Spatial Analysis of HAWAII’s Natural Hazards

Lab Reports, Maps, Visualization
Image 1 – Watercolor Carto map of the Hawaiian Islands


This March the Hawaiian Islands saw higher than average rainfall and at a dam overflowed, rivers rose to historic heights, and there was major flooding in several communities. The Opaeula Stream near Haleiwa on Oahu’s North Shore rose from 4 to 16 feet in one day. On Maui, the Kaupakalua Dam was breached and caused evacuations and washed out streets. Additionally, the North Shore of Oahu, home to the world famous Banzai Pipeline wave break, has seen years of coastal erosion and elevated tides. Climate change affects sea level rise and the majority of Hawaii’s population live along or near the coast. I seek to calculate the total population of people who live in three major coastal flood zones across the islands. The following geographic visualizations are preliminary work for a more detailed final project.


Response to the flooding on Maui and Oahu included anger at the local government for lack of oversight of potential flooding hazards. Hawaii is full of rivers, shown by the screenshot below in image 2. On the mainland, New Orleans has been a focus of GIS and spatial analysis since Hurricane Katrina in 2005. There are many interesting maps the show the flooded areas, high risk area, and projected flood zones due to sea level rise and increased storm intensity. Images 3 and 4 below are examples of those visualizations. Global sea level rise has already caused climate migrants and refugees to flee from their homelands. As the climate continues to warm, the ocean will continue to rise, putting more communities at risk of flooding and disappearing altogether (image 5).


Locating data of interest that doesn’t take days to clean is often the most time consuming part in this type of project. This time, I got lucky. Hawaii’s Statewide GIS Program from the State Office of Planning hosts a beautiful, and free, data portal of shapefiles. I used multiple datasets from this page: lava hazard zones, fire station locations statewide, coastal flood zones, erosion lines, economic impact of floods, and a few that I didn’t include in this project: marine unexploded ordnance and NOAA marine protected area. In addition to this heaven sent data, I used the United States Census Bureau’s data tables. From there I found census tracts, population, housing information, and poverty numbers. I’ve used this site in another class and have experience searching for data and organizing it in Google Sheets, it is user friendly if you use the advanced search. The majority of the work was done in Carto, which I also used to display the final visualizations. I found it easy to figure out due to the immediate visual feedback, and once I realized that you could edit the data tables. Carto would be even better if they added a “back/undo” button. I also utilized QGIS for data manipulation and adding CSVs to shapefiles.


Step 1 – Census Tracts. I knew that I wanted to use census data and place it on the map by census tract. There are more census tracts than zip codes or counties and produces more interesting data. Initially, there was a section beyond the coastline and circling the island that I needed to remove. I uploaded the shapefile into QGIS and in the attribute table I deleted the tracts that had “0” land elevation data. I exported it as a new shapefile, replaced the original in my folder and compressed the folder so that I could upload it back into Carto. This new shapefile had strange edges that didn’t look quite right (map 1). I downloaded a coastline shapefile from Hawaii’s GIS program and in QGIS clipped the census tracts file to the coastline file. Now I had a clean shapefile of the islands (map 2). I did most of my work focused on Oahu.

Step 2 – Make it pretty. The watercolor Stamen baselayer is beautiful (image 1), but not practical for showing technical data. The HERE satellite day baselayer is also interesting. I used it in map 3 to display the tract altitudes with triangle symbols. It’s visually cool, but the information displayed is redundant. Ultimately, I chose the Carto Voyager base layer for most of my maps because of the minimal labels that are on top of the data and simple color (map 4).

Step 3 – Tell a story. The following slideshow contains maps of various stories as I tried to analyze the data and make it mean something interesting. It was also a chance to play around with color, gradients, transparency, and legend properties. The coastline zones in map 5 are striking, but I didn’t understand the difference until I found a report with the chart in image 6. I used a cool color palette with high contrast colors to distinguish the zones. To make the data more compelling I added a coastal erosion shapefile and zoomed in on Sunset Beach on Oahu’s North Shore where the annual Billabong Pipe Masters surf competition is held. The beach changes in size throughout the year but is progressively shrinking. To the extent that some million dollar homes are at risk of falling into the water (locals named a surf spot after them, Monster Mush). Map 7 is another attempt to tell more of a story with the coastal flood zones with the addition of potential economic loss data. It looks cool but it’s difficult to see what’s really going on. Map 8 is a full view of the coastal flood zones around Oahu, as a single color, done in QGIS. It’s hard to tell what’s going on here but I want to continue to work with this map in my final project and find population totals within each zone.

