and the depths of the human misery left in the wake of the massive Tsunami have left many people awestruck. The size of the quake itself was enormous and many people have had a hard time comprehending just how big this earthquake was. Some sites like Japan Quake Map help us to comprehend the magnitude of this event. As a result of the earthquake and tsunami the nuclear reactor at Dai-ichi was severely damaged and has been leaking radiation. The radiation readings have been made available by WIDE and Japan's Nuclear Safety Division.
The idea was to use R to create an informative map of Japan showing the radiation levels of the different prefectures. Python was used to import the data from both of the web sites and insert it into a MySQL database. The format of both of the pages was understandably quite dynamic and resulted in the python script needing to be tweaked quite often. Sometimes it was easier to just copy and paste the data in a spreadsheet and then export as a csv to import into the database.
For the map, the shapefiles included in the R distribution were not working out so shapefiles for Japan from Harvard Asia Studies were used. These shapefiles combined with the plotPolys() command produced a higher quality map than the standard shapefiles.
The readings for most prefectures were rather reliable however in Fukushima and in Miyagi the readings were sporadic. Miyagi was hardest hit by the tsunami and most of the area was destroyed. It appears that most of the readings were from mobile units and there are gaps in the coverage. If there were no readings available for a given day they were estimated using the surrounding readings both spatially and temporally. In Fukushima which is the location of the reactor, there were many monitoring sites set up but they seemed to come and go over the course of time. For the purposes of this map, the sites located between 20km to 30km from the reactor were averaged together to give a reading for the Fukushima prefecture.
R code for Map
The radiation level values were each assigned a hex color value and then merged into a vector that matched the prefecture names in the shapefile. Nested for loops are usually a bad idea, especially in R, suggestions for a more elegant solution are welcome. plotPolys() takes care of making the map and the only thing left was to reverse the logarithm function to get the real values back and add a legend.
All of the maps from 16 March to 4 April were combined into an avi using:
mencoder mf://*.png -mf fps=1:type=png -ovc lavc -lavcopts vcodec=mpeg4 -oac copy -o output.avi
Shapefiles from China Historical GIS Project, "Tokugawa Japan GIS, Demo Version." Feb 2004
-- Greg Szalkowski