Sylvia GENEROSO, Pascal ACHIM, Mireille MORIN, Philippe GROSS, Gilbert LE PETIT. Commissariat à l'Energie Atomique et aux énergies alternatives (CEA, ...
SCIENCE AND TECHNOLOGY CONFERENCE 2015
Identifying the source of specific events: a major challenge for French NDC Sylvia GENEROSO, Pascal ACHIM, Mireille MORIN, Philippe GROSS, Gilbert LE PETIT (T1.3-P6) Commissariat à l’Energie Atomique et aux énergies alternatives (CEA, DAM, DIF, F-91297 Arpajon, France) Abstract The French National Data Center (NDC) makes use of atmospheric transport modelling to locate areas that may contain the source at the origin of a set of measurements at IMS stations. The wide variety of possible solutions – ranging from a single release to a combination of several sources, local or remote, from industrial or military applications – makes it a challenge.
1. Possible Source Region method
A fictive scenario is considered: 1013 Bq of Xe-133 is released at (2,59W; 51,45N) Feb. 7 2015 between 08:00 and 10:00 (UTC). Its dispersion is simulated with GFS/FLEXPART (0.5° resolution). Activity concentrations at the location of IMS stations are sampled. SEX63Stockholm, RUX61-Dubna and MN01 have detected the plume. The resulting activity concentrations at the location of IMS stations are used as the set of measurements in this study (Table 1). The aim is to evaluate the ability of our method to locate a source.
Table 1 Set of measurements (fictive) used in the study: seven detections (highlighted in yellow) and three non-detections (in grey) are considered. MN01 is a fictive noble-gas station located at the location of RUP54-Kirov.
Spatial localization: Retro-dispersion is calculated for each detection/non-detection in Table 1. Overlap of retro-plumes identifies areas most likely containing possible sources. Retro-plumes from nondetections exclude areas, where releases could not occur. Areas in red in Fig. 1 identify possible source locations explaining the seven detections. Dispersion is calculated for 8 days using GFS/FLEXPART (0.5° resolution). Fig. 1 Number of overlap of retro-plumes calculated from detections, constrained with a source term of 5x1013 Bq, occurring within the simulation period of 8 days, considering exclusion areas from non-detections. Points 1 to 4 are selected as reference points. The blue open circle locates the actual release. Fig. 2 Time series of required source terms for Xe-133 at Point 2 (See Fig.1) for each measurement in Table 1.
Fig. 3 Xe-133 activity concentrations simulated considering a release of 1013 Bq at Point 2 on Feb.7, 13:00-15:00 and fictive measurements.
Time of release and source term: Fig. 2 shows time series of source terms at Point 2 for example, required to explain the measurements. Assuming a single release, the most likely solution is given by areas, where curves from detections intersect, below curves from nondetections. One area verifies these conditions (red-dashed circle) for Point 2: Feb. 7, 8:00-16:00 / source term of 1013-5x1013. A forward simulation with this release agrees with the fictive measurements (Fig. 3). Similar agreement is obtained for Points 1,3 and 4.
From a scenario describing a release of Xe-133, this study presents: 1. the capacity of our localization method to characterize a source 2. additional developments to narrow solution areas 3. the influence of collection periods of measurement systems on localization results Application to a real case, FRX27, is presented.
3. Sensitivity to collection periods
2. Additional developments zoom
12-hour averages
Fig. 4 Diagrams represent times series of source terms calculated at a given point, as in Fig. 2, considering two detections (d1 and d2) and one non-detection (nd1). The red box represents a moving window of 12 hours.
The method in Section 1 counts the number of overlap of retro-plumes from detections, not excluded by retro-plumes from non-detections, occurring during the simulation period. The additional approach described here verifies the same criteria but within a narrow moving window of 12 hours (red box), moving with a time step of two hours, constrained with a source term (TSmax). Treatment within the boxes applies to 12-hour averages and allows to refine the exclusion criteria by considering margin distances for exclusions (Fac1 and Fac2). Fig. 5 presents results integrated on the simulation period (8 days). Fig. 6 shows results per time step (2 hours). Spatial localization of single releases: Fig. 5 represents the maximum number of overlap of retro-plumes from detections that passed a given location (possible source) within a time window of 12 hours. Areas in red identify source locations explaining seven detections with a single release. Compared to Fig.1, this result narrows the solution area (in red) around the point of actual release. Fig. 6 presents the results per time step of 2 hours for a limited period. Fig. 5 Maximum number of overlap of retro-plumes calculated from detections, constrained with a source term of 5x1013 Bq, occurring within a 12-hour time- This result shows that the time and window, considering exclusion areas from non-detections. location of the actual release is captured. Fig. 6 Same as Fig.5 except that results are shown per moving window of 12 hours, moving with a time step of 2 hours, from Feb. 7, 14:00 (top left) to Feb. 7, 00:00 (bottom right). Dates are centered to the middle of the time window of 12 hours (equivalent to the t0 in Fig.4). The actual release took place at the location of the blue open circle. The time of the actual release (08:00-10:00) is included in the time window of all panels except the first (14:00) and last (00:00) ones of the series.
Fig. 7 (Top left) Measurements at SEX63 considering three collection periods (6h, 12h and 24h) given the fictive scenario considered in this study. Maps of the resulting localization (method in Section 2) with 6h- (top right), 12h- (bottom left) and 24h- (bottom right) collection period.
Influence of the collection periods of measurement systems on localization results: Fig.7 shows that the most likely solution (in red) is slightly affected by the collection period. In this particular case given the specific meteorological conditions, a 6-hour collection period narrows possible solution areas compared to a 24-hour collection period, and has limited effect compared to a 12-hour collection period.
Application to detections at FRX27-Tahiti
Three successive detections of Xe-133 were measured from June 5 to 7, 2013 at FRX27 station, while the station never detected radioactive xenon before (top left). The method described in Section 2 defines a solution area, which includes ANSTO, a major medical isotope production facility (bottom right). The time of the release is estimated May 31, 2013, 10:00-14:00 with a source term of 5x1012-1013 Bq of Xe-133 (top right). It is compatible with releases at ANSTO.
▲Required source terms at ANSTO to explain detections at FRX27 ◄Detections at FRX27 ▼Max. number of overlap of retro-plumes constrained with 1013 Bq Number of overlapping concomitant retro-plumes :
ANSTO
Application of Possible Source Region method to a known (fictive) release of Xe-133 was able to identify an area of possible solutions, which includes the location and time of the actual release. Additional developments aiming at identifying the number of retro-plumes, which passed possible sources within a time frame of 12 hours, helped narrow the solution area. In this specific case, collection periods of measurement systems influence slightly localization results: 6-hour and 12-hour collection periods have narrowed possible solutions compared to a 24-hour collection period. Applied to Xe-133 detections at FRX27, the method has identified a major medical isotope producer as a possible source.
