This page will describe the state of the Supernova analysis. A plan for getting everything working during the summer 2016 NuMI shutdown.


There are numerous different models of supernova explosions each of which predict a different energy, time and flavour dependence of the resultant neutrino flux. Currently we have spectra for two theoretical models, the Livermore and Garching models. The source of both of these is this website. A real source should be identified soon.

Flux prediction

The models discussed above provide the supernova neutrino flux for each neutrino species in terms of instantaneous luminosity, mean energy and a parameterised energy distribution. In order to be of use these must be transformed into a neutrino flux, or fluence, at NOvA. This is done using the python scripts here. Specifically the script converts the input distributions into histograms of time vs energy vs flux. The script then parses this into a series of energy vs fluence tables in the Snowglobes input format. More details can be found in the README, and in the comments inline with the code.

Interaction cross section

The above will provide a neutrino fluence per unit of time that can be input into a simulation. For reasons of speed and due to a lack of low energy neutrino cross sections in Genie the Snowglobes globes based software package is used to convert this into a particle probability distribution function that will be used in detector simulation. The current implementation of this involves taking the output of the above and running each of these through Snowglobes and outputting the positron distribution function for each unit of time. The reason for this choice is that the dominant interaction mode of these neutrinos in NOvA is via inverse beta decay (IBD). Sub-leading modes could also be highly important and should be included as soon as possible. A script for doing this is provided as and is again described in the README, and in the comments inline with the code.

There are a number of caveats with this method. Including:

  • The lack of sub dominant interaction modes.
  • The model of NOvA in Snowglobes - currently we're using a 14kt scintillator detector.
  • The output IBD (and all other modes) energy distributions are in terms of reconstructed neutrino energy, not true neutrino interaction product energy. For the IBD decay case this is a fairly good approximations (e+ energy ~ neutrino energy - 1.3 MeV), but this won't be the case for the other interaction types.

Detector simulation

There are currently two supernova simulations that are being actively developed:

The resultant signal can be overlaid with minimum-bias data using the SNMixer module in the SupernovaUtilities package.


The challenge associated with identifying a supernova will be spotting it in the first place. There are two ways to do this at NOvA: using the data driven trigger and plugging into the supernova early warning system.


SuperNova trigger system is currently being developed.


The supernova early warning system. This will broadcast an indication of a supernova if detected by a coincidence of neutrino detectors around the world. The delay on this being issued is expected to be only around 10s, so should be within our buffer window. A trigger will be developed with records NOvA data based on this alert.

In a similar vein is Egads (p147, s4.3.3) which hopes to publish alerts within 1s! (Not quite - EGADS is a plan to get one large detector loaded to the gills with Gd for increased SN-nu sensitivity. Any one experiment can publish alerts within a second subject to the design of its DAQ, but then it loses the noise rejection feature of a coincidence network)


Now we've got the signal what can we do with it. Pointing information would be really useful, especially if it can be achieved quickly and to with a resolution of a few degrees. A recent summary paper on observing supernovae can be found here.

Useful talks

  • Krissie Nosbisch docbd-7767
  • Matthew Tamsett docbd-10109


The list of outstanding tasks:

  • Get better models.
  • Cross check the flux prediction code.
  • Look into adding low energy neutrino cross sections into Genie.
  • Revisit the NOvA detector model in Snowglobes.
  • Adjust the Snowglobes IBD efficiency and detector response matrices so they are appropriate.
  • Include subleading interaction modes.
  • Implement appropriate angular distribution of positions in SupernovaGen - can likely be done via fcl.
  • Expand SupernovaGen to include other interactions (when we have these rates).
  • Refactor SupernovaGen so that the positron PDFs are in a separate file.
  • Investigate cosmic overlays to avoid duplication and underestimation of Cosmic overlays.
  • Test the existing DDT with simulations.
  • Further develop the DDT. Note: In order to interface with SNEWs we need to have a fake trigger rate of < 1 per month! Can we get any directional information?
  • Offline analysis.