OpenSpectra. Open Source GUI application for interactive review of Particulates and GeHP Noble Gas Spectra. A Users Guide. Version 1.

Transcription

1 OpenSpectra Open Source GUI application for interactive review of Particulates and GeHP Noble Gas Spectra A Users Guide Version May 2013 CTBTO/ IDC Division/ Software Applications Section/ Scientific Methods Unit

2 Document Version History: Version Date Author Comments / main reason(s) for baseline changes Abdelhakim Gheddou, Radionuclide Officer, Scientific Methods Unit, Software Applications Section, International Data Centre Division CTBTO Preparatory Commission Tel.: (+43 1) Abdelhakim.Gheddou@ctbto.org First version for training and testing purposes Reviewed by Jeffrey Given, Chief, Software Applications Section, International Data Centre Division CTBTO Preparatory Commission Tel.: (+43 1) Jeffrey.Given@ctbto.org Page 2 of 70

3 Table of Contents 1. Introduction Classification of spectral data Sample type Data type Spectrum qualifier Spectrum status Login to the database Options for spectrum selection Using the assignment queue Browsing from the database Loading a specific spectrum Main spectrum display area Graph Assistance panel Detailed Information panel QC Flags panel Spectrum category Drop-down menus Comments menu: Options menu: Calibrations menu Policies menu Reports menu Interactive review functionalities Interactive Reprocessing Peak search dialog The Nuclide Review facility The Technetium Tool (TcT) Comparing Spectra Sample release Xenon spectrum analysis Appendix 1: Operational policies for CTBTO particulates data analysis Page 3 of 70

4 List of Figures Figure 1: Database connection window... 9 Figure 3: Database user change Figure 4: Sample assignment window Figure 5: Sample selection dialogue window Figure 6: Load sample ID window Figure 7: Graphical Assistance panel Figure 8: Detailed Information panel Figure 9: QC Flags panel Figure 10: Sample category window Figure 11: Station comment window Figure 12: View general comment window Figure 13: Add general comment window Figure 14: Color preferences window Figure 15: Pick color dialog Figure 16: Energy calibration plot Figure 17: Resolution calibration plot Figure 18: Efficiency calibration plot Figure 19: Review policies window Figure 20: Activity summary screen Figure 21: Sample category screen Figure 22: MDC report screen Figure 23: Nuclide library window Figure 24: Spectrum processing parameters Figure 25: QC flags report Figure 26: Spectrum reprocessing window Figure 27: Peak search dialog window Figure 28: Peak comment insertion Figure 29: Peak comment validation Figure 30: Remove nuclide functionality Figure 31: Remove nuclide options Figure 32: Remove nuclide validation Figure 33: Nuclide review window with Ag-110m selected Figure 34: Nuclide review window with Pb-212F selected Figure 35: Nuclide review window with Zr-97 selected Figure 36: TcT window with unsupported Tc-99m Figure 37: TcT window with Ge-75m Figure 38: main display in comparison mode Figure 39: sample release window Figure 40: Release without category window Figure 41: Change status to Viewed Figure 42: Change status to Others Figure 43: QC panel for SPALAX samples Figure 44: Xenon window in default mode Figure 45: Xenon window in single mode Page 4 of 70

5 1. INTRODUCTION Within ongoing technical support activities to States parties, the Provisional Technical Secretariat (PTS) of the CTBTO Preparatory Commission has recently developed software applications for Radionuclide data processing. The Radionuclide components are integrated into the NDC-in-a-BOX package. The aim is to allow National Data Centres to replicate analysis results of the International Data Centre (IDC) and to perform independent processing based on raw data from the International Monitoring System (IMS) as well. Both particulates stations and Noble Gas systems are covered by the NDC-in-a-BOX Radionuclide processing pipeline. OpenSpectra is an interactive software GUI tool for radionuclide data analysis (both particulates stations and SPALAX based Noble gas systems) of the IMS monitoring network. The tool is intended for interactive review of automatic results as generated by AutoSaint software. OpenSpectra is a java-based license-free application. It runs on Linux Operating System under the standard configuration of the NDC-in-a-BOX environment. This includes a structured file system and a database schema which are customized from the current configuration of the IDC platform for the Radionuclide monitoring technology. The main functionalities of the tool include interactive review of automatic peak search and nuclide identification results. OpenSpectra offers relevant features and dedicated functionalities which allow Analysts to introduce necessary corrections to automatic processing results as appropriate. In addition to sample spectra analysis, the GUI is also used to interactively check all auxiliary spectral data (detector background, Quality Control, calibration, blank and spike spectra). OpenSpectra software allows assignment of spectra to individual users with required roles and permissions as configured in the database. Interactive changes are only allowed when the spectrum is assigned to the user. The current User s Guide provides a functional description of OpenSpectra software application. This first version is mainly intended for testing and training purposes. Before focusing on OpenSpectra GUI, classification parameters of spectral data are introduced in the first section of the document. General features of the GUI, ranging from database login to sample release, are then described from the user perspective. The last section is dedicated to more detailed presentation of OpenSpectra main functionalities as involved in the interactive review process. Both particulates data and SPALAX based Noble Gas systems are covered. The current Users Guide is also accessible in pdf format via the menu item Users Guide of the menu Help under OpenSpectra GUI. Page 5 of 70

6 2. CLASSIFICATION OF SPECTRAL DATA Spectral data of the CTBTO Radionuclide monitoring technology (particulates stations and Noble Gas systems) are classified unambiguously in the file system and the database tables. This allows differentiated processing flowcharts and proper access mechanisms to raw data and analysis results. The classification scheme is based on four main filtering parameters: sample type, data type, spectral qualifier, and status SAMPLE TYPE Three high level identifiers are used for sample_type to distinguish between different monitoring systems used in the IMS radionuclide network: - Spectra from Particulates stations: P - Ge-HP based Noble gas (SPALAX) spectra: G - Beta-gamma coincidence based Noble Gas systems (SAUNA and ARIX) spectra: B The sample type is recorded as SAMPLE_TYPE in the GARDS_SAMPLE_DATA database table DATA TYPE Spectra are also classified according to data type. This reflects the spectrum measured entity. The followings settings are used: - Sample spectrum : S - Detector background: D - Blank filter : B - Efficiency calibration: C - Quality control: Q The data type is recorded as DATA_TYPE in the GARDS_SAMPLE_DATA database table SPECTRUM QUALIFIER A spectrum is "qualified" according to acquisition time in the measurement cycle. Two qualifiers are used for spectrum qualifier (commonly called spectral qualifier): - Final, or FULL spectra have nominally 24- (or >22-) hour acquisition time. - Preliminary or PREL spectra have acquisition times of 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, or 22 hours that is, the normal two-hourly spectra sent from IMS stations, excluding the final FULL 24-hour spectrum. The spectral qualifier is recorded as SPECTRAL_QUALIFIER in the GARDS_SAMPLE_DATA database table. Page 6 of 70

7 1.4. SPECTRUM STATUS In addition to sample type, data type and spectral_qualifier which are inherent to the data message itself, each spectrum is classified according to its status along the processing workflow. The status identifier is recorded as STATUS in the GARDS_SAMPLE_STATUS database table. There are two categories of status in terms of triggering actions: Status value as applied by the processing pipeline in automatic mode. Status value which can be set interactively by Analysts, either via the review tool or by running SQL statements directly in the database. Status conditions applied during automatic processing: Processed P The spectrum has been processed successfully through the automatic pipeline and, for sample FULL spectra, Automated Radionuclide Report is generated. The spectrum is ready for interactive review. Analysis A The spectrum was meant to be processed but it has stopped in the Analysis (post parsing) phase of the automatic pipeline. Remedial actions may be possible to recover from failure. Unprocessed U The spectrum was not intended to be processed automatically. This is the case of most PREL spectra which are not all processed in order to optimize the usage of disk space. under Quality control Q After a spectrum has been released by the Analyst, the status is changed from P to Q and the spectrum it enters a queue for quality-control assessment. During this period, the spectrum can be manually released by a user with appropriate permissions of quality-control officer as defined in the database. Otherwise, once the pre-set waiting time is elapsed, the spectrum status changes automatically from Q to R. Reviewed R The interactive review process has been completed, and the spectrum has been released (with or without categorisation) and reviewed product(s) are generated. This status applies (only) to FULL sample spectra. Page 7 of 70

