Summer of Maps: Raster Versus Vector Visualization

Now in its third year, Azavea’s Summer of Maps Program has become an important resource for non-profits and student GIS analysts alike.  Non-profits receive pro bono spatial analysis work that can enhance their business decision-making processes and programmatic activities, while students benefit from Azavea mentors’ experience and expertise.   This year, three fellows worked on projects for six organizations that spanned a variety of topics and geographic regions.  This blog series documents some of their accomplishments and challenges during their fellowship.  Our 2014 sponsors, GoogleEsri and PennDesign helped make this program possible.  For more information about the program, please fill out the form on the Summer of Maps website.

Raster Versus Vector Visualization

As a Summer of Maps fellow I worked with two non-profit organizations: Girlstart in Austin, Texas which empowers girls with Science, Technology, Engineering and Math, and City Harvest in New York City which rescues food all over the city and distributes it to hunger programs. Both wanted to identify areas that are in the most need of their services. Girlstart also wanted to determine areas for fundraising.

One of the tasks for both of my projects was to create composite layers built from different, but related, variables. For example, I made a layer of relative wealth for Austin’s Girlstart that took into account: median home value, educational attainment, and median household income. Since this data was at the census tract level I was working with vector data but actually converted to raster because I thought a surface of wealth would be both intuitive and pleasing to the eye. A couple examples of well-known raster maps are Yelp heat maps or weather maps. I was striving for a similar look and feel.

What’s vector?

“a representation of the world using points, lines, and polygons. Vector models are useful for storing data that has discrete boundaries, such as country borders, land parcels, and streets” (ESRI GIS Dictionary).

What’s raster?

“a representation of the world as a surface divided into a regular grid of cells. Raster models are useful for storing data that varies continuously, as in an aerial photograph, a satellite image, a surface of chemical concentrations, or an elevation surface” (ESRI GIS Dictionary).

It definitely wasn’t quite as beautiful as I had hoped, nor quite as meaningful. I thought it would provide a nice smooth surface across Central Texas and show more detail by being a stretched gradient.  Instead it just looks like really fuzzy tract boundaries. This is because my data attributes were not continuous. They are polygons and quite large polygons at that. When rasterized, the values in the cells are all the same within each polygon which doesn’t signify much. The process of rasterizing did not add any additional information or aesthetics. The vector format below is the better choice. It looks neat and is appropriately symbolized by a color gradient. The tract boundaries are distinct and the wealth ranking is distinguished across the features.

Recall that I created composite layers for both of my projects. For City Harvest I made a combined layer of vulnerability based on the percent of people living below the poverty threshold and the percent of people receiving SNAP benefits. It was a very similar task and used census data at the census tract level again. When I made a density raster, however, this is what happened.

The raster looks significantly different, and better, than the Girlstart raster. It is successful because this surface conveys information in a different and effective way.  That is, a more continuous surface shows the patterns in a smooth fashion. The data is from census tracts just like with Girlstart, but the actual size of the polygons in New York are much smaller than those in Austin.  That translates to more ‘pieces’ (and more data) to visualize.

Scale and size played a major role in whether to use raster or vector for me, but there are a couple other criteria to consider. While both my datasets were in vector to start, one should recognize how data is originally formatted as a good hint as to what may be appropriate. This has a lot to do with context. Just as the definitions referenced, certain topics lend themselves to one or the other. My starting demographic topics make a lot of sense as vector because census information is gathered from people who live in places that are normally categorized into geographic regions like counties and states. Other subjects like environmental monitoring are often rasters because, much like the real world, the earth is a continuous surface. Of course these are simply general guidelines. It’s all about how you perceive the data and want to visualize it. That last part is key. My first Girlstart raster simply didn’t look right because the unit of analysis (census tracts) was too large to visualize complex variation in the data.

Through my experience I’ve determined four recommendations that are good starting points to consider when contemplating between raster and vector.

• Scale and size of features
• Original formatting
• Context
• Aesthetics