Overview

The SEPEM server hosts three types of statistical models:

mission and worst event integrated quantities, peak quantities (quantities are particle flux or radiation effects);
time above threshold;
event duration.

Three statistical model types (Monte Carlo, (JPL), virtual timelines and ESP) are available. For time above threshold and event duration analyses, the virtual timeline method is preset.

The event duration analysis is performed on the time history of events only (i.e. start and stop times of events). For the first two model types, a physical quantity has to be selected on which the statistical analysis will be performed: either particle flux, or a radiation effect quantity.

The web page uses tab panes to define the varies quantities and parameters needed for a model run: model parameters, response functions and associated shielding and device geometries, and spacecraft trajectories for magnetopsheric shielding. Tabs will only be visible when required for a model run. The model parameters tab is always displayed.

In order to identify runs, a unique name and an optional description need to be entered at the bottom of the respective panes. The model run is started by clicking the Run Model button (or the Store Geometry button on the geometry pane) If the name has already been used for another run, the user will be prompted to either use another name or to delete the existing run from the list on the My SEPEM page.

Caution: deleting a run stored in the database, will also delete all other runs that are dependent on it.

Once a run has been started, no other activity (except browsing the help pages) is possible on the server (with the current user account) until the run is completed. While the process is running, a page is presented where the user can perform a refresh to check for completion, or kill the running process. The user can log out and return to the server later.

After completion of a model, response function or magnetospheric shielding run, an embedded page is shown with the outputs, which depend on the various selected parameters. All output files (PNG plot files and text files) can be downloaded as a zip archive. All files are stored in the database and can be retrieved at any time from the My SEPEM page.

Model parameters pane

Depending on the selected model type and analysis quantity, a number of model parameters need to be specified.

When not all input parameters or selection options have been set, the Run Model button is disabled, and the missing input is higlighted in red font. Labels of tab panes where input is incomplete are displayed in red font. All tab labels need to be green before a model run can be started.

Event based and time series models

The first model type can be run in two ways:

event based (H+ions): by performing statistics on event based quantities (event integrated and peak values); with this option, the SEPEM reference datasets for H and He are automatically selected. For each event in the selected event list, H and He fluence and peak spectra are obtained. For the calculation of radiation effects quantities, the He spectrum is extrapolated from the last He reference energy (80 MeV/nuc) up to the last H reference energy (240 MeV/nuc). Fluxes for heavy ions are obtained by applying abundance ratios to the He fluxes.
time series (H): with this option, effects quantities are calculated for each time record in the selected proton dataset and the statistical analysis is performed on the time series of the effects quantity. With this option, only protons can be treated, as it is not possible to reliably extrapolate the He spectra for individual data records. For proton flux models, this model type allows the user to select any dataset that contains proton fluxes.

When a radiation effect quantity is selected for the analysis, the Response function and Geometry panes will appear for definition or selection of an existing response function.

For the event based model type, geomagnetic shielding can be folded into the analysis.

Event duration analysis

This model type uses the start/stop times of SEP events to calculate the probability distribution of event duration during a mission. Either the SEPEM reference event lists or a user defined event list can be selected for this type of analysis. No data selection is required.

Time above threshold analysis

This model type calculates the probability distribution of the durations of episodes during which a particle flux or radiation effect quantity exceeds a user defined threshold.

Model selection and parameters

Quantity

For all model types except event duration, a physical quantity has to be selected for the statistical analysis, either the primary environment particle flux or radiation effect quantity. For effects quantities, an ion range needs to be specified (for TNID this is currently limited H only; for shielded ion flux, a single ion needs to be selected).

For event based analyses, the datasets for the H and He environment are automatically set to the SEPEM reference datasets. For a time series or time above threshold analysis, the user needs to select a proton dataset. If a radiation quantity was selected for analysis, the Response function and Geometry tabs will only appear when a proton dataset has been selected.

Ion abundances

When a response function has been selected which includes heavy ions, an abundance table has to be selected: either a table produced during the ESHIEM project by TRAD, France, or a table of flux ratios based on the CREME96 5 minute worst case ion spectra.

Geomagnetic shielding

For event based analyses, geomagnetic shielding can be applied to the primary interplanetary environment flux to simulate the environment experienced by an Earth orbiting spacecraft. When this option is activated, the Geomagnetic shielding will appear where a shielding table can be selected or created.

A number of additional parameters need to be set when geomagnetic shielding is activated:

selection of an ionisation table: this defines the charge state of the ions in the calculation of their rigidy. Three options are available:
- CREME96 ionisation states up to Zn, fully ionised heavier ions
- CREME96 ionisation states up to Zn, half ionised heavier ions (the default)
- all ions are fully ionised
selection of Kp values, as the external magnetic field model (Tsy89) used for the cut-off maps depends on Kp. Two options are available:
- a fixed Kp value selected by the user
- calculation of Kp on an event by event basis: this method calculates an average Kp value for each event in the event list by weighing the historical Kp time series with the H proton flux in the second channel of the SEPEM proton reference dataset (7.2306–10.4564 MeV).