8 Status conditions applied manually: Failed F The failure during the automatic pipeline analysis is irresolvable: the spectrum is effectively faulty and the failure is permanent. No remedial action can be undertaken to overcome the failure. Duplicate D Occasionally, two Sample Pulse Height Data (SPHD) messages are sent from IMS stations to describe the same sample (e.g., as a corrective action after a format problem with the first message or just because of a software issue at the station). The Analyst decides which one of the two spectra is more appropriate for the interactive analysis. The other spectrum is then labeled as Duplicated. With this labeling only one RRR per measurement cycle will be generated for the station. This interactive action will change the status to D in the database so that the spectrum in question will be automatically filtered out by future runs of queries that are used for various purposes as well as by those involved in different applications. Bad B A spectrum is marked as Bad if faulty data is sent inadvertently from a station (during a maintenance visit, for instance) and for which products should not be released (lack of reliability in terms of CTBT verification value), although, ARR would have been generated if the automatic processing was successful. This also includes situations where a full-sample spectrum is faulty and the Analyst decides to use a PREL spectrum preceding a FULL sample as an alternative option. In such a situation, the original FULL spectrum is labeled as Bad so that any confusion of apparently having two sample spectra on the same day will be avoided. A third category of samples that will be marked as Bad covers cases where SOH data analysis clearly shows a malfunctioning of key modules of the system. Finally, samples that are accompanied by unphysical values for at least one sample metric (such as negative processing time or air volume) are also to be marked as Bad. Viewed V The spectrum has been marked Viewed that is, it has been processed successfully and examined by an Analyst but is not intended to be released, and no products are to be generated. This status applies to QC, calibration, blank, background, and preliminary spectra which have been examined and which need to be removed from the Analysts assignment queue. Page 8 of 70

9 3. LOGIN TO THE DATABASE Upon launching OpenSpectra application (either by double clicking on the related shortcut on the Desktop of the NDC-in-a-BOX Virtual Machine or by invoking rms_openspectra in the command prompt), the user is prompted for the login (DB account username & password) as shown in Figure 1. Alternatively, the database connection window is also brought up via the menu item DB connection of the File menu. File Sample Comments Options Calibrations Processing Policies Reports Peaks release Special nuclide review Help DB Connection DB Disconnect Change User Change Role Save Spectrum Analysis to file Save Spectrum Analysis to file Exit Figure 1: Database connection window The default settings include MySql, Port number, Host name, DB name for a standard configuration. Page 9 of 70

10 Login with default settings: 1. Enter your username and password in the Edit fields next to User Name and "Password". 2. Then click on the button "Connect" to proceed with the login. Note: the NDC-in-a-BOX virtual machine is delivered with a default configuration containing a generic database account ndcuser / ndcuser for User name and Password. The following information message is then displayed for confirming successful DB connection. Figure 2: Database connection confirmation For a user defined configuration, the database connection settings can be changed either through the window or in a configuration file, which can then be loaded by clicking on the button Read new Conf. File. The user can logout from the current session by selecting the menu item DB Disconnect under the File menu. If a new session (different user and/or DB settings) is to be started, you first need to logout from the current session, by selecting the menu item Change User under the File menu. A confirmation dialog message as shown in Figure 3 below is then prompted with Yes and No options. Figure 3: Database user change Upon confirmation, the user will be prompted to login with a different account. Page 10 of 70

11 4. OPTIONS FOR SPECTRUM SELECTION Three possibilities are available for selecting a spectrum to load: (a) from the sample assignment queue, (b) using Open sample selection window to browse the database (c) by entering a valid sample identification (SID) number 4.1. USING THE ASSIGNMENT QUEUE Automatic assignment of spectra to individual analysts (users) can be performed in the automatic processing pipeline. Furthermore, spectra can also be (re)assigned either externally or via OpenSpectra related link. The user can display spectra that are previously assigned to him or her by selecting the submenu item Sample Assignment Queue under the menu item Sample. This results in a display of samples that are assigned to the current user (Analyst) as in Figure 4, below. Figure 4: Sample assignment window Page 11 of 70

12 The display window provides information related to the following sub-set of spectrum identification parameters: - Station code (Station) - Detector code (Detector) - Sample collection stop timestamp (Col. Stop) - Spectrum acquisition start timestamp (Acq. Start) - Spectrum acquisition time, in hours (Acq. (Hr)) - Spectrum analysis status (Status) - Spectrum qualifier (Qualifier) - Data type (D. Type) - Sample type (S. Type) - Sample identification number (Sample_ID) Assigned spectra can be sorted out in different possible ways by cliquing on header items of the grid. Note that ascending and descending modes are also available. To load a spectrum from the queue: 1) Select the corresponding row in the grid, 2) Then, press the button Load Sample BROWSING FROM THE DATABASE Activating the Open Sample sub-menu item of the Sample menu brings up the selection facility window as shown in Figure 5, below. The following filtering parameters can be used according to individual cases of selection criteria: - Station code (Station): individual station or All available stations. - Detector code: individual detector or All available detectors. - Sample collection stop or Spectrum acquisition start. - Spectrum analysis status (Status): individual status or All. - Spectrum qualifier (Qualifier): PREliminary spectra, FULL spectra or both. - Data type (D. Type): individual data type or All available data_types. - Sample type (S. Type): Particulates, Noble Gas or both. When the selection parameters are set as appropriate, the search is activated by pressing the button Query. The list of spectra matching the pre-selected filters will then be displayed with relevant information. Page 12 of 70

13 Figure 5: Sample selection dialogue window Notes on Figure 5: 1. Station selection: Click on the station of interest, using the scroll bar through the list. For including all stations s in the search, keep the selection on the - (dash) item at the top of the combobox. 2. Detector selection: Click on the detector of interest, using the scroll bar through the list. For including all detectors in the search, keep the selection on the - (dash) item at the top of the combobox. 3. Ordering options: Pressing any one of the headings will order available spectra according to the corresponding parameter. 4. Time interval: The time period of interest are be defined by the User via From and To text fields. These filters are assigned either to sample collection stop date or to spectrum acquisition start date, depending on which of the radiobuttons Collect Stop or Acquisition Start is activated on the window. - The default setting is a one week window back from the current date, based on collection stop. From date is set to sysdate 7 days. - If To date is left empty (as by default), it will be set to the current date (sysdate). - In order to include data_type(s) Q, C, D, B in the query, the user needs to check Acquisition start radiobutton. The reason is that Collection start is NULL for these data_type(s) no sampling is involved. 5. Spectrum status selection: select the data type of interest (for example P, R, B, ). Status values are explained with more details in Section 2. For including all status values in the search, keep the selection on the - (dash) item at the top of the combobox. Page 13 of 70

14 6. Spectrum qualifier selection: select whether the spectra should be FULL, Preliminary (PREL). For including both FULL and PREL spectra in the search, keep the selection on the - (dash) item at the top of the combobox. 7. Data type selection: select from the pop-up menu (S: Sample, B: Blank filter, C: Calibration, D: Detector background, Q: Quality Control, K: spiked sample). For including all data types in the search, keep the selection on the - (dash) item at the top of the combobox. 8. Sample type selection: select P for particulates or G for Noble Gas. For including both sample types in the search, keep the selection on the - (dash) item at the top of the combobox. 9. The button Load Sample : for loading the currently selected spectrum from the queue. 10. The button Assign Selected Sample : for assigning a selected spectrum to the current user. 11. The button Assign All Samples: for assigning the whole list of spectra in the window to the current user. 12. The Reset button is used to clear the currently set pre-selection filters. 13. The Close button closes the window. 14. The button Assign Station : (currently disabled) 4.3. LOADING A SPECIFIC SPECTRUM If the spectrum identification number of the spectrum of interest is known, the best option is to use the sub-menu item Open Sample_ID of the menu Sample. A dialog message window is then prompted (Figure 6, below) with a text field where the sample_id of interest can be entered. The spectrum can then be loaded just by clicking on the OK button. Figure 6: Load sample ID window Page 14 of 70

15 5. MAIN SPECTRUM DISPLAY AREA The main window of OpenSpectra (Figures 7, 8 and 9 below) consists of the following items: Drop-down menus One of threee information panels (Graphical assistance, Detailed information, QC flags), selected from tabs in the panel immediately below the drop-down menus. Display area for Gamma spectra. In OpenSpectra default display, the Single Channel Analyzer Curve,SCAC, baseline and LC are superposed in the same window along with the spectrum original counts. The main chart onsists of four series with the following default color-coding convention: - YELLOW: Original spectrum counts, where o Peaks ROIs identified in automatic mode are highlighted in BLUE. o Uncommented peaks (no nuclide is associated and no comment is provided) are shown in WHITE. - ORANGE: Spectrum baseline. - RED: Critical Limit Curve (LCC). - MAGENTA: Single Channel Analyzer Curve (SCAC). Default scaling for horizontal and vertical axis is Automatic: o Horizontal axis in channel number mode, from first channel to Multi Channel Analyzer (MCA) size. o Vertical axis in logarithmic scale from 0 to max count over the spectrum range. o All chart series are in visible mode (counts, SCAC, baseline, LC) The spectrum display area provides basic step-wise zoom in/out functionalities: o Pressing the mouse left button zooms to the area of a rectangle you draw (from left to right). o By clicking on the Reset button, the scaling switches back to automatic mode. The following sections describe specific information panels of the main window. Page 15 of 70

