Introduction
The Federal Emergency Management Agency (FEMA) has four main phases of disaster management which include mitigation, preparedness, response and recovery. In the event of a disaster in Salt Lake County, the Emergency Management (EM) office would be the first responders. To efficiently use their manpower and resources during the response phase, the EM office wants a hazard reporting application. This application is meant to be a quick data collection tool to supplement the comprehensive survey form that FEMA developed for the recovery phase. This project will be divided into three main parts: the data collection application, the dashboard, and risk assessment maps. The data collection application will use ESRI’s ArcGIS QuickCapture platform. The QuickCapture survey will speed up the initial damage assessment and allow for greater coverage of the affected area. The dashboard will be customized to display the desired hazard information in real time. An organized dashboard will help the EM team to distribute resources across the disaster area. The final part of the project is the risk assessment maps for the mitigation and preparedness phase of emergency management. Risk assessment maps are a valuable resource when conducting the mitigation and preparedness phase. They assist city planners to identify buildings and areas that may be vulnerable in a disaster. For Salt Lake County, earthquakes are the main threat and mitigation is very important to protect infrastructure and save lives.
The objective is to develop a customized QuickCapture form to be used for basic information gathering during emergencies. The current survey forms do not cover all the desired hazards for Salt Lake County. The EM team needs customized hazard survey forms for quick damage assessments. The QuickCapture application will interface with a web map and dashboard for quick visualization of a disaster area. This visual interface will allow the County to easily deploy FEMA resources to areas in need. The application/dashboard combination will supplement the FEMA Survey123 for ArcGIS form used for comprehensive damage assessments. The final goal is to test the QuickCapture application in The Great Shakeout Drill on April 16th. This will be the first wide scale use of the application/dashboard to see how well it functions.
The risk assessment maps will be split into three different groups. There will be a set of maps displaying unreinforced masonry structures in Utah townships. Another set will display Emergency Management employee housing versus shake and liquefaction potential. Lastly the predicted earthquake damage from the FEMA Hazus earthquake model will be compared with the actual damage assessment from the March 18th earthquake.
Methodology
The first step was to acquire the hazard categories from the Salt Lake County Emergency Management team. These categories were separated into damage survey, hazards, human interaction and mission specific groups. There were two main layers that needed to be developed with ArcGIS Online for the QuickCapture application and dashboard to function. The point layer will hold the collected hazard points and the line layer will contain the tracks from each user. The point layer was more complex because domains had to be created for each hazard category. It was important to create domains to properly link the hazard categories to each QuickCapture button. Without a hazard type for each button the point will not link to the correct symbology. The application was setup by groups that divided the hazards into damage survey, hazards, human interaction and mission specific categories (Figure 1). This makes it easier to select the desired button without having to scroll through every hazard. Each button has to be formatted with the proper layer and domain. Using ArcGIS Online the layer symbology was formatted and each point was tested to ensure the input data was from QuickCapture.
The basic template for the dashboard started with a map and simple counters. The most important part was to make sure the correct layers were linked to the dashboard map. After making the basic template, it was time to acquire the categories and information that the EM team wanted the dashboard to display. Feedback on the first dashboard was received and the decision was made to make additional dashboards. There would be three dashboards total, one horizontal and two vertical. The horizontal dashboard was a simple overview with counters, lists and pie chart. The overview covers some basic information about hazards, structures and human interaction (Figure 2). The other two vertical dashboards were for specific information either structures or human interaction(Figure 3-4). The most important part was for these dashboards to be visible across a room. The size of the text had to be increased to make the headers and counters readable. Once the dashboards were completed to the EM team standards additional testing was performed. One of the first changes was to adjust the refresh rate of the layers from manual to every two minutes.
The risk assessment part of the project was split into three separate map sets. The first was a map set for six different townships identifying unreinforced masonry structures. The EM office supplied the structure layer that contains all the structures within Salt Lake County. The first step was to use a definition query to select only unreinforced masonry structures which were coded URML and URMM. The selected data was then clipped to the township boundary layer. After adding the township boundaries a map layout was created for each township (Figure 5).
