Академический Документы
Профессиональный Документы
Культура Документы
Technical Paper
Introduction
subject, with several core stages of disaster management before, during, and after the occurrence
of a disaster. There are numerous key stakeholders involved in emergency management, and the
scale of the efforts related to natural disaster management can range from local responses to
global efforts. As such, it is obvious that having perfect preparedness, response, recovery, and
mitigation efforts in place is impossible. Still, there are several different avenues that emergency
This project attempts to provide a new methodology for helping emergency managers in
the preparation stage of the management cycle. During this stage, emergency management
specialists attempt to develop the ability and preparedness to respond and recover to a natural
disaster event. This phase includes using geospatial tools and methodology to help understand
and interpret areas with greater risk and susceptibility to disaster events. One potential way to
identify these areas is through the use of composite indicators and modeling to assess risk.
The importance of a composite assessment is strongly linked to its interpretability and its
combination of multiple indicators of risk. Merz et al. points to this operationalizing of multiple
factors, stating that “using composite indicators bridges the gap between theoretical vulnerability
framework and decision-making within disaster risk management.” As they later mention,
composite modeling does not result in absolute or definitive statements, but shed light upon the
underlying indicators (Merz et al., 1080). The use of composite risk modeling is a popular
concept, although they are often fairly vague in both description and classification. Often the
composite index is limited to poor spatial resolution or broad rankings, such as the Rhode Island
Composite Exposure Indicator (CEI) approach, in that the approach is “flexible and the list of
indicator variables could be modified easily” (Multihazard Risk Assessment, 316). Additionally,
the creation of the composite risk map should be easily scalable by the locally perceived threat of
each natural hazard. It is important to note that the results of this project are intended to display
the methodology of developing a high resolution, generalized composite risk map rather than
provide significant insight as to the risk levels that were found throughout the case study area.
The selected case study for this project was the Los Angeles, United States county. This
region was selected due to two main factors: moderate levels of publicly available data and the
presence of enough potential hazards to develop a composite risk map. The availability of spatial
data provided enough information to be able to develop perceived risk layers, while the lack of
refined data available for use provided the requirement to make the methodology flexible enough
to adapt to other spatial data. This would ensure that the composite risk mapping methodology
would be able to be ported to other locations with ease. The second condition gave cause for a
composite risk map to be created, with earthquakes, fires, flooding, and landslides all as potential
natural disasters for the Los Angeles county region. While these hazards were used for this case
study, it would be simple to incorporate other hazards into the composite risk mapping
methodology.
The methodology for the creation of the generalized composite risk map incorporated
several different geospatial functions and processes to ensure the smooth conversion of
individual risk indicators into one composite map. The following functions and tools were
implemented:
● Clipping and masking of spatial data was used to limit the extent of datasets to the case
study area of interest. This was done for raw datasets of both polygon and raster file
types.
● All raw data was reprojected into the appropriate projection, which was UTM 11N for
● Any datasets that were initially of polygon file types were converted to raster format
The need for high spatial resolution for this project necessitated the conversion of any
generalized spatial data to raster format. Raster formatting is ideal for local composite risk
mapping, this allows for the finer combinations of individual perceived risk indicators into one
composite output.
For this project, a standardized workflow was used for creating all layers (figure below).
First the appropriate data had to be gathered for each layer. This is a vital decision to make as
the quality of the data highly affects the results. Next, each layer had to be converted to raster
format because generalized risk is calculated for all locations in the study area. Raster format is
best for phenomena that exist everywhere. The conversion process varies depending on the
format of the original dataset. Then all the rasters are reclassified into five categories with a
value of five being very high risk and a value of one being very low risk. This is done to
standardize each layer so when the composite is created, every layer will have equal weight
(unless intentionally weighted otherwise). Finally, all the rasters are added together and the
value is divided the number of layers. The values in this final output represent the generalized
risk of natural hazards using the 1-5 scale. For the case study of Los Angeles county, wildfires,
earthquakes, flooding and landslides were selected as the relevant natural hazards. The World
Bank’s ThinkHazard!, a tool that models natural hazard risk at a more global scale, was used for
guidance on this decision (Douglas et al. 42). It is possible that Los Angeles county is
susceptible to other natural hazards not included in this model, but they were deemed
insignificant or adequate data could not be found. In the figure below, drought and cyclones are
shown but they were not incorporated into the final version of the model. Drought was dropped
because its impact on human lives tends to be equal at the scale being used in this project.
Cyclones were not included because they rarely affect the study area.
Wildfire
The intentions of this project are to establish the framework for creating localized tools
for the generalized risk of natural hazards. The case study used is for Los Angeles County, but
the specific methodologies discussed below will differ by location due to different data sources
and different natural hazards. California is notorious for its wildfires which have been known to
cause significant destruction to property. Los Angeles County government has a data portal for
GIS data on several topics which would be expected to be invaluable for finding high resolution
localized data. However, much of their data on natural hazards was irrelevant to this
methodology or it linked to other sources. Through this platform, relevant data was found at the
California Department of Forestry and Fire Protection’s Fire and Resource Assessment Program
(FRAP). The dataset used from this source is titled Fire Hazard Severity Zone, and considers
fire frequency as well as potential fire behavior (California Department of Forestry and Fire
Protection 1). Conveniently this dataset was already available in raster format with five
assigned threat classes. Below the chart shows how the original dataset was reclassified, and
there are before and after images of this layer below. The primary negative aspect of this data is
its 100x100 meter resolution. The final output for the project is 10x10 meter (a fairly precise
resolution) so in the conversion process, the pixel size of the original fire dataset was made
smaller even though the source data was not gathered at this resolution.