16 5.1. GRAPH ASSISTANCE PANEL This panel as shown in Figure 7 allows the user to customize the spectral display area and perform interactive review. Figure 7: Graphical Assistance panel Notes on Figure 7: - The button Linear/Log 10 allows switching the vertical axis between linear and logarithmic scales. The button caption changes accordingly depending on the scale mode. - The button Channel/Energy changes the bottom axis unit from Channel number to energy in kev, and vice-versa. The button caption switches accordingly depending on the axis unit. - The button Cursor enables and disables horizontal and vertical cursors. By moving the cursor along the spectrum, a tooltip under the bottom axis displays the current channel number, energy and counts for the cursor position. - The button LC shows and hides the chart series related to the Critical Limit Curve (default Red line). The button background color changes accordingly. - The button Baseline shows and hides the chart series related to the spectrum Baseline (default Orange line). The button background color changes accordingly. - The button SCAC shows and hides the chart series related to the Single Channel Analyzer Curve (default Magenta line). The button background color changes accordingly. - The button Scatter/Line changes the chart type of all series (spectrum counts, SCAC, Baseline, LC) between two plot modes: Scatter with markers and Scatter with lines. The button caption changes accordingly. Page 16 of 70

17 - The button Go under the label Energy allows a zoom in around the corresponding spectrum channel. This assumes a valid energy value is entered in the Edit field on the right. - The button Reset sets the display area back to the default scaling for the whole energy range. Compare options sub-panel is described in the Section dedicated to Interactive Review Functionalities DETAILED INFORMATION PANEL This panel displays detailed information on spectrum identification parameters. These consist of the following items as shown in Figure 8, below: Notes on Figure 8: Figure 8: Detailed Information panel - Sample_id: spectrum identification number as assigned by the automatic processing pipeline, - Station code, - Detector code, - Flow rate: air sampling average flow rate, in m3/h, - Sample Status: processing status of the spectrum (U: Unprocessed, P: Processed, R: Reviewed, V: Viewed, A: Analyzed by the parsing program but not processed yet or failed the processing, B: marked as Bad, D: marked as Duplicate, H: Hold on for review). Details on status convention settings are provided in the Section Classification of spectral data, above. Page 17 of 70

18 - Spectral qualifier: FULL for the ultimate acquisition time (usually ~ 24h) and PREL for preliminary spectra that are usually generated after each 2h of acquisition time. - Data type: type of spectral data (S: Sample, B: Blank filter, C: Calibration, D: Detector background, Q: Quality Control, K: spiked sample). - Quantity: air volume of the sample, in m3. - Collection stop: date and time of air sampling stop - Sampling time: duration of air sampling, h - Acquisition start: date and time of gamma spectrum acquisition start - Acquisition time: duration of gamma spectrum acquisition (real acquisition time), h - Category: sample category (from 1 to 5) based on interactive review of the spectrum. - Auto. Cat.: sample category (from 1 to 5) based on automatic processing. - Acq. Life: live acquisition time (real acquisition time dead time), h - SRID: Sample reference identification number. - Sample type: P for particulates data and G for Noble Gas. - Rep. Time: Reporting time (from sample collection start until data is processed). - STA. Rep. Time: Station reporting time (from sample collection start until data is transmitted to IDC) QC FLAGS PANEL This panel, as in Figure 9 below, shows Quality Control (QC) flags for FULL sample spectra. These are compared to reference thresholds of operational requirements and displayed in color-coded mode for test item pass/fail: - Flag caption in green: the parameter is within the limits (PASS), - and in red: the parameter value is beyond the required range (FAIL). Rolling the mouse pointer over a QC flag caption displays relevant information on the sample metric in white text. Current sample metrics for particulates consist of: - Sample collection time (ctime): represents the actual air sampling time, in hours. - Air sample volume (volume): Integrated sample volume, in m3, reflects the air flow for a sampling time within acceptable limits. - Spectrum acquisition time (atime), - Station reporting time (rtime). The color of each label caption of QC flags is either green or red, depending on whether the parameter value is within the IMS Operational requirement or beyond specified limits. These sample metrics measure the spectral data quality from the CTBT verification value point of Page 18 of 70

19 view. In case the sample reliability is seriously compromised, the sample in question has to be released without category. Example of QT flag color coding convention: The relevant IMS specification related to collection time is that the sampling duration must be in the range 24 ± 10% h (which means in the range h). If the sampling time is within these limits, then the label caption is in green. Otherwise (sampling time beyond this specified range), the label caption will be in red. Figure 9: QC Flags panel The following table summarizes IMS specifications for major sample metrics as defined in CTBT/PC/II/1/Add.2, p. 48. Sample metric Collection time Sample volume Decay time IMS specification The sampling time must be in the range 24 ± 10% h (which means in the range h). the average flow rate during sampling must be greater than 500m 3 /h. The decay time must be less than or equal to 24 h, although a 10% tolerance is allowed, which means in effect that the decay time must be less than or equal to 26.4 h. Acquisition time The acquisition time must be greater than or equal to 20h, although a 10% tolerance is allowed, which means in effect that acquisition time must be greater than or equal to 18h. Reporting time Two reporting times are to be distinguished: - (a) station reporting time: time elapsed between Page 19 of 70

20 completion of spectrum acquisition and transmission from the station to the GCI. - (b) overall reporting time: time delay when the spectrum was received at the IDC data base. Here the reporting time is defined as the time between collection-start and transmission of the data from the station ( transmit time the date/time when the data were delivered to the GCI transmission device at the station). IMS specification on Reporting time sets the upper limit so that the three components (sampling, decay, and acquisition times) can vary within their respective tolerance limits but with an upper limit imposed by the reporting time requirement. The relevant IMS specification is that the reporting time must be less than or equal to 72h. Another set of QC flags reflects the spectral data quality itself: - Spectrum receipt check: this constitutes a check that all 11 Preliminary spectra were received, and that the 4-hour preliminary spectrum is processed successfully. A check on consistency of acquisition-start data for the preliminary spectra is included in the suite of tests deployed here. A warning Any problem with any of the tests results in a red warning signal, and the cause of the alarm can be found by right-mouse clicking on the signal. - Be-7 resolution check. this checks that the FWHM value for the Be-7 peak is less than 1.7 kev (the value set in agreed performance measures incorporated into the first-generation of IDC software). - Ba-140 MDC check: this checks that the minimum detectable concentration for Barium-140 in the current spectrum is below 30 micro-bq/m 3, as prescribed in IMS specifications for station performance. - QC10D check. This is a check involving a comparison of the QC spectrum (received just before the first preliminary sample spectrum) with the corresponding QC spectrum received 10 days previously. It is a check that there have been no long-term drifts in detector efficiency, resolution or gain. The program compares area, width (FWHM), and centroid energy for all peaks in both QC spectra with areas of 1000 counts or more. If there is a difference of more than 5%, a red signal is given to indicate a possible drift. - QCMRP check: this check is analogous with the QC10D check except that it compares the day s QC spectrum with that received the day before, as a check against sudden changes in detector performance. - Peak identification check. This facility checks the percentage of peaks for which nuclides are associated during automatic processing, to the total number of found peaks in the spectrum. If less than 80% of peaks are identified, the related QC flag turns to red indicating that energy and/or calibrations need to be checked and updated as appropriate. Page 20 of 70

21 6. SPECTRUM CATEGORY FULL sample spectra of particulates are categorized according to a five-level categorisation scheme as described in Appendix 1. The spectrum categorization level is shown with a prominent font under the Graphic Assistance panel of the main window. P3 The display consists of two parts: - an upper case letter (P, R) which reflects the spectrum status (P: Processed, R: Released), - followed by a number between 1 to 5 which represents the current categorisation level. - Examples: o P3 represents a sample under status P (Processed) and categorized at level 3. o R5 represents a sample under status R (Released) and categorized at level 5. o "P-" represents a spectrum under status P (and without category). This applies to spectra that are not intended to be categorized: sample PRELs, QC, Calibration, Detector background, Blank and Spike. o "R-" represents a sample spectrum after release without categorization. The sample category is updated after any relevant interactive action (for example, removing a CTBT radionuclide from peak identification, reprocessing with substantial change in energy and/or resolution calibration parameter sets). In such instances, all sample related category items are updated accordingly on OpenSpectra GUI (all parameters are refreshed from the database). A mouse-click on the Category text field results in a new window which summarizes categorization parameters specific to the radionuclides of interest that are (still) in the spectrum. The display, as shown in Figure 10 below, includes: - upper and lower abnormal limits, nuclide category along with the activity concentration of each nuclide in the sample, - automatic and current categorization levels for the sample. Page 21 of 70

22 Figure 10: Sample category window Notes on Figure 10: Auto-category is the automatic categorization based on automatic processing results of AutoSaint, as reflected in ARR (Automated Radionuclide Report). Sample category is the current spectrum category which may differ from autocategory, depending on interactive analysis actions via OpenSpectra tool. The Final spectrum category before sample release will be reflected in RRR (Reviewed Radionuclide Report). Page 22 of 70

