The IC86-2017 physics run: better neutrino alerts and a brand-new monitoring system

It’s that time of the year. Down at the South Pole, our team is in the darkness of the austral winter, enjoying beautiful auroras while monitoring IceCube data taking. Up north, the team has completed all updates and checks to the new data systems running live in the IceCube Lab (ICL), sitting on top of the IceCube detector on Antarctica’s surface.

Each year, just after the summer polar season ends, the operations team shifts its focus to preparations for the new run. During the last three months, every update to IceCube’s data-taking and monitoring systems has been checked and tested numerous times. And on May 18, 2017, everything was in place to start run IC86-2017, the seventh year of operations with the completed detector.

“We keep the detector data-taking configuration stable for a full year, which is ideal for IceCube science analyses. So the physics run start is our best opportunity to roll out significant improvements that we’ve been working on throughout the year,” says John Kelley, who is the IceCube detector operations manager. “It’s a major milestone for our team.”

The 2017 run brings great upgrades to the IceCube follow-up alert systems, including a new real-time starting-track neutrino selection, called ESTReS.  “The ESTReS online selection looks for starting tracks below the energy of HESE,” explains Kyle Jero, a PhD student at UW–Madison who developed the new real-time selection. HESE was the selection used in the study that led to the IceCube discovery of astrophysical neutrinos, which selected events with at least 60 TeV of energy. “We now can find astrophysical neutrinos down to 10 TeV, and we expect to find up to three more astrophysical neutrinos a year in real time.”

The array of IceCube sensors as displayed in the monitoring system. The left image show a “good” run, when most sensors are working well (represented in blue). During a “bad” run, as displayed in the right image, many sensors are not performing as expected. This right image corresponds to an incident during the IC86-2016 run that caused a loss of 1.5 hours of data­ (further details provided in text).

Also deployed this year is a new GCD (geometry, calibration, detector) database and generation system that is flexible and easy to maintain. GCD files are the basis of every analysis in IceCube since they calibrate the response of each sensor in the detector to measure with precision the light produced by the interaction of high-energy particles with the ice.

 Another upgrade to highlight is the new run monitoring system, developed by a team at UW–Madison, with contributions from colleagues in Maryland, Alabama, and New York, in the US, as well as in Mainz, Germany, and Uppsala, Sweden. “This new system allows gathering, testing, and displaying more information than previously, while improving on user experience and maintainability,” explains Michael Frère, one of the software developers that led the design of the new system. “It’s been a long road to get here, and there are lot of people to thank for making this happen,” adds Colin Burreson, also a software developer who worked on the user experience and usability.

The IC86-2016 run in brief

After major improvements during the first years of data taking with the full detector, IceCube is now steadily working at full potential for approximately 98% of the time.

The previous run, IC86-2016, ended very successfully. It was another full year of data without major failures in the hardware. The last IceCube sensor to be lost was in May 2013, after a power outage in the Dark Sector, an area near the NSF-managed Amundsen-Scott South Pole Station that is kept clear of electromagnetic signals that could hamper radio telescopes. This February, a plane accidentally created radio-frequency interference that produced a 50% increase in the signal rate in IceCube, and a loss of 1.5 hours of data but no further damage to the data-taking systems.

This past year presented challenges in terms of power supply failures and power outages, but IceCube still ran smoothly, with uptimes consistent with last season’s performance. The detector uptime remained at an outstanding 99.75%, with a full-detector “clean uptime,” in which all strings are taking data, of 97.65%.

It has been a tradition that every winterover team tries to break new records, and in recent years they have succeeded. Christian Krueger and Mack van Rossem, the IC86-2016 winterovers, achieved unprecedented >99% clean uptimes for four months in a row and a smashing record of 99.54% full-detector clean uptime for the month of October. “Many subsystem experts around the world contribute to the stability and operation of IceCube, but at the end of day, our winterovers are the glue that holds everything together,” explains Matt Kauer, the IceCube run coordinator.