Declining radiation levels in Fukushima?

[ferdo]

Hi,

While working on my Windows App (which will be named: 'Radiation Explorer for Safecast Data', or RadExSaDa) I discovered declining radiation levels around the Fukushima site in Japan.

With a new option in my app I can select all datapoints within a selection rectangle (max 6.000.000 measurements) and show them as a timeseries.

See as an example the following image. Above the chart you can see the timeseries which starts at 5/5/2011 and ends at 1/12/2018. It has 569138 measurements.

When I look at other locations nearby I see the same decline in time.... I don't know if this is a valid conclusion, but at least it seems hopeful!

So this is FYI

[Sean Bonner, Safecast]

Very cool Ferdo!

Yes, it’s legitimate and we show the same thing on our online visualization at map.safecast.org with the time snapshots, the levels are actually decreasing slightly faster than predicted but that’s expected as well.

Thanks!

-s

 

Sean Bonner

Co-Founder, Director of Global Operations

Safecast.org | pgp key

[Mat Schaffer]

Love the time series and map combo here. 

Might have to see if we can get something similar going. 

--

-Mat

matschaffer.com

[ferdo]

I've added another cool option to change the date ranges on the map. When there are less then 2000000 points visible the results of the daterange changes are displayed relatively instantly.

See a video here: 

http://bigroses.net/video/Radiation Explorer for Safecast Data 22-3-2018 23_50_29.mp4

regards,

Ferdinand.

[Christian Michel]

Cool visualization. However it needs to be taken with a grain of salt. If you're using a selected area and put all data points into a time series without looking at their position then the result could look like it's decreasing (or even increasing) when the data points for differently contaminated areas (e.g. streets) were collected over different periods of time. At the end the selection area needs to be fairly small (or best stationary) to make a qualified statement.

So for best results there should be an indicator how many points are actually very close together and then put those into maybe different time series to discover trends.

Regards,
Christian

[ferdo]

Yes, of course, you are absolutely right!

In the app posted to the Store the number of data points in de selection rectangle is displayed. So that could help identify valid regions.

The distribution of the contamination is not equal to a rectangle. Would it help to let the user draw a free from path? 

The inside/outside calculations will then become (much) slower, but when the results are significant better, I could try it...