Step 4 – Make it interactive. Below are two different visual attempts to tell a story with volcano and fire station data on the big island of Hawaii, called Hawaii island. I used a shapefile of volcano lava flow hazard zones, a shapefile of fire station locations, a baselayer map that shows elevation and terrain, and a CSV that I made from Volcano peak data. In Carto map 1, I added analysis to the hazard zone layer with the SUM of fire stations. I used a soft green gradient to display the zones with the most fire stations, in theory that can respond to lava flow fires more quickly. Something went wrong in the analysis because there are not 261 fire stations near Kilauea, I’m not sure what happened there. The analysis and widget parts of Carto stumped me. I added pop-ups with volcano details and fire station names, displayed the stations with flame icons, and made the zones transparent to show the elevation of the basemap. In Carto map 2, I changed some cosmetic variables including color scale, icons, and labels, and colored the zones by highest to lowest hazard. Because the value was by zone the legend displayed 9 icons and labels instead of a nice gradient scale, so I didn’t include the zones in the legend. During a peer review, my classmate suggested different colors for the gradient, so I changed them to what you see instead of red-purple. I agree with her that it makes the scale more intuitive since it’s not explained in the legend.

Carto map 1 – Lava flow hazard zones on the Big Island of Hawaii. Analysis of # of fire stations within each zone.
Carto map 2 – Lava flow hazard zones and fire station locations on the Big Island of Hawaii. Colored by hazard zone intensity.


The following Carto maps are the finished product from all of the experimentation above. They are clean, straightforward visualizations of poverty rates and housing characteristics in Hawaii. Carto map 3 is colored by poverty rate. I initially added the poverty data as a csv and added geocoding analysis, but it only displayed as points. Then in QGIS, I joined the csv to the census tract and population shapefile. The data only had the total number of people living under the poverty scale, so I added a formula in the attribute table to calculate the percentage. I included pop-ups with the total population and poverty population numbers to communicate more of a story. For example, census tract 9806 has a poverty rate of 0%, but the pop-up tells that the total population is only 87. It’s probably a national park or something. Carto map 4 is similar but is colored by population totals and the pop-ups display the housing owners and renters. I couldn’t figure out how to display both owners and renters on the map without pop-ups. I also could have calculated the owner vs. renter rates, but making new attribute table fields in QGIS is a bit of a headache. More steps to continue for my final project.

Carto map 3 – 2015 poverty rates in Hawaii by census tract
Carto map 4 – Housing owners v renters in Hawaii


My biggest takeaway from this project is that I want to spend more time on it. There is a lot of potential to add more analysis to the layers and tell a more focused and rounded story about who lives in flood and lava hazard zones. It might be easy to assume that the hazard zones include more impoverished communities, but sea level rise flooding will also affect the millionaires on the beach. That’s why I also want to include river flood zones because that might show the disparity of natural hazards and economic status. It won’t change where people live, but it might affects emergency planning and response resources. Floods, tsunamis, volcanoes, and lava flow are not easy to predict, but who am I kidding I’d move there in a heartbeat despite the natural hazards.


  • The following datasets and shapefiles can be found at Hawaii geospatial open data portal:
  • Coastline data. Source:  USGS Digital Line Graphs, 1983 version.  Extracted from USGS Digital Line Graphs by Office of Planning staff, 1988.  
  • Volcano lava flow hazard zone data. Source:  USGS, 1991.
  • 1 Pct Coastal Flood Zone with 3.2 ft Sea Level Rise – Statewide. Source: Hawaii statewide GIS program.
  • Fire Stations – Island of Hawaii. Hawaii statewide GIS program / office of planning
  • Coastal erosion. Anderson, T.R., Fletcher, C.H., Barbee, M.M., Romine, B.R., Lemmo, S., and J.M.S. Delevaux (2018). Modeling multiple sea level rise stresses reveals up to twice the land at risk compared to strictly passive flooding methods, Scientific Reports, 8:14484, doi: 10.1038/s41598-018-32658-x. Anderson, T.R., Fletcher, C.H., Barbee, M.M., Frazer, L.N., and B.M. Romine (2015).
  • Potential economic loss. Data compiled by the Pacific Islands Ocean Observing System (PacIOOS) for the Hawaii Sea Level Rise Viewer hosted at For further information, please see the Hawaii Sea Level Rise Vulnerability and Adaptation Report:

Leave a Reply

Your email address will not be published. Required fields are marked *