23 7. DROP-DOWN MENUS OpenSpectra drop-down menus items are gathered under 12 menus which allow various interactive functionalities of relevance for the data review process: File, Sample, Comments, Options, Calibrations, Processing, Policies, Reports, Peaks, Release, Special nuclide review, Help. The current section provides an overview on menu items which are not systematically involved in interactive spectrum analysis. Key review functionalities will be described in details in the next section. Page 23 of 70

24 7.1. COMMENTS MENU: - View station comment : displays the comment under #Comment block of the PHD message as sent from the station (Figure 11). Figure 11: Station comment window - View general comment : displays general comment(s) provided interactively by Analysts during spectrum review via OpenSpectra (Figure 12). o Available comments are displayed in chronological order, with timestamp and analyst name. o A comment can be removed by clicking on glyph. Figure 12: View general comment window Page 24 of 70

25 - Add general comment : open a comment window editor (Figure 13) which allows Analysts to add general comments for the spectrum. Figure 13: Add general comment window Notes on Figure 13: - A drop-down combobox shows predefined text for usual comments. - By selecting a comment item, the full content is displayed in the lower text panel in edit mode which provides the possibility to modify the text. - Clicking on the button Insert comment will validate comment insertion in the database for the current spectrum identification number (table name: GARDS_USER_COMMENTS). - If cancel button is clicked, the changes will be rolled back. - In addition to predefined comments, the Analyst can type in a user comment in the text panel and validate in the same way. - Note: additional predefined comments can be inserted in the database table GARDS_COMMENTS_DEF. The default configuration already includes 44 following generic comments as detailed in Annex. Page 25 of 70

26 7.2. OPTIONS MENU: By clicking on the (single) menu item Graph colors, a window as shown in Figure 14 is brought up. This allows the user to customize colors of different series item of the main spectral display area. Figure 14: Color preferences window To change the color of any item, just click on the color label and a standard pick color dialog window will be prompted as shown in Figure 15. The user can then set a customized color and validate. Figure 15: Pick color dialog Page 26 of 70

27 7.3. CALIBRATIONS MENU Three menu items allow the user to visualize the plots of spectrum calibrations: (a) Energy calibration plot The plot displays the stations datapairs under #g_energy block of the PHD message along with the fit. A second degree polynomial fit is applied and fitting coefficients are also displayed (Figure 16). Figure 16: Energy calibration plot Note that this plot doesn t necessarily represent the actual energy calibration used in the processing. That might be updated by the competition algorithm implemented in Autosaint software which considers different candidates. More details are provided in the Reprocessing section of the current Users Guide. Page 27 of 70

28 (b) Resolution calibration plot The plot displays the stations datapairs under #g_resolution block of the PHD message along with the fit. A square root of second degree polynomial fit is applied and fitting coefficients are also displayed (Figure 17). Figure 17: Resolution calibration plot Note that this plot doesn t necessarily represent the actual resolution calibration used in the processing. That might be updated by the competition algorithm implemented in Autosaint software which considers different candidates. More details are provided in the Reprocessing section of the current Users Guide. Page 28 of 70

29 (c) Efficiency calibration plot - The displays the stations datapairs under #g_efficiency block of the PHD message along with the fit. A Log-Log polynomial fit (default degree 5) is applied and fitting coefficients are also displayed (Figure 18). Figure 18: Efficiency calibration plot Note that if VGSL simulated efficiency entries are available in the database for a detector setup (table name GARDS_EFFICIENCY_VGSL_PAIRS), these will be used as default option in the processing. Otherwise, Autosaint reverts back to efficiency pairs from the station. Available options in interactive mode via OpenSpectra are described with mode details under Reprocessing section of the current Users Guide. Page 29 of 70

30 7.4. POLICIES MENU The window contains a selection of instructions on a specific set of regularly occurring cases. Figure 19: Review policies window 7.5. REPORTS MENU The Reports menu contains a listing of the following different functions: Radionuclide activity Categorization Minimum Detectable Concentrations View Spectrum File Sample information window Nuclide Library Processing parameters QC flags report Automatic analysis log Analyst analysis Log View ARR View RRR View OpenSpectra Log Page 30 of 70

31 The following Figures (20 To 25) illustrate screenshots of the most important menu items under Report. Figure 20: Activity summary screen Figure 21: Sample category screen Page 31 of 70

32 Figure 22: MDC report screen Figure 23: Nuclide library window Page 32 of 70

33 Figure 24: Spectrum processing parameters Figure 25: QC flags report Page 33 of 70

34 8. INTERACTIVE REVIEW FUNCTIONALITIES After spectrum selection and checking of state-of-health and quality-control flags as described in related sections above, the main task of Analysts is to go through the automatic processing results in order to ensure that all peaks have been correctly identified. Suitability of processing parameters needs also to be checked and, if necessary, the spectrum has to be reprocessed interactively with adjusted parameters. Autosaint peak identification algorithms are designed to ensure a conservative behavior in automatic mode when CTBT-relevant nuclides are involved as possible associations. The final peak identification is left to Analyst judgment. Therefore, a special attention is paid to CTBT relevant fission or activation in peak association. In addition, correct nuclide(s) and /or explanatory comments need to be provided to peaks left without nuclide identification after the automatic processing. The following sections provide a contextual description of OpenSpectra main interactive functionalities as involved in the review process. These include: 1) Interactive reprocessing of data via OpenSpectra GUI; 2) Checking of peak search results in automatic mode; 3) Nuclide review facility for updating Autosaint processing results; 4) Germanium-Technetium discrimination tool (TcT); 5) Comparing to reference spectra; 6) Sample release options; 7) Xenon review window. Page 34 of 70

35 8.1. INTERACTIVE REPROCESSING OpenSpectra s Processing window (Figure 26 below) allows the Analyst to reprocess interactively the currently loaded spectrum. Note: This assumes that the spectrum is assigned to the current user which provides him/her all required write/update permissions on database related tables. By pressing the Reprocess button, OpenSpectra will invoke AutoSaint to reprocess the sample with updated options/parameters on the command line. Figure 26: Spectrum reprocessing window Options for energy (ECR) and resolution (RER) calibrations: The Analyst can chose between two options: (a) Run the competition algorithm of Autosaint by activating the Competition radiobutton in the lower left panel. The following related parameters can then be set: o Reference peak area threshold counts (only peaks with peak area above the defined threshold will be considered for updated calibration based on the current spectrum itself). o Minimum number of peaks that should satisfy the condition on area threshold for generating updated calibration based on the current spectrum itself. o Tolerance window for nuclide association to reference peaks. This can be a linear function of energy and the Analyst can set both the offset and the slope. (b) Force calibration coefficients of choice for energy and/or resolution by activating the radio-button Direct in the upper left panel. Page 35 of 70

36 o Depending on which of energy and resolution the Analyst wants to update, respective checkbox ( ECR, RER ) need to be checked/unchecked accordingly. o For both ECR and RER, the Analyst can either select one of the already available calibrations from the drop-down combobox or select CMDLine item and then type in new values for as fitting coefficients in respective Edit fields. A dummy entry (0.01) is also needed for Err field(s). o The following terminology is used for ECR and RER calibration options: INPUT: based on calibration pairs in the PHD message; MRP QC: calibration from the Most Recent Prior Quality Control spectrum; MRP A: calibration from the Most Recent Prior Automatically processed spectrum; MRP M: calibration from the Most Recent Prior Manually reprocessed spectrum; INITIAL: based on reference peaks in the current spectrum. CMD Line: calibration coefficients provided by the User (in the command line) Options for efficiency calibrations: In automatic processing configuration, Autosaint uses VGSL efficiencies when available, as a default option. It s only in case there is no VGSL entries for a detector that Autosaint reverts back to station efficiency pairs. If VGSL based efficiencies are available in the database (table name GARDS_EFFICIENCY_VGSL_PAIRS) for the detector_id in question, the check box Use VGSL in the upper right panel of the window gets enabled. In this case, the Analyst can choose between two sets of efficiency data sources for reprocessing in interactive mode: (a) station pairs as provided under #g_efficiency block of the PHD message (stored in GARDS_EFFICIENCY_PAIRS database table), by unchecking the check-box Use VGSL. Optionally, the polynomial degree (poly. Degree) can also be changed (the default is set to 5). (b) VGSL based efficiencies, by checking the check-box Use VGSL. Page 36 of 70