The second risk assessment map set was for employee housing locations to determine the risk of shake and liquefaction during a 7.0 magnitude earthquake. An Excel file with employee housing information was acquired and loaded to ArcGIS Pro as a table. The table was geocoded using the given addresses. After the table was converted to points the shake and liquefaction layers were added. The shake layer was from the USGS ShakeMap scenario of a 7.0 magnitude earthquake and the liquefaction layer was provided by the EM team. Shake and liquefaction fields were added to the housing point layer. A system of definition queries, cross referencing, select with lasso and the field calculator was used until each point received a shake and liquefaction value. Maps were developed to identify areas where high shake and high liquefaction zones overlapped (Figure 6).
The final risk assessment task was to run the FEMA Hazus earthquake model and compare the results to the actual damage assessment surveys. The data layers were collected after loading the Hazus software. To match the required inputs for Hazus, fields and values had to be added to many of the layers. The Hazus program begins with setting the study region and adding the shake layer. The study area included all of Salt Lake County. For this run, the model used the USGS ShakeMap for the Magna 5.7 earthquake on 18 March 2020. The liquefaction layer only required the conversion of “liquefaction potential” from categorical to numerical. The data layer that took the most work was the user-defined structures layer. This layer had a lot of fields that needed to be converted or added to accommodate the Hazus software. The most time consuming field to create was occupancy because there were over 160 property types to be convert into 33 Hazus categories. The fields that only required simple field calculations were currency, area, Latitude and Longitude. The only error was the inability to post process the user-defined structures layer. The model was still able to run it simply did not use that layer in the analysis. The outputs from the Hazus model were then compared with the public survey feedback from the March 18th earthquake (Figure 7).
Results
The results of the QuickCapture application and dashboards were very promising. The application responded great under testing conditions with only some minor issues. When the horizontal accuracy on the application is 50 feet or higher the collected data has inaccurate location information. The inaccuracies may be corrected by using mobile hotspots instead of the devices network. This should improve the reliability of the points and tracks even inside buildings. The final test was to assess the offline capabilities of the QuickCapture survey. Once the initial login was completed online, the devices airplane mode was turned on. Then track and hazard data were recorded and stored on the device until service was regained by turning off air plane mode. The recorded values were then submitted and check for accuracy. Both the points and lines were accurate using QuickCapture’s offline capability.
The track portion of the application was tested during the March 18th earthquake. Prior to the earthquake there had not been time to teach the various departments about the new application. Although it’s unfortunate that the full application was not used, it was really meant for a larger scale disaster. It was not meant for individual structure damage assessment. The resulting tracks showed a number of places where poor service made the tracks zigzag. Even though only a part of the application was used it was still valuable to receive feedback.
The results of the risk maps could be used for the preparedness or the recovery phase of emergency management. If the township has the funds to retrofit the required buildings then changes can be made prior to a disaster but otherwise the township can apply for grant funding to rebuild structures after a disaster. Each township has to develop an emergency management plan to identify all risk areas in order to receive grant funding and these maps will help that process.
The housing risk assessment maps will help the EM and Unified Fire Authority (UFA) to determine the safety of their employees. The maps show a combination of housing locations and how they intersect shake and liquefaction zones. This analysis will help identify employees who may not be able to respond in an emergency based on their location. The Hazus model was an extremely challenging program to use but the results of the comparison were interesting. There are four maps comparing the severity of damage predicted by Hazus to the actual damage survey locations. The Hazus distribution for light and moderate damage followed the actual damage very closely with only some minor deviations. Even though the user-defined structures did not work properly the results were still very good. Both damage estimates closely follow the shake contours. The comparison proved to be quite accurate given the actual damage survey points.
Conclusion
This project required a diverse range of GIS skills. It was challenging to work with new programs but the experience was priceless. The focus of the project revolved around ArcGIS Online to deploy the application and dashboard. A lot of the layers were edited using ArcGIS Online which was a big change from other projects. Having to use so many different tools within ArcGIS Online was a great experience. Besides using ArcGIS Online the Hazus earthquake model was employed to run other analysis. Although the results were pretty accurate, the task of loading and running the model were extremely difficult. It was time consuming and some of the analysis did not work. Given the time it took to develop the layers it may not be worth it to run this analysis. It was a good experience but between the software and the layer development the challenges were extensive.
Skills
- Project Management
- GIS Analysis
- Project Design
- GIS Workflow
- Cartography and Graphic Design
- Communication
Melissa Gfeller 