Using Non-authoritative Sources During Emergencies in Urban Areas - Computational Approaches for Urban Environments - page 339

Fig. 14.7 Classification of flooding (high, medium, low flood severity/damage) in New York using
an artificial neural network

Because the inundated schools, USGS, and Civil Air Patrol data represented
maximum flood extent, it was possible to generate only one layer from each dataset;
therefore, these data were used for both days. The initial training and testing datasets
produced results indicating flooding along the coastlines of New York City with the
greatest damage identified in lower Manhattan and southern edges of Brooklyn and
Queens (Fig. 14.7 ).

14.2.3 Damage Assessment After Emergencies

After an emergency, remote sensing and volunteered data can be employed to
provide a damage assessment. In this particular work, the official FEMA flood
map is color coded to show not only which areas have been flooded but also which
areas have been most affected. In addition, the damage assessment surface is then
used to identify roads which may be compromised or may require site inspections
(Schnebele et al. 2013 ).
Crowdsourced data (CAP photos) and VGI (YouTube videos), which are il-
lustrated in Fig. 14.8 b, are fused together using a kriging interpolation. Kriging
allows for spatial correlation between values (i.e., locations/severity of flooding)