37 Risk level setting: Autosaint uses the SCAC/LC Curve (LCC) method for deciding on presence of peaks at any energy of the spectrum range. The decision criterion is based on the relative position of SCAC with respect to LCC. It is enough to have only one channel above the LCC for a peak to be recognized as present. This risk system is channel based and the sensitivity of peak-search is set by the height of the LCC above the baseline. Wherever the SCAC curve touches or crosses the red Lc curve, the structure is deemed to be a real peak (at the pre-selected level of risk defined by the height of the Lc curve above baseline). Assuming a well-known baseline, the nominal risk is defined as the risk of making an incorrect judgment that a channel represents part of a real peak when, in fact, it does not (i.e., a Type 1 error). The higher the risk setting, the greater the number of expected false peak detections. In routine Operations, the default risk level used in LC calculation is set to 0.001%. This can be changed interactively as appropriate, by selecting from a predefined set of standard values in the drop-down combobox labeled Risk Level (central right panel of the window). If the spectrum is reprocessed with a new value for the risk level, this new value will be used in LCC calculation. Consequently, this will produce different results of peak search algorithms Tolerance window for peak identification: Autosaint supports a nuclide location window as a linear function of energy (Tol. = a +b.e). This allows a better compensation for lower resolutions in the high energy range. The default peak association is based on a static tolerance window of 0.8 kev. OpenSpectra reprocessing window offers the possibility to change both the offset and the slope of the tolerance window Other parameters: In addition to the high-level parameters described above, OpenSpectra allows wider possibilities for changing the setting of any processing parameter of interest. This can be done by simply typing the parameter name, followed by = and the value, in the Edit field under Other parameters in the lower-right panel of the window. The new settings will then be communicated to Autosaint when invoked for reprocessing. Page 37 of 70

38 If more than one parameter needs to be used, they just need to be separated by a blank space: <PROC_PARAM_NAME1>=<setting1>{white space}<proc_param_name2> =<setting2> Top level processing parameters are listed in the header part of Autosaint logfile. The complete list is stored in the database PEAK SEARCH DIALOG The Peak search dialog window, as shown in Figure 27 below, displays peak search and nuclide identification results of the spectrum. Before starting the interactive review, the display reflects automatic processing findings. The content will then get updated after each relevant action during the interactive review. Figure 27: Peak search dialog window The display is structured into two levels of information and functionalities: the left part is peak oriented and the right side is mainly nuclide oriented. (a) Peak oriented panel: The grid on the left side shows relevant information on each peak found in the spectrum: Peak identification number (Id), energy (kev), centroid channel, FWHM (kev), peak area, detectability (det), number of peak added comments (#Cmnts) and nuclide(s) associated to the peak. Interactive functionalities: o By selecting a peak line in the Peak search window, the spectrum display area gets automatically synchronized in zoom in mode with the current energy. o The red arrows " <<<<" and " >>>>" on top of the peak list allow left and right browsing of peaks with possibly anthropogenic nuclide association. These need to be interactively validated by the Analyst. Page 38 of 70

39 o The black arrows "<<" and ">>" below the peak list allow left and right browsing of uncommented peaks. Further actions on these peaks will consist in interactive nuclide association and/or appropriate comments. o The button Nuclide review window activates a graphical based decision support tool which assists the Analyst when dealing with peak association. This will be presented in detail in a dedicated section below. o The button Add peak comments allows the Analyst to provide comment(s) to peaks of interest as appropriate. o The button Add general comment allows the Analyst to provide a general comment for the spectrum. (b) Nuclide oriented panel: The right panel displays details on nuclide association and nuclide comments. Interactive functionalities: o The button Remove nuclide allows the Analyst to disregard a previously associated nuclide to a peak. o The button Add nuclide comment allows the Analyst to provide a comment to a nuclide of interest as appropriate. o The display changes accordingly between Nuclides and Comments control mode in a synchronized way with activated action Adding a peak comment To add a comment to the currently selected peak: - Click on the button Add Peak Comment. Then the right panel switches automatically to Comments control mode. The Analyst can either select a predefined comment from the drop-down combobox or type in a user defined comment in the text Edit field (Figure 28). Figure 28: Peak comment insertion Page 39 of 70

40 - To validate comment insertion, click on the button Insert Comment. Selected text will then be stored in the database and the number of peak added comments #Cmnts gets incremented for the peak in focus (Figure 29). Figure 29: Peak comment validation Removing a nuclide To remove a wrongly associated nuclide from the currently selected peak: - Select the nuclide to remove from the list under the right panel Nuclides (example: IR-190 in Figure 30); Figure 30: Remove nuclide functionality - Click on the button Remove Nuclide. The Analyst will then be prompted to choose between two possible options (Figure 31): (a) the nuclide will be removed only from the peak in focus or, (b) all peaks where the nuclide in question is also available will be affected. Figure 31: Remove nuclide options Page 40 of 70

41 - (a) To remove a nuclide (IR-190 in the current case) ONLY from the currently selected peak, click on the left button Remove Nuclide IR-190 from this peak only. After validation, a generic comment as shown in Figure 32 in will be inserted in the database and a comment ID will appear in the #Cmnts column for the peak in focus. Figure 32: Remove nuclide validation - (b) To remove a nuclide (IR-190 in the current case) from ALL peaks where IR-190 is wrongly given as a possible association, click on the central button Remove Nuclide IR-190 completely from the spectrum. After validation, a generic comment will be inserted in the database and a comment ID will appear in the #Cmnts column for the peak in focus and all other peaks with the nuclide in question is (IR- 190 in the current case) is given as a possible association. - Note that when an anthropogenic nuclide is removed from a peak, all categorization related parameters are updated automatically. Page 41 of 70

42 8.3. THE NUCLIDE REVIEW FACILITY The Nuclide Review facility is a decision support system which assists the Analyst while exploring peak identification options. Relevant information is displayed in a graphical mode for other gamma lines of all possible nuclide associations that have peaks within the tolerance identification window around the energy in question. Available possibilities are compiled in graphical mode so that the Analyst can consult each option interactively one by one. Nuclear data of interest are made available on the same window with the possibility to extend the search to other nuclides in the Nuclide Library by just changing the energy tolerance setting. In addition to the peak in focus, additional lines of each possible nuclide are co-plotted in a contrasted way depending on whether a peak is present or not: Combining all elements of such a user friendly dashboard environment allows the Analyst to take the right decision in a reliable way. The Nuclide Review facility can be activated in two possible ways: - By selecting the menu item Particulates Nuclide Review under the menu Special Nuclide Review. - From the Peak search window, by clicking on the button Nuclide Review Window. The Window as shown in Figures 33, 34 and 35 (see the sections below) consists of the following items: - The plot in the upper-left corner displays a zoom in around the current peak of interest. - The Nuclide listbox to the right contains possible nuclide candidates for peak association. These have energy lines within the tolerance window (default is 0.8 kev). - The grid in the upper-right corner provides nuclear data of the selected nuclide. These include nuclide type, half-life and energy line details. - The plots in the lower panel correspond to other emission energy lines (if any) for the selected nuclide. If more than three lines are available, the Analyst can scroll left and right for additional energy plots by clicking the arrow buttons < and >, respectively. - Each plot footer shows contextual information for the selected nuclide and energy line: o Peak detectability in the spectrum, o Gamma emission probability (abundance), in %, o Detection Probability Index (DPI), defined as the product Abundance x Efficiency. Page 42 of 70

43 Key features of the Nuclide Review Window: - Each plot displays original spectrum counts in a zoomed in mode around energies of interest. - In each plot, a shaded (grey) Gaussian peak is fitted to the spectral structure, with the peak centroid being at the energy associated to the selected nuclide, where the peak would be expected. - In lower plots, shaded (grey) areas represent theoretical peaks, based on the size of the peak under investigation (upper left window) and taking into account the abundance ratio, detector efficiency ratio, and coincidence correction factor(s) if applicable. - The plot background is coloured to allow immediate recognition of real peaks from insignificant SCAC structures: o blue background means the peak is real (SCAC above LC: detectability 1), o black background indicates there is no significant peak in the spectrum (SCAC below LC: detectability<1) around the energy in question. o This color coding convention applies to both the peak under investigation and other peaks of the selected nuclide. - The peak location tolerance window can be set interactively to a different value in the spin Edit field next to Tolerance, as appropriate. Then, by clicking the Search button, a new query is run and the list of possible nuclides is updated on the display. - The arrows buttons << and >> in the upper panel provide an internal option for browsing all uncommented peaks in the spectrum (without getting back to the Peak Search window). Note that possibly anthropogenic peaks are skipped by these browsing arrows. Such peaks are only accessed from the Peak Search dialog window, as described above. ILLUSTRATION CASE The screenshots in Figures 33, 34 and 35 below illustrate an example on how the Nuclide Review Window can be used to check nuclide identification for a peak around kev. The window shows all three nuclides with possible association to the peak: - Ag-110m, with a gamma line at kev, - Pb-212F, with a gamma line at kev, - Zr-97, with a gamma line at kev Although, all three nuclides have energy lines within the tolerance peak search window, the following considerations will allow the Analyst to provide the correct identity to the peak in question: Page 43 of 70

44 - An isotope is to be disregarded if observed peak areas at energies where other lines are not in consistent ratios with respect to the peak under investigation. - The right nuclide association will be the one where all its other lines are in consistent peak area ratios to the peak in focus. The Analyst will ultimately decide the correct peak identity by applying these key rules to each of possible nuclide associations. (a) Assuming Ag-110m By selecting AG-110M in the listbox, the Window displays related information as in the following Figure 33. Figure 33: Nuclide review window with Ag-110m selected. All lower plots show BLACK background for other peaks of Ag-110m (884.69, and kev in the screenshot) which means that these peaks are NOT present in the spectrum. Taking into account that the DPI of the peak at kev is approximately 80% of the DPI at the peak kev (under investigation), associated detectabilities would also be in the same proportion if the peak in question belongs to Ag-110m. However, actual detectability at kev is only ~ 15% compared to kev. Therefore, Ag-110m is unlikely to be the correct peak identification. Page 44 of 70

