Damage Assessment - Hurricane Sandy

Module 5 for GIS Applications continues our focus on Hurricane Sandy and explores damage assessment for the storm's impact in the Garden State.

Our first task was to create a formal hurricane track map showing the path Sandy took from the Caribbean Sea to the Northeastern U.S. The symbology uses custom color coded coordinate points showing the hierarchy of storm intensity. Included are the maximum sustained winds and the barometric pressure shown in 12 hour increments to improve legibility.

The next section of lab 5 was the creation of a damage assessment survey using Survey123. This was a pretty straightforward process, with options to add multiple choice questions pertaining to damage to be documented, a field for describing that damage surveyed, the option to include an image or have the mobile device take a photo, and a required location setting either through GPS or map locator. Following the form creation, we determine what the survey application does after the submission of a completed survey and we set the restrictions on what an individual viewing the survey can see.

Our next task is the preparation of raster data of air photos showing an area of New Jersey both before and after Superstorm Sandy. An array of .SID raster images of pre-storm photos created the first mosaic dataset using geoprocessing. .JPG images of the post-storm photos were compiled into the second mosaic dataset.

With both mosaic datasets in place, we revisit the Flicker and Swipe tools (located in the Compare group below the Mosaic Layer tab), which were previously used in the Remote Sensing course, to alternate the display between the pre and post storm imagery. These are both fast methods to visually compare the two imageries.

An example of the Swipe tool showing pre-storm imagery above and post-storm imagery below

Step 3 of the lab focuses on the creation of data. For this, we revisit the concept of Domains previously covered in Intro to GIS last Fall. Attribute domains constrain values allowed in an attribute for a table of feature class. Domains create a rule set of acceptable attribute values, or in the case of Module 5, a range of integers associated with predefined aspects of damage assessment:

Helping to ensure data integrity, domains limit the number of acceptable values for a field.

Attribute domains are store in a geodatabase. They can be utilized by multiple feature classes, tables and subtypes in a geodatabase. Through Catalog in ArcGIS Pro, Data Design>Fields, these can be added to an existing feature class.

Using the aforementioned air photos showing Seaside Heights, NJ before and after Superstorm Sandy, we were tasked with conducting damage assessment within a study area of parcel data using the preset domains for the various damage categories. Symbolization uses a continuous color scheme from green for no damage to red for destroyed.

Damage assessment study area for Superstorm Sandy at Seaside Heights, NJ

Given the four domains of Structure Damage, Inundation, Wind Damage and Structure Type, each parcel within a seaside neighborhood was evaluated for damage based upon the two air photos. This was a tedious task due to relatively low image resolution and long shadows in the post-Sandy aerial imagery. Without in-situ data collection, evaluating parcels for wind damage was impractical given details such as missing roof shingles was not possible.

Expanding our analysis, we aggregate the damage assessment points into buffers of within 100 meters of the coastline, between 100-200 meters and between 200-300 meters. Using the Multi-ring Buffer geoprocessing tool, created the three storm surge zones. Proceeded to run a Spatial Join on the Structure Damage point file with the MultipleRing Buffer polygon file to quantify the damage type by buffer zone. The Summary Statistics geoprocessing tool does the tabulation for us:

The final results of our damage analysis confirms the penetration of Superstorm Sandy's storm surge varied as the distance from the coastline increased from 100 to 300 meters. Structures facing the ocean were generally pulverized, while buildings located around 300 meters inland fared much better, some with seemingly no visible damage. This damage estimation appears to be consistent along other parts of the barrier island where the elevation and slope are similar. Exceptions were noted, such as further south of the study area in Seaside Heights, New Jersey, where the barrier provided by a boardwalk and two piers protected adjoining neighborhood areas from the storm surge.