After two years in the temporary IceCube Counting House, the IceCube Team moved cables and computers into the new IceCube lab (ICL). This move required that the surface cables from the nine strings from the past two seasons be disconnected and then relocated through new trenches to a tower of the ICL. Similarly, the surface cables from the 13 new strings were drawn into the ICL. The ICL (formerly used as a dormitory) sits at the center of the whole IceCube array. In addition, 34 DOM Hub computers that communicate with the DOMs and 44 server class computers for DAQ, processing and filtering, data handling and network services were installed. The servers each have 2 dual core Opteron 64 bit CPUs and with the DOM Hub computers have 420GB of RAM. The 184 hard drives have a raw storage capacity of about 20 TB. Altogether, the system requires about 25 kW of power and is backed up by enough UPSs to keep all systems going for at least 30 minutes if station power is lost. This completes the computer hardware required for all of the IceCube strings.
The IceCube team deployed thirteen strings in the 2006-07 season–exceeding the 12-string goal. Without the weather related one-week delay at the beginning of the season, a 14 string season was clearly possible. This year the time to deep drill each hole averaged 41 hours, and more importantly, the average time between holes was reduced to 75 hours. In the end, the drillers could drill a hole less than every 3.5 days. An independent firn drill was new this season and drilled firn for holes 12, 13, and 14. Hole 14 will be first hole for next season. The use of this independent firn drill should reduce the time between holes by a half-day. Deploying time for each string averaged 10 hours. There are 1,424 deployed DOMs (InIce and IceTop). Only one DOM does not communicate from this season. Of the total, 1,417 DOMs or 99.5% are working.At the end of the season, the drill camp was placed in a ready location for next season, instead of being towed back to the winter storage site, saving about one week next year. Ready for next season are about 500 DOMs at the South Pole and McMurdo.
In November 2006 the project adopted a plan to streamline the DAQ software. Principally the new DAQ merged the hub and string processor and changed the distributed control framework. An XML based deployment simplified the DAQ configuration and the server architecture system auto discovers and configures itself based on minimal input and few rules. Coinciding with the computer and server upgrades in the ICL, the new DAQ software integration began in late January 2007. By the close of the South Pole season on February 21, 2007, the 22 in-ice strings, 26 IceTop stations and the AMANDA detector were integrated into the DAQ. The new software has fewer lines of code and requires fewer computing hosts. The configuration server system is simple and flexible allowing uniform deployment across several platforms. Each developer can bring up real instance of DAQ (simulated DOMs) on a system as simple as single laptop for development work. Turn-around time is very fast: 30 seconds for complete deploy and start-up to running state compared to the previous system which took 10 minutes. This enables quick edit-compile-debug looping.
Until one year ago the priority for simulation was the implementation of all of the ingredients for the generation of physics events, their propagation through the Earth and ice, the simulation of Cherenkov photon propagation in the ice, the photomultiplier tubes response to photons, the DOM functionalities, and the data acquisition trigger decisions. The comparison of simulated physics events with the real data taken during 2006 with the first nine strings provided precious information for the refinement of simulation software. Then the software underwent a major structural change last year requiring adaptation to a new data structure. This transition led to the implementation of an efficient and well scoped series of unit software tests to ensure stability and reliability. Consequently, simulation quality improved substantially in terms of how the physics is described and how the detector functions. We are now able to simulate coincident events between IceTop surface arrays and IceCube in-ice strings. We can simulate events with IceCube and AMANDA responses and merge them in a single event, as the real detector does. The quality of simulation continues to improve.
Using a second drill tower and and improved firn drill, the IceCube drill team was able to drill eight holes this year. By coordinating their efforts they increased their rate from one hole every six days at the beginning of the season to one hole every three and one-half days at the end of the season. If drilling can begin by December 1, then it will be possible to drill an average of 14 holes each season. Contributing to the efficiency of drilling was the increase of drill speed from 1.3 m/min at the beginning of the season to 2.2 m/min at the end. The second season also led to better performance in drill camp set-up and shutdown. Plans for next season include three nine-hour driller shifts instead of two twelve-hour shifts and dedicated heavy equipment for IceCube operations. Other improvements included a new hose, improved heater reliability, and elimination of fuel system air problems.
The yield of fully functional DOM from this year's deployment of 480 in-ice and 48 IceTop DOMs was excellent. Approximately 99% of the DOMs that were deployed continue to operate as of April 2006. Four DOMs failed and three of the failures appear to be related to in-ice connector, cable or penetrator failures and not a failure of the DOM internal electronics. One in-ice DOM suffered a high voltage failure soon after turn-on. Cold-temperature communications problems with a number of DOMs were encountered this season and changes were made in the software and there is no longer an effect on data taking. Data can now be taken continuously with all DOMs using the testing and commissioning data acquisition system. Locations where DOMs have taken a long time to freeze in are correlated with higher drill dwell time at that depth. Within hours of the turn-on of the 9th string, it was possible to obtain clear multi-string events (many of the DOMs were in water at the time). Locations where DOMs have taken a long time to freeze in are correlated with higher drill dwell times at those depths.