45 (b) Assuming Pb-212F By selecting PB-212F in the listbox, the Window displays related information as in the following Figure 34. Figure 34: Nuclide review window with Pb-212F selected. The lower plots show BLUE background for all other peaks of PB-212F (238.6, and kev in the screenshot) which means that these peaks are also present in the spectrum. Furthermore, expected areas (in grey) are in excellent consistency with actual peaks in the spectrum for other lines of PB-212F. The combination of these elements provides enough confidence for deciding that the peak in focus belongs to PB-212F. Page 45 of 70

46 (c) Assuming Zr-97 By selecting ZR-97 in the listbox, the Window displays related information as in the following Figure 35. Figure 35: Nuclide review window with Zr-97 selected. The lower plots are shown with BLACK background for all other peaks of Zr-97 (743.33, and kev in the screenshot) which means that NONE of these peaks is present in the spectrum. Taking into account that the DPI of the peak at kev is around 130% of the DPI at the peak kev (under investigation), associated detectabilities would also be expected in the same ratio if Zr-97 is the correct nuclide identification to the peak in question. However, actual detectability at kev is only ~ 40% compared to kev. Based on these findings, Zr-97 is also to be disregarded as possible peak identification. Page 46 of 70

47 8.4. THE TECHNETIUM TOOL (TCT) One of the most common issues when dealing with IMS spectra is related to the discrimination between: - Tc-99m, one of the 83 CTBT-relevant nuclides, with a gamma energy at kev and a half-life of 6 h. - and Ge-75m, produced by interaction of neutrons from cosmic radiation with the Germanium crystal of the detector itself, with a gamma peak at kev. Another aspect of the issue is to distinguish whether Tc-99m is present alone in the sample or it is supported by its parent Mo-99 with a gamma energy at kev and effective half-life 66h Theoretical considerations In theory, such cases can be handled on the basis of peak position around 140 kev and peak area at kev. However, the usefulness of these elements require the following conditions: - Energy discrimination between Ge-75m and Tc-99m only applies if the peaks are large enough which allow centroïd determination with low uncertainty. - The primary line of Mo-99 at kev is only exploitable if the nuclide is present at relatively high concentration (expected ratio 140 kev:739 kev is in the range 4 to 13 for detector types used in the IMS radionuclide stations). In practice, Analysts have also to deal with spectral data where observed peaks are close to the detection limit. In such situations both above discrimination elements are compromised: - Uncertainties in peak shape cause uncertainties in centroïd position larger than the required 0.81 kev (difference between Ge-75m and Tc-99m). - Expected Mo-99 peak area at kev is much below the detection limit. Taking into account these inherent difficulties, OpenSpectra is equipped with a Technicium- Germanium analysis tool (TcT) which is based on four practical decision elements. (a) Ge-75m in the detector background Background and blank spectra provide peak area of Ge-75m with relatively low uncertainty. Daily variation of cosmic neutron flux is around 10% due to fluctuations in atmospheric pressure. Page 47 of 70

48 (b) Ge-71m in the sample spectrum The presence of kev peak due to de-excitation of Ge-71m is expected with comparable peak area than Ge-75m at ~140 kev. The reason is that the product abundance x production cross-sections x gamma emission probability x detection efficiency is almost the same for these Ge isotopes. In other words, for any Ge-75m peak at ~140 kev, there should be an accompanying peak due to Ge-71m of similar size at kev. (c) Time development of peak detectability Peak detectability around 140 kev during the normal 24h spectrum acquisition time will vary following differentiated patterns depending on whether the peak is produced by Ge-75m, Tc- 99m alone or supported Tc-99m: - If the peak results from Ge-75m, the peak area is expected to grow quasi-linearly with the acquisition time, due to the too short half-life of Ge-75m and which is produced continuously with a constant rate. - If the peak is produced by unsupported Tc-99m, peak detectability is expected to increase only slowly because Tc-99m will be decaying significantly due to its 6h halflife. - Supported Tc-99m, on the other hand, will decay only slightly because of the 66h half-life of the Mo-99 parent. This will be reflected by a faster increase of 140 kev peak detectability Operating principles The TcT window (as shown in Figures 36 and 37, below) of OpenSpectra is activated by selecting the menu item Te/Ge Review under the menu Special Nuclide Review. The TcT translates the above discrimination concepts into quantifiable decision support elements which allow a reliable association of a peak around 140 kev, even close to the detection limit. (a) Energy discrimination - The plot in the upper left corner displays original spectrum counts in a zoomed in mode around 140 kev. - Two vertical dashed lines are plotted at respective energies of Ge-75m and Tc-99m where peak centroïds would be expected. Page 48 of 70

49 (b) Background subtraction The first point to consider is whether or not the peak at ~140 kev is above the background Ge-75m level. The TcT consults filed data on background spectra and corrects the sample spectral peak by applying 90% of the background level (to conservatively allow for the 10% variation in neutron flux). The SCAC/LC method is then applied to the residual for deciding whether or not a substantial peak is left there. If no residual peak is present, the peak is deemed to be due to normal background Ge-75m. This is shown by the upper and lower left plots in the display screen (Figure 18) which contains: - the existing spectral peak in SCAC form, - and the residual structure after background subtraction. - The reduction factor is also shown this is the ratio of peak areas after and before background subtraction, so a ratio close to 1.0 indicates little effect of background (i.e., the peak is probably not due to ge-75m). (c) Peak ratio 140:198 kev: The spectral structure at 198 kev (gamma line of Ge-71m) is displayed in the upper right screen to enable the analyst to judge the degree of similarity. The ratio of peak areas, 140keV:198keV is also indicated: - The presence of a peak at 198 kev with a ratio around 1 compared to 140 kev means the peak around 140 kev is due to Ge-75m. - The absence of a peak at 198 kev or a ratio considerably greater than 1 with respect to 140 kev indicates the possible presence of Tc-99m. (d) Preliminary spectra The TcT calculates expected detectability at 2-hour intervals during the acquisition time, corresponding to preliminary spectra. These are plotted in the lower-right plot. The diagram shows four colored areas: - A pink band (Tc-99m band): expected detectability for unsupported Tc-99m, - A dark blue band (Mo-99 band): expected detectability for supported Tc-99m, Observed peak detectability around 140 kev in the PREL spectra of the sample under analysis are also co-plotted on the same so-called detectability diagram. Notes: - The Tc and Mo bands are curved, showing the effect of decay during acquisition on expected peak size. This expected curvature would obviously be affected by the presence of Pb-212 in the sample filter because it also decays simultaneously during the spectrum acquisition, albeit with a longer half-life, and so affects the apparent change in count-rate. - Band boundaries correspond to two extreme scenarios: (a) Assuming that there is no Pb-212 in the spectrum region (constant background under the 140 kev peak); (b) Assuming that the background is entirely due to Pb-212. Page 49 of 70

50 Illustration cases The following examples illustrate how the TcT window is used for peak identification in two contrasted sample spectra in terms of 140 kev peak association. Spectrum A: The peak around 140 kev is due to unsupported Tc-99m Figure 36: TcT window with unsupported Tc-99m Interpretation elements of Figure 36: (a) The upper-left plot shows a peak of significant size and the centroid is on the side of Tc-99m energy. (b) The net SCAC obtained after background subtraction, is still above LC, as indicated by the black background color of the lower-left plot. (c) The 198 kev Ge-71m peak, as shown in the upper-right plot, is much smaller than the 140 peak. This is translated by a 140:198keV ratio of ~ 30 which is considerably greater than 1. (d) On the detectability diagram plot, the 10 dots associated with the sample PREL spectra fit well within the Tc-band (boundaries in pink). The combination of these findings is interpreted as clear evidence that the 140 kev peak is due to unsupported Tc-99m. Note: If Tc-99m is supported, instead, the same logic as above applies. The only difference is that detectabilities of the PREL spectra would fall within the blue band of Mo-99. Page 50 of 70

51 Spectrum B: A peak around 140 kev is due to Ge-75m Figure 37: TcT window with Ge-75m Interpretation elements of Figure 37: The absence of Tc-99m is indicated by the following: (a) The peak centroid energy is rather close to Ge-75m line ( kev). (b) The Ge-75m and Ge-71m plot (upper right) shows peaks of similar size, as indicated by a peak-area ratio of (c) The net resulting SCAC after background stripping is below LC, as highlighted by the blue background color of the lower left plot (a reduction factor is 1 means that the peak in the sample is completely subtracted). In other words, this clearly indicates that the peak is due only to Ge-75m from the background. (d) Observed detectability values in the PRELs, as displayed in the lower-right plot, are scattered and don t fit with any of the Tc or Mo bands. The combination of these findings is interpreted as a clear evidence that the 140 kev peak is due only to Ge-75m. Page 51 of 70

