56
A. Hysa
The second phase consists of a multi-criteria inventory procedure (Hysa et al.
2017), through which there are measured absolute values of each criterion joining
to a cumulative multi-criteria inventory. As it is expected, the range of values of
different criteria are diverse among them. For example, the range of values for criterion “distance to urban centers” (ranging from 0 m to several km) is much different
than the values range of “maximum temperature” (ranging from 0 °C to at most
50 °C).
The variety among absolute values of different criteria is normalized during the
Analysis phase of the workflow. This is achieved by redistributing them into 10
classes based on Jenks natural break method as provided under QGIS layer properties.
The Jenks natural breaks classification method, also known as the Jenks optimization
(Jenks 1967), is a data classification method aimed to define the best reorganization
of values into sub-classes by reducing of the variance within classes and maximizing
the variance among them (Stefanidis and Stathis 2013). This is done by minimizing
the standard deviation within each class and make the most of the standard deviation
between classes (McMaster and McMaster 2002).
Since the analysis phase is the last one before the final double Indexing phase, the
criteria are clustered into two main groups based on their relation with the ignition
and spreading phases of the wildfire. Some criteria may have no effect on either
ignition or spreading phase of the forest fire. As a result, there are two separate sets
of criteria for WIPI and WSCI indexing.
Furthermore, based on the assumption that not all criteria have the same impact
on the wildfire events, criteria of each set are introduced into a weighting process
based on Analytical Hierarchy Processing (AHP) (Mu and Pereyra-Rojas 2017).
At the final stage, Jenks classification values of each criterion are introduced into
the equation (Eqs. 1 and 2 from Table 4.1) multiplied by their respective impact
factor as weighted via AHP. Finally, each representative point is loaded a final value
of WIPI and WSCI. In principle, the locations possessing relatively highest values
indicate for areas of high wildfire ignition and spreading risk.
4.3 Results and Discussions
The ten steps introduced above have been applied in the case of Durana metropolitan
area, with each of four main phases lead to distinct results. First, the preliminary work
and inventory phases both deliver a final shapefile layer of reference points, which
consist of all inventory measured values for 13 criteria. As a result, raw statistical
data on each criterion is produced.
Then, during the phase of the analysis, some inferential statistical data about
the value distribution of each criterion are generated. During this phase, a set of
maps that show the spatial distribution of Jenks classification of values per each
criterion are produced. Besides, during this phase the weighted impact factor values
via AHP are produced by being calculated as the average values as evaluated by eight
professionals.
A. Hysa
The second phase consists of a multi-criteria inventory procedure (Hysa et al.
2017), through which there are measured absolute values of each criterion joining
to a cumulative multi-criteria inventory. As it is expected, the range of values of
different criteria are diverse among them. For example, the range of values for criterion “distance to urban centers” (ranging from 0 m to several km) is much different
than the values range of “maximum temperature” (ranging from 0 °C to at most
50 °C).
The variety among absolute values of different criteria is normalized during the
Analysis phase of the workflow. This is achieved by redistributing them into 10
classes based on Jenks natural break method as provided under QGIS layer properties.
The Jenks natural breaks classification method, also known as the Jenks optimization
(Jenks 1967), is a data classification method aimed to define the best reorganization
of values into sub-classes by reducing of the variance within classes and maximizing
the variance among them (Stefanidis and Stathis 2013). This is done by minimizing
the standard deviation within each class and make the most of the standard deviation
between classes (McMaster and McMaster 2002).
Since the analysis phase is the last one before the final double Indexing phase, the
criteria are clustered into two main groups based on their relation with the ignition
and spreading phases of the wildfire. Some criteria may have no effect on either
ignition or spreading phase of the forest fire. As a result, there are two separate sets
of criteria for WIPI and WSCI indexing.
Furthermore, based on the assumption that not all criteria have the same impact
on the wildfire events, criteria of each set are introduced into a weighting process
based on Analytical Hierarchy Processing (AHP) (Mu and Pereyra-Rojas 2017).
At the final stage, Jenks classification values of each criterion are introduced into
the equation (Eqs. 1 and 2 from Table 4.1) multiplied by their respective impact
factor as weighted via AHP. Finally, each representative point is loaded a final value
of WIPI and WSCI. In principle, the locations possessing relatively highest values
indicate for areas of high wildfire ignition and spreading risk.
4.3 Results and Discussions
The ten steps introduced above have been applied in the case of Durana metropolitan
area, with each of four main phases lead to distinct results. First, the preliminary work
and inventory phases both deliver a final shapefile layer of reference points, which
consist of all inventory measured values for 13 criteria. As a result, raw statistical
data on each criterion is produced.
Then, during the phase of the analysis, some inferential statistical data about
the value distribution of each criterion are generated. During this phase, a set of
maps that show the spatial distribution of Jenks classification of values per each
criterion are produced. Besides, during this phase the weighted impact factor values
via AHP are produced by being calculated as the average values as evaluated by eight
professionals.