The South Pole Test System (SPTS), a mirror site for the South Pole System, operates in Chamberlin Hall, UW-Madison Physics Department. The SPTS is the final test bed for the data acquisition and data handling software and is used to test patches and upgrades to the software operating on the South Pole System (SPS). An additional test facility is located at the UW-Madison Physical Sciences Laboratory that includes a fulllength surface to DOM cable and DOMs in portable freezers to simulate actual DOM operating conditions. This additional test facility will eventually be moved to Chamberlin Hall. All the electronic and computer equipment required for the 2006-07 season is received and assembly of the actual system will be completed in Madison to confirm performance prior to disassembly and shipment to the South Pole. Ordering has been completed for the cluster that is being built to support simulation analysis. The cluster is currently scheduled to be online sometime in June 2006.
Data continues to flow from the detector, with about 25GB/day of raw data being archived to tape at pole and 5GB/day being transmitted over satellite to the data warehouse at UW. The event count for the IceCube 9 string array, with 595 DOMs currently reading out into the data stream, surpassed 250 million events on 4/4/06 and is upwards of 360 million events two weeks later. Detector uptime peaked at 99% on 4/4/06 and holds at approximately 80% averaged over week intervals–a 122% increase over the same period in mid-March. This increase in uptime is due to improvements to both DOMHub hardware components and DAQ software. In the Northern Hemisphere, the DAQ application level software (not resident in the DOM itself) will migrate to 64-bit computing platforms in order to increase computing density and decrease power consumption. Preliminary testing has run smoothly and there are predictions for an easy transition. (Above as of April 4, 2006)
IceCube staff shipped the hose reel and some of the components of the EHWD from Wisconsin to the South Pole in the 2003-2004 Austral season (∼Nov. 1-Feb.15). The hose reel had to be disassembled to fit into the cargo bay of a LC-130 aircraft for the final leg of the journey from McMurdo to the Pole. A team of seven reassembled the hose reel in January 2004. The following Austral summer the rest of the EHWD, 900,000 lbs., arrived in 30 separate flights. In addition to the hose and heating plant, a drilling/deployment tower was shipped and assembled. Twenty-four IceCube and Raytheon staff members spent two months putting the system together and field testing the heating plant and control system. The working temperature at the pole averages -35°F during the summer. Drilling for the first hole began on January 15, 2005.
Drilling on the first hole stalled at about 1200 meters. The team determined that damage to the drill head prevented further progress so they relocated the tower and started a second hole. After 38 hours they reached a depth of 2450 meters. Reaming the hole on the way up took another 19 hours before they turned the tower over to the deployment team. Within 20 hours the first string of 60 Digital Optical Modules (DOMs) was deployed. In addition, 16 DOMs were installed in eight IceTop tanks. These DOMs will be part of an array that will identify IceCube events that are coincident with cosmic ray air showers and examine cosmic rays with energies up to 1018 eV. Only three weeks of the usual 12-week Austral summer were available for drilling and deployment during this first season. Early analysis indicates that all DOMs are working and that the DOM clocks are calibrated to within 2 ns.
Ten strings are planned for the 2005-2006 season and over 800 DOMs are under construction in Wisconsin, Germany, and Sweden. One hundred and eighty DOMs are stored at the South Pole. Over 1800 components comprise each DOM, including 10 inch Hamamatsu photomultiplier tubes. The DOMs undergo extensive final acceptance testing (FAT) at their respective assembly sites. This testing includes three weeks in a dark freezer to establish the lowest noise levels for each DOM. In the 2005-2006 season the average IceCube population at the South Pole will be 45 people. Each will have completed medical and dental exams to be physically qualified (PQ'd) to work at the 10,000 ft. equivalent altitude of the South Pole. While installation of instrumentation proceeds at the South Pole, scientists and engineers in the northern hemisphere are developing the software to analyze the 54.7 terra bytes of data that will be produced annually once the telescope is finished. This data is stored on magnetic tape at the South Pole and shipped to the University of Wisconsin for archiving and retrieval by IceCube scientists. Researchers are developing tools that will enable scientists around the world to access and analyze the data for evidence of gamma-ray bursts, WIMPS, and dark matter.
The first new optical sensors of the IceCube neutrino observatory, 60 on one string and 16 in four IceTop stations, were deployed during the austral summer of 2004-05. Analysis of the first few months of data collected by this configuration is underway demonstrating that hit times can be determined across the whole array to a precision of a few nanoseconds. Coincident IceTop and deep-ice events are recorded and the capability to reconstruct muons with a single string has been verified. Muon events are compared to a simulation. The performance of the sensors meets or exceeds the design requirements. Fig. 1 shows an event involving all 76 DOMs. The circle size is proportional to the signal amplitude, while the color (from blue to red) indicates relative times of the hits recorded in the DOMs. All hits are consistent with an air shower on the surface coincident with a deepice muon, traveling down at a zenith angle of 3 +/- 2°. From the ice scattering-length profile shown next to the detector string, one sees that most of the detector is located in very clear ice. In fact, the lower 25 DOMs are in ice that is up to 2 times clearer than that available to the AMANDA [2] sensors located at depths of 1500-2000 meters. (from the introduction of Dima Chirkin's paper for the 29th International Cosmic Ray Conference in Pune, India, August 2005)