The deep Antarctic ice is the most important detector component as it is the detection target and light propagation medium in IceCube. Unlike the DOM hardware, which has been calibrated and studied in the laboratory, the glacial ice can only be measured in-situ and is therefore more difficult to describe. The optical properties arise from the dust concentration and the amount of air bubbles in the ice as well as the crystalline and molecular structure of ice. The first data was obtained using ice cores of the deep regions below a depth of 2000 m from drill sites approximately 1000 km away from the South Pole [208]. The deep South Pole ice was first measured with the LED calibration system of AMANDA [209]. During the construction of IceCube, direct measurements of the dust concentration in the glacial ice were obtained by a dust logger that had been deployed into some drill holes [210]. In addition to the dust-logger data, the flasher LED calibration system of the IceCube detector has been used for fitting a detailed South Pole Ice Model (SPICE). It is the default ice model of IceCube, for which multiple iterations (SPICE1, SPICE2, SPICEMie, SPICELea) have been developed so far [211,212]. The optical properties described by each ice model are the scattering and absorption of light in ice. They are quantified by the scattering length ls and absorption length la, respectively, as well as an average scattering angle. The scattering length is the average propagation distance after which a photon is scattered and the absorption length after which the survival probability of a photon decreases to 1/e. In practice, only an effective scattering length leff
s = ls/(1− ⟨cos θ⟩) is experimentally accessible, for which the average scattering angle ⟨cos θ⟩ is taken into account. The effective and geometric scattering lengths are equal (leff
s = ls) if the light is scattered isotropically. Measurements with the in-situ LED calibration system of IceCube have yielded lseff ≃ 6 m . . . 90 m and la ≃ 20 m . . . 300 m, depending on depth [211]. Consequently, the deep ice is so clear that photons may travel for hundreds of meters before being absorbed. However, the much shorter scattering lengths cause the light to diffuse relatively quickly, with the result that a point-like light source looks nearly isotropic at a distance of ∼ 100 m.
correlation between Cdustand the contribution from dust to
be(400), were used to derive the dust profile for absorption
(Figure 21, right) from the dust profile for scattering (Figure 21, left). The third parameter in our model,a, is used to calculate scattering at any wavelength from be(400)
via a power law:
beðl nm½ "Þ ¼ l=400ð Þ&abeð400Þ: ð25Þ
The remaining three parameters (k, AIR, andl0) are used to
calculate absorptivity from adust(400) through the two-
component model:
aðl nm½ "Þ ¼ l=400ð Þ&kadustð400Þ þ AIRe&l0=l: ð26Þ
Maps of effective scattering coefficient and absorptivity, generated from our model and summarizing our knowledge of optical properties of South Pole ice, are shown in Figure 22 for depths between 1100 and 2300 m.
[79] Our measurements of depth dependences of the
optical properties had a resolution of on the order of ten meters, and our methods probed up to two hundred meters of ice between emitter and receiver. The techniques used in this work could not resolve individual dust layers much thinner than ten meters, such as highly absorbing layers of ash deposited by volcanic eruptions. Such thin ash layers may affect the performance of AMANDA and IceCube as neutrino telescopes. Building on the remote sensing techni- ques presented here, a dust logger [Miocˇinovic´ et al., 2001; Bay et al., 2001] was developed and used in both Antarctic
and Greenland boreholes, where it was able to resolve centimeter-thick layers of volcanic ash. Analysis of data from a dust logger operated in the first hot-water-drilled IceCube hole confirmed that ash layers are also present in South Pole ice and can be detected with the logger tech- nique [Bramall et al., 2005]. However, the South Pole ash layers are weaker and less numerous than those detected at Siple Dome (West Antarctica) [Bay et al., 2004], which is partly explained by the higher altitude of the South Pole and greater distance from Antarctic volcanoes. Highly absorbing ash layers will affect light propagation, mainly by localized depletion of photons traveling at an acute angle relative to a layer, which modifies the angular dependence of the photon yield. Scattering in thin ash layers should be similar to scattering by dust and the effect on timing should be small. Furthermore, unambiguous identification of ash layers in the depth profiles at boreholes up to one kilometer apart in the IceCube array would make it possible to measure deviations of optical properties from the horizontal. In the present analysis, we assumed that the dust structure is horizontal over the length scale probed and within the sensitivity of the measurements. However, isochronal maps made with deeply penetrating radar at the South Pole [Blankenship and the Instrument Definition Team for a Europa Radar Sounder, 2001] show that dust layers can tilt by up to 50 m over a square kilometer. Given the strong fluctuations in optical properties over such a depth scale, tilting dust layers would strongly affect IceCube perfor- mance and must be fully mapped. This could be achieved by using dust loggers in several widely spaced boreholes along the perimeter of the array and matching up features in Figure 22. Maps of optical scattering and absorption for deep South Pole ice. The depth dependence between 1100 and 2300 m and the wavelength dependence between 300 and 600 nm (left) for the effective scattering coefficient and (right) for absorptivity are shown as shaded surfaces, with the bubble contribution to scattering and the pure ice contribution to absorption superimposed as (partially obscured) steeply sloping surfaces. The dashed lines at 2300 m show the wavelength dependences: a power law due to dust for scattering and a sum of two components (a power law due to dust and an exponential due to ice) for absorption. The dashed line for scattering at 1100 m shows how scattering on bubbles is independent of wavelength. The slope in the solid line for absorptivity at 600 nm is caused by the temperature dependence of intrinsic ice absorption.
