1. What we do
The purpose of this site is to investigate a scale and possible consequences of an accidental release from a nuclear power-plant under various meteorological conditions and increase public awareness on this important topic. Generally, it is assumed that all serious consequences will take place inside of so called emergency planning zones which are typically circular in shape and 10-30 km in diameter. We want to examine whether an accident under adverse meteorological conditions can cause serious consequences also at larger distances from a power-plant (from tens to hundreds km). We created an automated system which runs a simulation of an accidental release from a nuclear power plant every day and evaluates basic radiological quantities on a computation grid. We use a stat-of-art Lagrangian particle dispersion model Flexpart and force it with NOAA's Global Forecast System (GFS) operational data with horizontal resolution 0.5 deg and 26 vertical levels provided by NOAA. On this page, anybody can see what would have been possible consequences of an accident (if any) in the place where she/he lives if it have had happened.
1.1 Source and computational domain
As a sample source we chose Czech nuclear power-plant Temelín with two VVER reactors. We use a rectangular computational grid with grid size approx. 10 times 10 km at the latitude of Temelín. The grid has dimensions 30 x 30 degrees centered at the NPP site and covers CEE region and also a part of Western Europe, Balkan and Ukraine. Besides grid cell values we also evaluate doses in a set of points of interest (POIs).
1.2 Calculation of radiological quantities
Atmospheric transport model Flexpart uses a large number of computational particles to calculate spreading of various pollutants in the atmosphere. We assume that the release starts at 6:00 UTC every day and calculate its propagation hourly over following 24 hours. We assume the release to be comprised of 16 nuclides (both noble gasses and aerosols), see the tables below for details. Each hour we evaluate following quantities:
- Activity concentration in air (Bq/m3)
- Deposited activity (Bq/m2)
- Gamma dose rate from activity dispersed in the air - cloudshine (mSv/h)
- Gamma dose rate from activity deposited on the ground - groundshine (mSv/h)
- Gamma dose rate from internal irradiation due to inhalation (mSv/h)
Results are presented in the form of transparent overlays on GoogleMaps over whole computational domain. In POIs we also show temporal profiles of doses and dose rates from all pathways. See Today's release Section.
To calculate dose rates we use a tool quickdose which is capable to apply radioactive decay in post processing, so instead of 16 species with with different physical half-lives we propagate just two species (one for noble gasses and one for particulates - we do not distinguish here between aerosol-bound and gaseous iodine since we do not consider it important here). On a six-core machine this takes about one hour with 0.5M of computational particles per one hour of release. Formulas and other details can be found in this BitBucket repository.
Dose conversion coefficients are taken from Radiation Protection Bureau, Health Canada, Atomic Energy Control Board, Atomic Energy of Canada Limited: Recommendations on Dose Coefficients for Assessing Doses from Accidental Radionuclide Releases to the Environment (1999).
1.3 Source term and nuclide properties
The source term we use was compiled by the Czech National Radiation Protection Institute and it serves as a reference source term for evaluation of operational intervention levels. From the comparison in table below (tab Comparison with major nuclear accidents) is obvious that it is by orders several of magnitude lover than accidents in Chernobyl or Fukushima.
|No.||Nuclide||Relese in 1st hour||Relese in 2nd hour||Relese in 3rd hour||Relese in 4th hour||Relese in 5th hour||Relese in 6th hour|
Physical properties of nuclides
|No.||Nuclide||Physical half-life||Physical type||Cloudshine coef.||Groundshine coef.||Inhalation coef.*|
|s||Sv/s * Bq/m3||Sv/s * Bq/m2||Sv/Bq|
Comparison with major nuclear accidents
Comparison of total emissions of selected nuclides with major nuclear accidents Chernobyl (1986) and Fukushima (2011). Our source term is always at least by one order of magnitude lower. All emissions are in petabecquerels (PBq)
|No.||Nuclide||Our source term||Chernobyl1||Fukushima2|
1 Data taken from Devell, L., S. Guntay, and D. A. Powers. The Chernobyl reactor accident source term: development of a consensus view. Organisation for Economic Co-Operation and Development-Nuclear Energy Agency, Committee on the safety of nuclear installations-OECD/NEA/CSNI, Le Seine Saint-Germain, 12 boulevard des Iles, F-92130 Issy-les-Moulineaux (France), 1995. [Overview] [Download]
2 Data taken from O. Saunier, A. Mathieu, D. Didier, M. Tombette, D. Quélo, V. Winiarek & M. Bocquet, An inverse modeling method to assess the source term of the Fukushima Nuclear Power Plant accident using gamma dose rate observations, Atmospheric Chemistry and Physics 13, 11403-11421 (2013) [Overview] [Download]
2. Today's release consequences (Release start: 2017/03/23 06:00)
2.1 Spatio-temporal distribution of quantities
Use controls to select a combination of a radiological quantity and a nuclide. Red circle denotes the zone of emergency planning of diameter 13km. You can click points of interest (POIs) to see dose graphs.
