Geographic Vulnerability Analysis Exercise

Vulnerability of Population to Flood Risk in Malawi
Vulnerability of Population to Flood Risk in Malawi

My experience reading national climate change adaptation plans from developing countries is that their vulnerability analysis and synthesis could be improved with geographic analysis! In response, Eliza Drury and I reviewed and developed guidelines for Geographic Vulnerability Analysis in Developing Countries. I took Eliza’s geographic vulnerability analysis work on Malawi in ArcGIS and translated it into an exercise in open-source QGIS: GIS software any developing country planner or NGO can afford! The exercise is designed for anyone who has already taken an introductory course in any GIS software. I gave the exercise to my seminar in Geographies of Climate Change Adaptation and Development in Spring 2016 and edited it based on student feedback. Several students then successfully applied GVA in their own research projects.

Teen Pregnancy

My wife, Kate, is currently interviewing Rhode Island school nurses to learn about school nurse communication with primary care physicians and how public schools and their nurses handle contraception. To help her understand the geographic context of teen pregnancy in Rhode Island, I used data collected by KidsCount to map the rates of births to teenage girls aged 15-17 and calculated linear regression trends from 2000 to 2010. As usual, I put these maps together in QGIS.

There are drastic demographic differences in pregnancy rates, with minority communities much higher than others in Rhode Island, but those communities are also decreasing teen pregnancy rates more quickly than any others. An exception from that trend here is West Warwick, at 20 births per 1000 teenage girls. However, there are pockets of impoverished children in West Warwick, so perhaps childhood poverty is a better predictor than race and ethnicity? See minority and childhood poverty maps in my Health Geography Primer for RI post.

teenBirths

teenBirthTrends

Thematic Mapping for Medical Residents

I’ve been facing an interesting challenge: is it possible for residents in medicine, with very limited time, to create reasonably high quality cartographic representations of health indicators?  I think so!  I need two things:

  1. A more or less fool-proof spatial database of health indicators with data at the sub-county level.  I’m working in Rhode Island here, and we only have five counties!
  2. An interactive thematic mapping tool that makes it easy for intelligent medical residents, inexperienced in cartography or GIS, to create quality maps.

For #1, I found the Centers for Disease Control and Prevention Social Vulnerability Atlas, which allows you to download the social vulnerability index and all of its component indicators by census tract for any state.

For #2, I found the online thematic mapping tool, indiemapper, recently made free to the public by its developers at axismaps.

Here are the results of my experiment, followed by some thoughts and specific steps taken to create this map.

CDCSVIRI

Benefits of indiemapper
  • No installation required: just run it in your web browser!
  • Save your work as an .imp file, inclusive of all your data, symbology and the page layout.
  • Default symbology is easy on the eyes, and choropleth maps use Cynthia Brewer’s ColorBrewer. You can choose equal interval, optimal breaks, or quantile classifications and the tool displays a small colorized histogram.
  • Supports multivariate choropleth maps, dot density maps, proportional symbols, and cartograms.
  • Supports several different projections and automatically customizes projection parameters for the extent of your data.
  • Exports maps as a flat jpg or png images or as layered vector svg images which can be edited in Adobe Illustrator or similar open source software like Inkscape.
  • Straightforward cartographic advice and explanation of cartography jargon in Learn More menus.
Limitations of indiemapper

(disclosure: this is based on knowledge accumulated through 4 hours or so of experimentation)

  • All uploads should use the WGS 1984 geographic coordinate system (epsg code: 4326)
  • I could only successfully upload single-part features, and this could be quite problematic for thematic mapping. On one hand, you can use GIS to convert multi-part shapefiles to single-part shapefiles and then upload them. On the other hand, this can seriously distort your data distributions for optimal breaks or quantile classifications by duplicating the data values for every part of a feature.
  • You cannot rename layers or attributes, and this includes legends.
  • You cannot select subsets of layers or exclude any data values from classification.
  • You cannot upload image/raster graphics.
  • Text is always aligned to the left.
  • There is no scale ratio or scale bar. To be fair, the app seems to have been designed primarily for global thematic maps, for which scale is not constant.
  • No support for legend customization, and thus no way to edit “RPL_Themes” on the map above. Each legend includes just one data layer, and legends are only available for layers symbolized with classes or gradients.
  • Unlike Google’s My Maps, I have not reached a limit on the number of attributes (maximum of 50 with My Maps) and files uploaded with the correct attribute data types, (My Maps requires you to duplicate any numerical field!)
Making this Social Vulnerability Map
  1. Create the water and service area layers in QGIS: Prior to making this map, I had downloaded 2010 Census Data for Rhode Island at the census block level from RIGIS.  Using QGIS, I saved the data layer in the WGS 1984 geographic coordinate system and separated water from land. I dissolved the census blocks into towns, and selected Pawtucket and Central Falls, the two towns in the Memorial Hospital service area based on the Dartmouth Health Atlas.
  2. Create the social vulnerability layer in QGIS: I downloaded the Rhode Island social vulnerability atlas data by census tracts from the CDC SVI Atlas, and downloaded a guide to the index and component attributes. In QGIS, I saved the layer in the WGS 1984 geographic coordinate system and then converted from multi-part to single part. I realized that two census tracts had no population and values of -999 for social vulnerability percentiles, which really distorts indiemapper’s classifications. Therefore I selected all the features with RPL_THEMES >= 0. In simpler terms, I selected just the census tracts with legitimate data and saved them to a separate file.
  3. Upload data to indiemapper: I launched indiemapper and uploaded my three data layers (hospital service area, water, and social vulnerability) with their shp and dbf files.
  4. Organize and symbolize data layers: I placed the layers in the following order: hospital service area on top, followed by water and social vulnerability.   Social vulnerability is symbolized as a choropleth data layer with the RPL_THEMES attribute.
  5. Page layout, map projection and extent: The portrait page layout is best for Rhode Island, and I zoomed the map extent to the social vulnerability layer. From the available map projections, the Transverse Mercator projection is best for Rhode Island, as the RI state plane system uses it.  Indiemapper’s suggested parameters for the projection’s central meridian is good, matching the RI state plane projection.
  6. Final touches: I made the graticule layer invisible, and added a north arrow, legend for RPL_THEMES, and annotations for the title, credits, and map notes.
Do it Yourself:

The Social Vulnerability Atlas is composed of four themes based on fifteen attributes, plus two ancillary attributes for health insurance and daytime population. Therefore, I’ve only shown one out of 22 possible maps here!

You can download all the data for this project in the form of shapefiles in a zip file here: www.josephholler.com/files/SVI_indiemapper_files.zip. Be sure to extract/copy the contents out of the zip file once it downloads. The zip file also includes the CDC’s documentation, or else download that here.

If you’re inclined to do even less of the work yourself, download my indeimapper map file for the project here: www.josephholler.com/files/CDCSVI.imp. Launch the indiemapper app, load my CDCSVI.imp map file with the “Browse for IMP” button on the start-up screen, and start customizing your own thematic maps!