GEOSTAT 2015 Visualisation Competition Entry
Robin Lovelace and Chris Brunsdon
August 21, 2015
This is our entry to the visualisation challenge at the GEOSTAT Summer School 2015.
Rather than single static visual outputs, we have built an online and interactive visualisation tool.
All the code is open source, allowing others to reproduce, modify or extend the tool. This has many benefits for geographical research and teaching (Brunsdon 2015).
Live online tool: http://geo8.webarch.net/robin/vizcomp/
Source code: https://github.com/Robinlovelace/GEOSTAT2015-viz-challenge
Images
Let’s take a look at the tool’s capabilities with 4 screenshots.
grid.raster(readPNG("figures/fig1.png"))
Overview of the visualisation tool
grid.raster(readPNG("figures/roll-mean.png"))
Illustration of the impact of changing the rolling mean window, using the new RcppRoll package (Ushey 2015)
grid.raster(readPNG("figures/solidraster.png"))
Raster layer
grid.raster(readPNG("figures/zoomrastersolid.png"))
Zoomed layer. Note the emphasis on ‘solid’ precipitation and the high values in mountainous areas.
Further work
There are many directions for future work opened-up by this research, including:
The addition of ‘hexagonal bins’ (note the placeholder for this in the first drop-down menu).
Addition of spatial sub-setting. This could be combined with some kind of ‘quadrant analysis’, whereby the points are divided into northeast, southeast, southwest and northwest sections, for example.
The integration of the tool with the mapView package for online GIS.
Exploration of changing the cell size and bandwidth of the raster interpolation layer.
References
Brunsdon, C. (2015). Quantitative methods I: Reproducible research and quantitative geography. Progress in Human Geography. doi:10.1177/0309132515599625
Ushey, K. (2015). RcppRoll: Efficient Rolling / Windowed Operations. Retrieved from http://cran.r-project.org/package=RcppRoll