Geoscience Reference
In-Depth Information
Fig. 4
Spatial distribution of the standard deviation of normalized least risk path lengths
in size. Taking into consideration that the total path length of both shortest and least
risk paths in indoor space are already quite short (109.22-113.69 m with standard
deviations of 45.89 and 48.74 m respectively) due to the restricted building size, the
found limited differences are even more intensified as a whole. These results point
to an at first sight equivalent path choice by both algorithms, implying that (1) either
the shortest path algorithm is already selecting paths that are least risky to get lost on
or (2) these results could also be a first indication that the least risk path algorithm
is actually not calculating routes that are less risky and as such might not be well
defined for use in indoor spaces. This will be investigated in the following paragraphs.
A second analysis focuses on the internal variability of the results over the
entire network. More specifically, we want to analyse whether certain areas in the
building return a significantly different result compared to the average. Figure
4
visualizes the spatial distribution of the standard deviation for all least risk paths
starting in that point. The standard deviations have been classified in five quin-
tiles (five classes with equal cardinality), similar to Duckham and Kulik (
2003
)
analysis. Low standard deviations (i.e. blue data points) indicate starting points
with little variation between their least risk and shortest paths in terms of total path
length. Figure
4
shows that these points with low standard deviations can mostly
be found on the first floor and in lesser connected areas of the building. The higher
standard deviations (red data points) generally occur on the ground floor in denser
connected areas and around staircases both on the ground and first floor. This
greater variability can be interpreted as a result of the deviations of the least risk
path from the shortest path being more pronounced at rooms with many options
like around staircases where paths can be significantly different in the final route.
Starting locations within isolated areas (e.g. on the first floor) have no option but
to traverse similar areas to reach a staircase and deviate from there onwards.