D13203 ACKERMANN ET AL.: OPTICAL PROPERTIES OF SOUTH POLE ICE
24 of 26
D13203
Figure 3.8: Effective scattering coefficient (left) and absorption coefficient (right) of light in South Pole ice as a function of depth and wavelength. Scattering in the shallow ice is large due to bubble columns. In contrast, the deep ice is very clear and absorption is minimal around a wavelength of 400 nm. The region of high absorption and scattering around a depth of 2000 m is called dust layer and likely due to volcanic ash. Figure reproduced from [209].
Both scattering and absorption depend on the wavelength of the light and on the depth of the ice as shown in Figure3.8. Here, the measured and modeled effective scattering and absorptions coefficients are shown. They are defined as the reciprocal values of the effective scattering and absorption length, respectively. These measurements were obtained by the in-situ LED calibra- tion system of AMANDA [209]. Down to a depth of ∼ 1300 m, light scattering is dominated by residual air bubbles. At larger depths, the increasing pressure leads to a compression of the air bubbles, and dust becomes the dominant source of scattering. The dust concentration changes with depth due to variations in climate and volcanic activity as snow accumulates on the South Pole glacier over time [213]. Near a depth of ∼ 2000 m, scattering and absorption is signifi- cantly increased due to a high dust concentration, potentially from a major volcanic eruption in the past. This region is called dust layer. Light absorption is caused by dust particles in the ice and therefore follows the same depth dependence as scattering for the ice below 1400 m. Light scattering decreases with its wavelength. It is larger for ultraviolet light and smaller for infrared light. Absorption of light is also dependent on its wavelength. The intrinsic properties of ice cause a strong attenuation of infrared light. The absorption is minimal for light with a wavelength around 400 nm and, again, larger for light with a shorter wavelength.
The SPICE1 ice model was fitted using full brightness LED flasher data and describes the effective scattering and absorption coefficients at a wavelength of 400 nm for ice layers between a depth of 1098 m to 2798 m in bins of 10 m. The width of the ice layers is limited by the vertical distance between two DOMs. Unlike the newer generation ice models, SPICE1 was completely symmetric and isotropic within one ice layer. In SPICEMie, the ice model fit was extended to include numerical calculations based on Mie theory [214]. Also, evidence arose that the ice layers are not exactly horizontal but rather tilted [211]. This ice tilt most likely related to an uneven surface of the rock at the bottom of the glacier on which the ice layers accumulated over time. The effect was parametrized symmetrically along an axis that is approximately perpendicular to the glacial flow and also dependent on depth. With the extension of an ice tilt, the scattering and absorption coefficients have effectively become dependent on the full three-dimensional position in the ice instead of just the depth. In the latest ice model, SPICELea, the isotropy assumption has been lifted and an ice anisotropy of the scattering coefficient depending on the direction of the photon propagation has been introduced [212]. The modulation of the nominal scattering coefficient was fitted to −8% along the horizontal direction of the glacial flow, +4% along the horizontal direction of the ice tilt and +4% along the vertical direction towards the surface of the ice. There has been no evidence that the modulation is also dependent on the position in the glacier. The ice anisotropy is an asymmetry of the model which is important for tau-neutrino reconstruction in IceCube. This will be discussed in greater detail in Chapter5.
3.3 (Re)construction of an IceCube Event
A highly energetic particle interaction in the IceCube detector can produce enough Cherenkov light to trigger hundreds of DOMs. Each DOM records a digitized waveform that carries information of both the arrival time and the amount of detected light. Together with the position of the DOM, the four-dimensional distribution of Cherenkov light in the detector is determined. The entirety of all triggered DOMs make up an event in IceCube. It can be visualized to display different hit patterns which are classified into various event topologies. Event properties such as the direction, energy, or interaction type can be derived by reconstruction algorithms and used to define event selections. In order to study the performance of reconstruction algorithms, hundreds of thousands of events are simulated in IceCube by Monte Carlo methods, and the true simulation parameters are compared to the reconstructed results.