The above graph is derived from data scraped from TEPCO press releases. Every hour or so for the first few days of the crisis, a TEPCO van would record radiation (probably Beta/Gamma, but the translation is unclear) at the front gate to the Fukushima complex. The series is less continuous than the graph would lead you to believe, but I don’t feel the actual series is so different as to warrant concern. I pulled the data down from here (some explanation of the original docs on a reddit thread) and converted the dates from a ragged MM/DD/YYYY HR:MN AM format to something a little more tractable using Google Refine. I used Google Refine because making such a change programmatically was much more challenging, with the irregular format trouncing my understanding of
strptime(). As you will see from the code below I actually only used the dates for labeling. But the y axis is completely true to the original data.