The Approach
Through a series of collaborative sessions, the Pro Bono Analytics (PBA) team worked with the C-10 Network Administrator to understand the meaning and application of the data collected from the sensors and further to understand the full range of data involved, and how the data was organized, stored, and prepared for analysis.
There are ~20 monitoring stations installed in communities within a 10-mile radius of the Seabrook Station nuclear power plant. The sensors capture readings of beta and gamma radiation levels, as well as wind speed and direction, at 1-minute intervals. The data is pushed to a central location in near real time and backed up by a daily log file. The data is retained in a series of daily folders containing the log files from reporting sensors.
The Solution
The core challenge expressed by C-10 was to enable better analysis of their existing data. Performing initial exploratory analysis, the volunteer team plotted the beta and gamma radiation time series and produced visualizations of the wind speed and direction data.
Through discussion with the team members and input from the user community (including C-10 board members and affiliated scientists) the team chose to create an interactive application using the Streamlit framework for Python applications. The application allows interested parties to access the data over the web and plot the radiation time series and related wind plots over the desired time frame. The radiation levels are shown as line charts, and wind speed and direction for the same time period are shown as radial scatterplots. The combination of charts allows interested parties to identify spikes and other anomalies in the radiation levels and see where those correlate with wind speed and direction.
Early in the project, the PBA team explored several options for managing the existing data and understanding the completeness of the historical record. The daily log files generated by the sensor array are retained on C-10’s FTP servers for access by government and community partners if needed, but access to data from multiple servers over longer periods of time is a manual process. The team elected to use the file system to continue to organize and retain the log files and built some log file maintenance support into the application, to allow for simple maintenance of the sensor log data.