Event list

For all model types except time above threshold, an event list has to be selected, either one of the SEPEM reference lists or a list generated by the user. Prior to starting the calculation, the system will check how many events are covered by the intersection of the event list and the epoch range of the input data. If less than 50 events are found, a warning message is generated and the model run is not performed.

Parameter for analysis

One of two analysis parameters can be selected:

time integrated quantity: the outputs will provide the probabilities of not exceeding a stated total quantity (e.g. particle fluence or total number of SEUs) over the complete mission duration (cumulative mission quantity) and over a single SEP event (listed as the worst event quantity).
event peak value: this analysis will yield the peak quantity (e.g. particle flux or SEU rate) which will not be exceeded over a mission.

Analysis method

Three analysis methods have been implemented.

Monte Carlo is the method used for cumulative fluence analysis in the JPL model. In SEPEM this can also be used for worst event and peak value analysis.
Virtual timelines is the new SEPEM modelling methodology and accounts for the non-negligible duration of SEP events as well as allowing the inclusion of the Levy distribution which has been shown to be a better fit in most cases and certainly more robust than the two Poisson distributions available.
The analytical ESP method is another widely used modelling technique which extrapolates results from the fitted distributions assuming a Poisson distribution of events. .the use of the truncated power law for the analytical extrapolation.

Distribution

Three distributions are available to fit the SEP event quantities and peak quantities: the cut-off power law, the truncated power law and the lognormal distribution. For the ESP method, the truncated power law is prescribed (consequently, the distribution selection is not shown for this method). In the output, plots for all three distributions are always included.

Waiting time distribution

With the virtual timeline method, a selection must be made between three distributions to fit the waiting times between SEP events (or the reciprocal event frequencies): the Poisson, the time dependent Poisson or the Levy distribution.

A Poisson distribution (used in all major models with the exception of the early King model prior to SEPEM) assumes events are distributed randomly while the others allow for periods of higher and lower average event rates. As only the virtual timelines method allows the use of non-Poisson waiting time distributions, the user must select this option to compare the fits. The Levy function will provide a good fit in all cases and is strongly recommended by SEPEM when performing virtual timeline runs.

Duration distribution

With the virtual timeline method, a selection must be made between three distributions to fit the durations of SEP events: the Poisson (more precisely, the exponential distribution), the time dependent Poisson (Fourier transform in the time domain) and the Levy distribution. As with the waiting times, the Levy function will provide a good fit in all cases and is strongly recommended by SEPEM when performing virtual timeline runs.

Time period

At present, the system allows the use of three time periods: total time period, active years only and solar minimum (quiet years). Most models are applied to active years as it is uncertain when a mission may launch and when exactly solar cycles will begin/end. This is the strongly recommended option. Selecting quiet years only assumes that the event frequency is lower but that the flux distribution is independent of activity (possibly conservative). The total time period method ignores any solar cycle dependence and therefore assumes an average event frequency over the complete time series.

Mission lengths

For every model type, a set of mission lengths needs to be defined. Up to 8 mission lengths can be specified, between 0.25 and 20 years in length. Input fields that are not required should be left blank. The system will sort the specified lengths in ascending order if required.

A minimum of 1/4 year is allowed but these results should be used with caution as the duration of a single SEP event can take up a significant portion of the time series and the complete flux profile for randomly generated events is not established. For short time periods the virtual timelines method is more likely to produce reliable results as it considers the duration of events based on the generated fluence or peak flux.

Thresholds for event selection

Low fluence/peak flux events are often badly fit by the flux distributions. Therefore, thresholds are used for each data channel: events during which a threshold is not exceeded, are removed from the event list prior to the statistical analyis.

The treshold specification depends on the selected model type and quantity:

event based analysis: for primary fluxes, a set of thresholds pre-calculated for the SEPEM reference datasets is used. For a radiation effect quantity, the user has to specify a threshold value (this may require an iteration of threshold values in order to optimize the distribution function; a value of 0 is allowed as a starting point for evaluating the distribution function).
time series: for primary H fluxes and fluences, a threshold needs to be specified for each data channel of the selected H dataset. The system present default threshold values as follows:
- for any of the reference datasets, the pre-calculated values are used;
- for the other datasets, thresholds are calculated as follows: f_th = 8.54E7 * E^-3.1209 for a fluence analysis, and
  f_th = 2.70E3 * E^-3.1209
  for a peak flux analysis (where E is the median energy of the channel).
For a radiation quantity, the user has to specify a threshold value as in the case of an event based analysis. However, for a time series analysis the system provides guidelines for threshold values by listing the minimum and maximum values in the quantity time series for a peak value analysis; for an integrated value analysis, the minimum and maximum values are multiplied by 86400 s to obtain daily values. Depending on the selected H dataset, there will be a delay in refreshing the page while the minimum and maximum values are first extracted from the database.
time above threshold: guideline values are shown as for the time series analysis.
event duration analysis: this method does not use thresholds

Error propagation

For event based and time series type analyses, error propagation can be activated. The options for the treatment of error propagation are:

no propagation: No error analysis will be performed for the statisitcal model generation process.
event statistics only: The fit to the event magnitude (flux, fluence, dose, etc.) is performed so that uncertainties in the fit parameters are determined in addition to the parameters themselves.
non-event statistics only: Here the uncertainties other than those associated with the statistical variability in the event magnitude (and the fit to it) are calculated and propagated. These include uncertainties in:
- cross-calibration process between different proton and He instrument data;
- the magnetospheric shielding transmission functions;
- physical shielding calculation to produce shielded fluxes and dose;
combined errors: The uncertainties in the statistical model output includes both event statistics and non-event statistics.