52 8.5. COMPARING SPECTRA Compare options functionalities are available under Graph Assistance panel of the main window (Figure 38 below). Comparing to Most Recent Prior (MRP) spectra provides additional information to check for new observed peaks of sudden changes in the spectrum baseline. Comparing a sample spectrum with a blank filter or a detector background spectrum provides useful input to the review process. This helps in checking for peaks that may be explained by contributions from either the surrounding environment or the filter material itself. In addition, comparing a FULL spectrum with its preliminary spectra is a helpful feature when deciding on identities for short-lived nuclides. Figure 38: main display in comparison mode OpenSpectra GUI offers three options for choosing the second spectrum against which to compare: (a) One of the available PRELiminary spectra for the same sample. (b) Most Recent Prior spectra from the same station. (c) Any spectrum in the file system (not necessarily from the same station). Just a valid spectrum identification number needs to be typed in the Edit field under Compare Options panel. Available spectra related to (a) and (b), can be selected from the drop-down combobox under MRPs/PRELs. Page 52 of 70

53 After spectrum selection, the button Compare is enabled. Clicking on it loads the second spectrum of interest. The Enable check box allows switching between compare on and off modes. The vertical shift in between the two spectra can be changed through the YDiff Edit field or associated spin-button. Note that this scaling display parameter can also be adjusted once the second spectrum is loaded. Page 53 of 70

54 8.6. SAMPLE RELEASE Depending on the data type and quality, four different options are available for releasing a spectrum: a. Release with category, b. Release without category c. Mark as Viewed d. Change status to others (a) Release with category After successful review of a FULL sample spectrum with good CTBT verification value (see policies in Annex), the Analyst can proceed with the normal release. The screenshot in Figure 40 below shows the Sample Release window which is activated by selecting the menu item Release with categorization under the menu Release. Figure 39: sample release window Page 54 of 70

55 The Sample Release Window contains three panel areas: (a) The upper panel shows overall sample category information (Auto-category as generated by the automatic processing; Category as updated after the interactive review of the spectrum). (b) The central panel provides nuclide specific categorization details after the interactive review: o Statistics parameters of the categorization algorithm (Lower and upper abnormal limits, central value), o Activity concentration, o Resulting nuclide category. o Hold Cat. checkbox if checked, this toggles the Hold flag on a nuclide. (c) The lower panel contains a text area which can be used to type in new comments regarding the sample. o Note that the general comment(s) provided through the main menu are not displayed here. However, these are stored in the database and reflected in the RRR related section after the spectrum is released. On the bottom of the window there are two buttons: Release sample This button allows the Analyst to validate the sample release action. After validation through the confirmation dialog message, the following information dialog message is displayed. The spectrum status will then change from P to R and a RRR is generated and stored in the file system. Once released, the spectrum can only be loaded in read only mode (no reprocessing or any interactive update action are allowed any more). (a) Cancel button for closing the window without releasing the sample. Page 55 of 70

56 (b) Release without category In case the data quality is deemed unsatisfactory from the verification value point of view (see policies in Annex), a FULL sample spectrum is released without categorization. By selecting the menu item Release without categorization under the menu Release, the window as in Figure 41 will be prompted. It contains similar information than the normal release window with the possibility to type in an optional comment on the spectrum. Figure 40: Release without category window After validation via the confirmation dialog message, the spectrum status will be changed from P to R and the category will be set to NULL in the database. Page 56 of 70

57 (c) Change status to Viewed This release option applies to spectra that are not intended to be categorized: (b) PREL sample spectra, (c) Quality Control spectra, (d) Calibration spectra, (e) Detector background, (f) Blank spectra, (g) Spiked samples. Note that neither auto_category nor (reviewed) category applies for these spectra types. By selecting the menu item Change status to Viewed under the menu Release, the window as in Figure 42 will be prompted. Figure 41: Change status to Viewed The Analyst can optionally add a general comment in the text field area. After validation, the status will change from P to V and the spectrum can only be loaded in read only mode. Page 57 of 70

58 (d) Change status to Others This release option applies offers open possibilities for changing the status of any spectrum to any upper case letter. By selecting the menu item Change status to Viewed under the menu Release, the window as in Figure 43 will be prompted. Figure 42: Change status to Others The Analyst can either select one of the conventional status values from the combobox Change status to, or simply select type in any letter in the Edit field on the right. Predefined status options of interest include D (mark as Duplicate) and B (mark as Bad) as explained in Section 2 of the document. A general comment can also be optionally added in the text field area. After validation, the status will change from P to the status of choice as provided by the Analyst and the spectrum can only be loaded in read only mode. Page 58 of 70

59 8.7. XENON SPECTRUM ANALYSIS After loading a SPALAX spectrum, OpenSpectra automatically switches to the Xenon mode. (a) Quality Control (QC) Flags for SPALAX The QC Flags panel on the main window shows check results for SPALAX relevant sample metrics (Figure 43): - Stable Xenon volume - Sampling time - Acquisition time - MDC for Xe Reporting time Figure 43: QC panel for SPALAX samples The result is displayed in traffic-lights like color coding for each QC flag caption: - Green: the parameter fulfills IMS operational requirements. - Yellow: the parameter doesn t meet certification requirements but the sample quality is considered still fair. - Red: the data quality is seriously impacted from the CTBT verification point of view and the sample is not reliable. Rolling the mouse pointer over a QC flag caption displays relevant information for the sample metric in white text, including the test result (Pass/Fail/Warning). Page 59 of 70

60 The current thresholds set for SPALAX samples are as follows: Sample metrics thresholds for SPALAX sample data Sampling Time 12 h h 48 h AcquisitionTime 12 h 21.6h 26.4h 48 h Xenon Volume 0.2ml 0.87ml MDC mbq/m 3 1mBq/ m 3 5 mbq/m 3 Reporting Time 10 h 48h 96 h (b) Xenon review window The Xenon screen as shown in Figure 44 is brought up automatically, upon loading a SPALAX sample spectrum. Note that this window can also be activated via the menu item Xe Nuclide Review of the menu Special Nuclide Review. Figure 44: Xenon window in default mode Page 60 of 70

61 The default Xenon window displays five plot regions: - X-ray range in the upper left; - Xe-135 gamma line (250 kev) in the upper right; - Xe-131m gamma line (164 kev) in the lower left; - Xe-133m gamma line (233 kev) in the lower central plot; - Xe-133 gamma line (81 kev) in the lower right. Analysis results are also displayed in the upper central panel for the four Xenon isotopes based on both methods: - Deconvolution method using the FULL spectrum; - Deconvolution method including the PREL spectra (if available). Note: These two methods are referred to as method_id 11 and 12, respectively, in GARDS_XE_RESULTS database table which holds processing results of SPALAX data. Activity concentrations and associated uncertainties are expressed in mbq/m3, backward decay-corrected to sampling time. (c) Display modes of Xenon window The user can change the display mode of all active plots through the following controls on the window: - The button SCAC shows and hides the chart series related to the Single Channel Analyzer Curve (default Magenta line). The button background color changes accordingly. - The button LC shows and hides the chart series related to the Critical Limit Curve (default Red line). The button background color changes accordingly. - The button Baseline shows and hides the chart series related to the spectrum Baseline (default Orange line). The button background color changes accordingly. - The radio-buttons Linear and Log 10 allow switching the vertical axis between linear and logarithmic scales. - The radio-buttons All isotopes and Select isotope allow to switch between two display modes: o By activating All isotopes, the window displays all five plots as described above; o By activating Select isotope, the user can select individual isotopes from the pop-down combobox. The window switches then to single display mode (Xray region in the upper-left plot and the gamma region for the selected isotope in the lower-left plot). Figure 45 is a screenshot for the case of Xe-133. Page 61 of 70

62 Figure 45: Xenon window in single mode (d) Enabled functionalities for SPALAX spectra The following features of the GUI work in the same way than for particulates: - Main display window. - Spectrum selection options - Graphical and Information panels. - Interactive reprocessing. - Items under the menus Comments, Options, Calibrations. - Peak Search window (in read only mode), since there is no need for peak identification in interactive mode. - Automatic/Analyst analysis logs under the menu Reports. - The menu items under Release : spectrum status is updated in the database but no report is generated in the file system. Page 62 of 70

