lots of these experiments
need to be carried out in the most extreme and, therefore, most dangerous conditions
so that all possibilities have been tested and can be protected and fought against
should a real fire of such an extent occur in the future.
The next step involves validating the recreation of the spatial field of the risk
index, by using any of the methods introduced throughout this chapter. The most
common approach involves excluding a certain of number of points (whose index is
calculated from meteorological data that has been recorded) from the experimental
sample (for example, the file on which the regression is calculated). This approach
also involves checking whether there are not too many residuals, with a residual
being the difference that exists between the actual recorded value and the value that
was predicted by the model.
Due to the fact that an index can be calculated directly but not measured directly,
the difficulty associated with validating it is not easy to solve. At this point,
validating the index in itself is more important than validating its spatialization.
In general terms, the use of geographical information is essential so that data
relating to the dangers associated with fires can be produced irrespective of the
method used to generate the information. Geographical information is used
differently, depending on the method that is used to generate the data, and often the
data are generated at different phases of the different methods. The risk is composed
with hazard and susceptibility that must be associated to spatial areas, and risk is
present only because of the relationship that exists between the vulnerability. As the
vulnerability must also be associated to spatial areas, the use of geographical
information is, therefore, required at all phases.
Search WWH ::




Custom Search