2.2 Doses in points of interest after 24 hours
What follows is a table with all POIs with non-zero maximum dose. You can use icons in column Map action to centre and zoom map on a particular POI. To see dose temporal profile in firt 24 hours after release start please click on a map markers.
2.3 Guide to dose values - what are these numbers comparable to?
On this website, units mSv (1000 mSv = 1 Sv) and mSv/h are used for radiation dose and dose rate, respectively. Risk caused by irradiation depends both on overall dose obtained (the lower the better) as well as on exposure time, i.e. how long did it take to get the dose (the longer the better). Examples of radiation effects for various dose levels and exposure times are in the following table:
|Dose (mSv)||Scientific notation on color|
scale under the map (mSv)
|0.1||10-1||instant dose (seconds to minutes)||Comparable to an average chest X-ray|
|1||100||1 hour||Hourly dose obtained at Fukushima site 1 day after the accident|
|10||101||instant dose (seconds to minutes)||Comparable to an average CT scan|
|100||102||1 year||Lowest annual dose where increased lifetime risk of cancer is evident|
|1000||103||hours||Temporary radiation sickness and nausea (still not fatal)|
|4000||4x103||hours||Bleeding, hair loss, death possible within 4-6 week if untreated|
3. Frequently asked questions
- Why do you do this?
- Simply, because I can. Or for the sake of science because scientia potentia est. It's a kinda a hobby...
- Why don't you calculate some longer release?
- Well, that would be nice but my computational and storage resources are rather limited. With current six-hour release it just works and does not take too long. Fun fact: a single day of operating this system takes approximately 1.3 GB of disk space:)
- How accurate is your model?
- Flexpart is one of the most advanced atmospheric transport models in the world. It is used by many scientific teams and institutions all around the world, including CTBTO . Of course, there is still a space for improvement. That is why its parformance and parametrizations of physical processes are continuously being improved. There has been carried out many evaluations of the model against measurements, e.g. here. The main factor which limits model accuracy is a limited amount of computational resources. What regards meteorological forecasts, the model is forced by GFS operational data with resolution 0.5 deg and 26 vertical levels. An alternative would be to use ECMWF data with higher resolution of WRF data. Unfortunatelly, ECMWF data are not provided for free and for computation of the later - WRF - we do not have computatinal resources. GFS data is provided for free and it performs well. What could be weakest point of our calculations is the gamma dose rate module wich is based on quickdose system which needs to be further validated. This validation is currently in progress.
- Do you use any meteorological measurements in your calculations?
- No. The model is forced by operational meteorological fields, i.e. predictions of the future state of the atmosphere - there are no "future measurements". However, these forecasts come out of so called analyses based on a waste number of previous observations of the atmosphere (satellite, radiosonde and others), see this for more details.
- Is it possible to see also older simulations or download the results?
- All the results are stored and it takes really a lot of space:). Generally, making all the simulations available is one of future goals. Unfortunately, the interface for this task is not ready yet. But I'm working on it:)
- Can you use/extend/modify the software for calculation of hypothetical accident consequences at a different power plant(s)?
- Yes, the software is quite versatile (and starts to be quite complex:). Contact me!
- Can you add some new points of interest for dose calculations, e.g. a town where I live?
- Yes, I can. Contact me!
- How the software works?
- Well, the automated calculation has following steps which are executed daily as a cron job on a linux server:
- Fetch of current meteorological forecasts from NOAA FTP server.
- Creation of a new directory tree with all Flexpart initialized and configure properly (configuration files are mostly generated from templates).
- Batch run of all Flexpart simulations in parallel.
- Extraction of deposition and concentration fields from runs, application of radioactive decay.
- Calculation of dose rates and doses.
- Generation of GoogleMaps overlays and dose plots.
- Compiling whole web together (map, data, tables etc.). Again, with an extensive use of a templating engine.
- Upload of the web on a production server.
- You f#!*&^g environmental faggot, your calculations are bull!#@$!
- Firstly, can you do better? Secondly, this is not a question:)
4. Contact and Disclaimer
My name is Radek Hofman. I run this site. I am a Python programmer with a PhD in mathematics but I want to be a data scientist:) Should you have any questions or comments please feel free to mail me at firstname.lastname@example.org.
All the releases on this webpage are just simulations (what-if scenarios) and they are not happening in reality. Hopefully, the dose calculation engine works correctly and does what it should. I can, however, not completely rule out any bug in my software. In other words, I do not guarantee anything, and using this web site is entirely at your own risk.