If either event statistics or combined errors is enabled, the uncertainties in the event statistics will be sampled as part of the Monte Carlo analysis. The user can examine the influence of using a perturbed fitted function for the event magnitude distrubution by enabling the Factor for first/second coefficient check boxes and modifying the number of standard deviations to apply to the fit parameter (in the text boxes); "0" means no perturbation is applied to the fit parameter, "+0.5" and "-1.75" will mean the fit parameter, μ, is modified to μ+0.5σ_μ and μ-1.75σ_μ respectively.

Monte Carlo biasing

For event based and time series type analyses, Monte Carlo biasing can be switched on or off for the random generation of event sizes. Under normal circumstances (biasing switched off) the statistical models are generated based on an analogue Monte Carlo sampling of the function fitted to the event magnitudes. When switched on, the sampling will favour the lower-probability, higher event magnitudes, whether analysis is treating fluxes, fluences, doses or SEU rates. Biasing therefore improves the accuracy (lowers Monte Carlo statistical uncertainty) of the models at higher event magntitudes, but with a slight reduction in the accuracy for lower magnitudes.

Random seed initialisation

By default, the IDL code that runs the statistical models uses a randomised initial seed for its random number functions. Users can force an initial seed value by ticking the corresponding box on the model tab, and supplying an initial value (a valid integer or float value).

Outputs

Model outputs

The model output files consist of a header and a series of data blocks, one block per mission length. Comment lines are identified by a '#' character in the first column. The header format is as follows:

number of data channels;
mission lengths as a comma separated list;
number of probabilities;
label identifying the data source
for each data channel, a label and units string on consecutive lines.

Each mission length output block is preceded by a comment line identifying the mission length. The output blocks consist of records starting with the probability, followed by the modelled quantities.

Event list files

The event list files consist of a header and a series of data records, one per event. Comment lines are identified by a '#' character in the first column. The header format is as follows:

event list label;
number of data channels;
number of events;
helioradial distances (AU) as a comma separated list;
label identifying the data source;
for each data channel, a label and two units strings (for event integrated and peak values, respectively) on consecutive lines.

For each event, a record is written with the following content:

event start and stop times and event duration (days);
for each radial distance, the event integrated values for each data channel, followed by the peak values.

Magnetospheric shielding pane

In order to implement geomagnetic shielding, a series of rigidity cut-off maps was produced using the Geant4-based MAGNETOCOSMICS model. The user can define a spacecraft trajectory over which the average particle transmission can be calculated - as a function of energy - based on interpolating over these rigidity cut-off maps.

The trajectory is defined by specifying the start date and duration, and the osculating elements. A number of shortcut orbit types (e.g. geostationary) are provided to simplify the trajectory specification.

Instead of generating a transmission function, a previously created one can be selected as well.

Response function pane

The SEPEM shielding effects tools use a response function approach whereby the effects parameters are calculated once for a unit flux in the energy bins defined by a set of data channels. Applications that use a response function will fold it on the fly with the flux records of the data channels for which it was defined.

A number of Geant4 radiation effects tools have been implemented on the server to evaluate Total Ionising Dose (TID), Total Non-Ionising Dose (TNID), shielded fluxes, and Single Event Upset (SEU) rates.

For event based analyses, H, He and heavy ion fluxes can be included in the calculation of the effect quantity. Heavy ion fluxes are derived from the extrapolated He spectrum using abundance tables.

Each response function requires a geometry definition on the Shielding geometry pane. Once a geometry has been created or selected, the response function parameters can be entered.

For the calculation of SEUs, cross section functions have to be defined for protons and/or ions (depending on the selected ion range). In addition, a device geometry needs to be defined. For the other effect quantities, no parameters are required.

Caution: at present, the MULASSIS tool only treats H ions in the calculation of TNID. Hence, the ion selection for TNID is pre-set to H only.

Shielding geometry pane

The shielding geometry is defined by specifying a sphere or slab configuration, and the number of shielding layers. For each layer, the shielding material and thickness (and thickness units) have to be specified. The NIEL curve is used to calculate non-ionising dose. The material selection list is the NIST list used in Geant4.

For spherical geometries, a central void is assumed with radius at least 10% of the total radius. The shielding layers are layered on top of the central void.

Last modified on: 5 December 2018.