63 APPENDIX 1: ANALYSIS OPERATIONAL POLICIES FOR CTBTO PARTICULATES DATA Spectrum categorization policy IMS gamma spectra are intended to provide Treaty-verifying evidence pertaining to the collection period, which is nominally a 24-hour period. On completion of interactive review of full-sample spectra, the Analyst is provided with two choices: to release the spectrum (i.e., generate the RRR) with categorization or without categorization. Categorization is an event screening tool intended to facilitate the work of NDCs by drawing attention to spectra containing relevant anthropogenic radionuclides. Level 1 and 2 spectra contain only natural or non-relevant anthropogenic nuclides, while level 4 and 5 spectra contain relevant nuclides and are therefore of prime interest. Although the releasing of a spectrum without categorization is relatively uncommon, it happens from time to time so guidelines and procedures are required in order to ensure that the IDC has a consistent response in such situations. The purpose of this document is to define such operating guidelines for application in the IDC. Common NDC practice is to scrutinize products pertaining only to level-4 or level-5 spectra. This places special responsibilities on the IDC: to ensure that all cases of anthropogenic nuclide detections, or even suspected or possible cases, are reported with level 4 or 5 categorization; and to ensure that there is the highest possible confidence, within the limits imposed by IMS operating criteria, that all spectra categorized at levels 1 or 2 do not contain relevant anthropogenic radionuclides. There are thus two different confidence concepts which apply in spectrum categorization: (1) that relevant anthropogenic nuclides should be reported on a conservative "proven or suspected" basis, in which NDCs are ensured of having the opportunity to decide for themselves in the latter case; and (2) the reporting of only natural or non-relevant nuclides must be conducted with the highest possible confidence that no anthropogenic nuclides are present. The over-reporting or mistaken reporting of anthropogenic nuclides is thus more excusable than the failure to report such nuclides. Spectra involving relevant anthropogenic nuclides at anomalous concentrations must be categorized as level-4 or level-5 spectra and released accordingly. Normal IDC operating practice is to maximize the confidence that all such cases are reported. This extends to declaring a nuclide as "present" if there is sufficient doubt that it is not present. This is conservative reporting, which will undoubtedly involve some "errors on the safe side", which is preferable to the alternative. Page 63 of 70

64 Factors influencing the decision The decision of whether or not to categorize a spectrum is influenced by three types of factor as described below: (a) IMS specifications (b) Spectrum quality (c) Presence of anomalous anthropogenic nuclides 1. IMS specifications The sensitivity of any analysis is fundamentally determined by the IMS monitoring specifications which were set originally to provide the required sensitivity for the monitoring network. Confidence in categorization is therefore likewise limited by the degree of adherence to these specifications. If monitoring practice does not meet the specifications then the spectra should not be categorized. The IMS specification used for spectrum categorization according to the minimum requirement specified in CTBT/PC/II/1/Add.2, p.48 (reproduced below), are as follows: (a) collection time is within the bounds hours (prescribed 24h with 10% tolerance) and; (b) decay time is less than or equal to 26.4 hours (prescribed 24h with 10% tolerance) and; (c) acquisition time is greater than or equal to 18 hours (prescribed 20h minimum with 10% tolerance) and; (d) reporting time is less or equal to than 72 hours and; (e) average air flow-rate is greater than or equal to 500 m 3 /h and; (f) peak FWHM is less than 2.5 kev at 1332 kev and; (g) 140 Ba MDC is less than 30 μbq/m 3. Three of these specifications reporting time, resolution, and 140 Ba MDC - are anomalous in their application to determination of data availability in that they either generally do not affect data quality or are beyond control of the station operator. Reporting time, however, controls the upper limits on acquisition time, which would otherwise be left unbounded - so it remains an important specification. Experience has shown that 140 Ba MDC regularly exceeds 30 μbq/m 3 at some stations due to high prevailing atmospheric 212 Pb concentrations. The lower limit of 10 μbq/m 3 is applied to the blank spectral measurement and is checked periodically according to the IMS Operational Manual, but is not applicable to sample spectra. A severe degradation of resolution would affect data quality, but whether or not it is significant enough for the data to be classified as Page 64 of 70

65 unavailable would be taken into account in the categorization of the spectrum as discussed below. The specifications of resolution and the 30 μbq/m 3 MDC limit for 140 Ba are therefore reduced to the status of flags to alert the PTS of possible problems occurring at the station, rather than being enforced in specification compliance and data availability determinations. 2. Spectrum quality The simple meeting of IMS operating criteria does not necessarily mean that spectra are fit for categorization. Experience has indicated several scenarios under which the IMS specifications may be met, but the data are meaningless. An example is the malfunctioning of a robotic sampler, where the robot failed to place the filter on the detector so the spectra sent to the IDC were effectively 24h background spectra. Other spectra are rendered useless by equipment malfunction causing noise spikes, peak broadening, empty channels, double peaks, and major gain shifts, for example all of which reduce or remove confidence in categorization. In addition to the meeting IMS specifications the data must therefore also satisfy IDC analysts that they are meaningful. 3. Presence of anomalous anthropogenic nuclides As one of the main purposes behind categorisation is event screening in order to draw attention to suspicious events, spectra containing evidence of nuclides which would result in level 4 or 5 categorisation must be categorised, even if conditions (1) and (2) above are not satisfied. Page 65 of 70

66 Operational policy The above philosophy carries through to operational policy on whether a sample spectrum is to be categorized or not. Spectra are categorized under the following conditions: 1. The following IMS requirements are met: - collection time within the bounds hours, and - decay time less than or equal to 26.4 hours (and greater than 6h), and - acquisition time greater than or equal to 18 hours, and - reporting time is less than or equal to 72 hours, and - average flow rate greater than 500 m3/h. 2. There are no problems with the spectral data which make nuclide detection and identification uncertain, such as: - power supply failure during acquisition, or - detector failure during acquisition, or - major electronic problems causing major gain shifts, or - resolution degradation making peak identity uncertain. Note that in case anomalous relevant anthropogenic nuclides are detected in the spectrum (i.e., nuclides with a categorization level of 4), or suspected to be indicated in the spectrum, the sample is categorized even if some of the above conditions are not met. Spectra are not categorized under the following conditions: (a). Any of the following IMS requirement is not met, through any one of the following conditions: - collection time is outside the bounds of hours, or - decay time is greater than 26.4 hours (or less than 6h), or - acquisition time is less than 18 hours, or - average flow rate is less than 500 m3/h, or - (collection time + decay time + acquisition time) is greater than 72 hours or (b). There are problems with the spectral data which make categorisation uncertain, as in (2) above; and (c). No anomalous relevant anthropogenic nuclide is detected in the spectrum. Page 66 of 70

67 Notes on IMS specifications 1. Time specifications allow for an uncertainty of 10%, except for the reporting time parameter. 2. This value can be reduced down to a minimum of 6 hours if a suspicious event is detected by other stations or techniques. 3. This value allows for authentication measurements for manual systems. 4. This global value includes 80% filter efficiency and collection efficiency of the incoming air circuitry. 5. The upper limit is intended for high-background areas. 6. Certification procedures to be defined for baseline sensitivities (a posteriori MDCs) as well as the efficiency. Sample preparation losses should not affect baseline sensitvities. 7. This format should make provision for auxiliary data, authentication data and state-ofhealth data. 8. Provision should be made for spare parts in particular areas where periodicity of transportation facilities is more than 7 days. Reference: CTBT/PC/II/1/Add.2 Page 67 of 70

68 List of CTBT relevant nuclides for particulate samples: The table below lists the 83 CTBT radionuclides (42 activation products and 43 fission products). ACTIVATION PRODUCTS FISSION PRODUCTS Nuclide Half-life Nuclide Half-life AG-106M D AG D AG-108M Y BA D AG-110M D CD D AS D CD-115M D AS D CE D AU D CE D AU-196M H CE D AU D CS D BA Y CS Y CO D EU Y CO D EU D CO Y EU H CR D I H CS D I D CS Y I H EU Y I H EU-152M H LA D FE D MO D GA H NB D IR D ND D IR D PD H K H PM D MN D PM D NA H RH D NP D RU D PB D RU Y RB D SB Y RB D SB D RH D SB D SB D SB H SB D SM D SB D SM H SC D SN D SC D SR H TM D TC-99M H U D TE-129M D W H TE-131M D Y D TE D ZN D Y D ZN-69M H Y H ZR D ZR D ZR H Page 68 of 70

69 Categorization scheme for particulates Particulate samples are categorized on the basis 5 level scheme as illustrated in the diagram below. The five categorization levels are defined as follows: Level 1 (normal background): The spectrum contains evidence of only natural radionuclides at atmospheric concentrations which are within the normal range for the station. Level 2 (anomalous background): The spectrum contains evidence of either natural radionuclides at atmospheric concentrations outside the normal range, or the non-relevant anthropogenic radionuclides (or both these conditions). Level 3 (normal anthropogenic conditions): The spectrum contains evidence of a relevant anthropogenic radionuclide which is regularly detected at the station, at an atmospheric concentration which is within the normal range for the station. Level 4 (anomalous anthropogenic conditions): The spectrum contains evidence of one relevant anthropogenic radionuclide which is either not regularly seen at the station, or is regularly seen but is above the normal concentration range. Level 5 (multiple anthropogenic conditions): The spectrum contains evidence of more than one relevant anthropogenic radionuclide, under anomalous conditions as at level 4, with at least one being a fission product. Page 69 of 70