Earthquake
The San Andreas fault runs through the northern portion of Los Angeles County,
increasing the area’s risk of human casualties and property damage from earthquake shaking.
The dataset used for earthquakes was gathered from California’s Department of Conservation,
and is called Low Frequency Shaking Potential. This dataset was chosen because the metadata
of this reliable source said that earthquake shaking at the 1 second period correlates well with
method would have been using a historic earthquake layer and estimating risk using the epicenter
and magnitude fields. However, that methodology ignores the underlying geology of the region
contained a field called g which represented ground shaking. The Polygon to Raster tool in
ArcGIS was used to convert the data into raster format with a specified cell size of 10 meters.
Because this dataset was originally made up of precisely drawn polygons, the cell size makes a
significant difference in the output. After this, the Reclass tool was used to break the data into
the five standardized categories. The g field ranged from 0.25 to 2.35, and because this data is
continuous rather than categorical, natural breaks was used to create the five categories. After
Flooding
Almost any location with a stream network or on the ocean is susceptible to flooding and
Los Angeles County has both of these characteristics. The Federal Emergency Management
Agency (FEMA) is arguably the most rigorous organization for flood mapping, although they do
not make their data easily accessible. After much searching, FEMA’s National Flood Hazard
Layer was obtained. This data is in polygon vector format, and the field that was decided to be
most relevant was Flood Zones (Federal Emergency Management Agency 1). Unlike the
earthquake data, the Flood Zones field was categorical. Therefore, the metadata had to be well
understood in order to properly assign the reclassified values. The chart below shows the results
of this decision making process. After running the Reclass tool, the flood layer results were
cross-referenced with a DEM and a stream network layer to ensure that the results made sense
(one would expect flood risk to be highest along streams and at low elevations). Flood
Modelling is a very developed field of study, and it would have been possible to implement a
more advanced system to determine flood risk. However, these capabilities were beyond the
Landslides are not often considered one of the major natural hazards, but California is
notorious for them as they cause significant damage to property and infrastructure. Out of the
four hazard layers of the study area, the landslide data has the lowest resolution (this can be seen
in the map below). All data available on this topic was global in scale, and the best compromise
found was the United States Geological Survey’s (USGS) Landslide Incidence and Susceptibility
Layer. Like the last two hazards, this was available in vector polygon format, and like the flood
layer, the data was categorical (United States Geological Survey 1). There were only four
landslide categories so the decision making process for reclassifying was not as difficult (see
chart). After converting to raster and running the Reclass tool, all four layers are ready to be
added together.
Composite Methodology
ThinkHazard! Tool was used to decide how to weigh the hazards. The chart below shows how
each hazard was weighted for the final composite. Earthquakes and Flooding are the most
concerning hazards for the study area so they are given the highest weight of five. Due to the
landslide layer’s relatively low quality of data and lower rate of incidence, it was weighted the
lowest at one. Looking at the final results (map below), the area of highest generalized risk can
be found along the San Andreas fault in the northern part of the county. Most of the southern
portion (where most people live) is relatively low, but there are interesting pockets of high risk
particularly on the Palos Verdes Peninsula. Other populated areas at high risk include Malibu,
The methodology developed for creating a generalized composite risk map fulfilled the
desired goals of this project. By combining multiple high resolution hazard datasets into one
composite risk indicator, it was possible to develop the equally local risk map. This process and
modeling was consistent and easily replicable, which satisfies the requirement to easily be able
to swap in different hazard datasets or transpose the model to a different area of interest. Still, the
development of composite risk maps is highly useful but limited. Merz et al. also come to this
conclusion, stating that they are “stepping stone[s] towards risk mitigation: by identifying the
most vulnerable sectors, risk management measures enhancing the resilience can be defined”
(1096). This project hopes to be an additional stepping stone, providing a convenient way to
identifying these vulnerable locations for improvement through modeled composite risk maps.
There are several components of this project that could be enhanced in the future,
although many of these improvements are reliant upon external sources: higher resolution
analysis, incorporation of more hazards, and providing an interactive interface. The first
improvement is highly dependent upon the availability of data; however, it would be beneficial
to see the full potential of an extremely high spatial resolution composite risk map. Additionally,
it would be both increasingly complex and useful to incorporate other hazards into the composite
map. Finally, providing a system for less tech-savvy emergency managers to interact with the
composite risk map could provide a way for more individuals to identify areas that are at high
risk of natural disasters. A separate project that would complement this composite risk map
modeling could be refining the ability to classify and scale natural disaster potential impacts.
References
California Department of Conservation. (2016). Earthquake Shaking Potential for California.
http://www.arcgis.com/home/item.html?id=eee5a39483604a4c8861e6e6433e009c
California Department of Forestry and Fire Protection. (2016). Fire and Resource
Assessment Program.
http://www.fire.ca.gov/fire_prevention/fire_prevention_wildland_zones_development
Douglas, John, et al. (2016). ThinkHazard!: Methodology Report. World Bank Group.
http://thinkhazard.org/static/2f1be9baecafa08c3823a654028d2dcd/documents/thinkhazar
d-methodology-report_v1_8final.pdf
Merz, M., Hiete, M., Comes, T., & Schultmann, F. (2013). A composite indicator model to
assess
natural disaster risks in industry on a spatial level. Journal Of Risk Research, 16(9),
1077-1099. doi:10.1080/13669877.2012.737820
https://www.fema.gov/media-library-data/20130726-1545-20490-4252/mhira_ra.pdf
Rhode Island Emergency Management Agency (2016). State of Rhode Island Hazard
United States Geological Survey. (2016). Landslide Overview Map of the Conterminous United
States. http://landslides.usgs.gov/hazards/nationalmap/