Pith. sign in

REVIEW 3 major objections 4 minor 12 references

Calibration LEDs in the IceCube Upgrade D-Egg Modules

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A Monte Carlo study shows that D-Egg downward LEDs, using the forward-to-backward photoelectron ratio in a likelihood fit, can recover the hole-ice bubble column's effective scattering length to within ±20% and diameter to within ±5 cm.

desk verdict Honest, clearly scoped MC sensitivity study for D-Egg hole-ice calibration; the precision claims are bin-limited closure results, not real measurement uncertainties. read the letter →

arxiv 1908.10780 v2 pith:JVR6UCUE submitted 2019-08-28 astro-ph.HE

Aya Ishihara , Ayumi Kiriki (for the IceCube Collaboration) This is my paper · ORCID
classification astro-ph.HE
keywords IceCubeUpgradeD-EggopticalmoduleholecalibrationbubblecolumnLEDflashereffectivescatteringlengthneutrinoreconstructionMonteCarlosensitivitystudy
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper seeks to establish that the downward-facing calibration LEDs on the D-Egg optical modules for the IceCube Upgrade can measure the optical properties of the refrozen drill-hole ice from the recorded light alone. Using a Monte Carlo simulation of photon propagation, the authors build a binned Poisson likelihood from the ratio of photoelectrons seen by the forward and backward photomultipliers, and show that the best-fit point lands in the Monte Carlo truth bin for every tested combination of bubble-column size, scattering, and module position. The projected precision is $\pm 20\%$ on the effective scattering length in the bubble column and $\pm 5$ cm on its diameter. A sympathetic reader should care because hole ice is currently a sizable systematic for low-energy neutrino reconstruction, and the same fitted parameters could be applied to the full decade of IceCube data.

What carries the argument

The load-bearing observable is the forward-to-backward photoelectron ratio: for each of the four downward-pointing LEDs, the upward-facing PMT in the lower module sees mostly direct light, while the upward-facing PMT in the upper module sees light that has been Mie-scattered inside the bubble column, so the ratio depends on both $D$ and $\lambda_e$ but divides out the LED intensity. The paper forms binned ratio histograms from the four LEDs and fits them with a binned Poisson likelihood whose parameters are $\lambda_e$, $D$, $r$, and $\varphi$, treating $r$ and $\varphi$ as nuisance parameters. The scattering is parameterized by the effective scattering length $\lambda_e = \lambda_s/(1-\langle\cos\theta\rangle)$, with $\langle\cos\theta\rangle = 0.95$ in the Henyey-Greenstein function.

What would settle it

After deployment, compare likelihood-fit values of $\lambda_e$ and $D$ from real D-Egg LED flashes against an independent optical probe of the same hole ice—for example, photographic images of the bubble column or scattering measurements from neighboring strings; a systematic disagreement larger than the quoted $\pm 20\%$ in scattering length or $\pm 5$ cm in diameter would show the simulation geometry does not transfer to the ice.

Watch

Extended reading notes

Core claim

The central claim is that the ratio of photoelectrons at two vertically separated PMTs encodes the two hole-ice parameters—bubble-column diameter $D$ and effective scattering length $\lambda_e$—while cancelling the unknown absolute LED intensity. The paper supports this by simulating two perfectly aligned D-Egg modules in a 60 cm borehole, with Geant4 photon propagation and Henyey-Greenstein scattering, and by fitting the simulated photoelectron-ratio histograms with a binned Poisson likelihood over $\lambda_e$, $D$, the module offset $r$, and the rotation $\varphi$. The fit minimum coincides with the Monte Carlo truth for all tested cases, and the contour size is limited by the chosen binning rather than by statistical fluctuations. From this the paper concludes that in-situ D-Egg measurements can determine $\lambda_e$ to within $\pm 20\%$ and $D$ to within $\pm 5$ cm.

Load-bearing premise

The entire sensitivity estimate rests on the assumption that each drill hole can be modeled as a centered, uniform bubble column of adjustable diameter inside a 60 cm borehole, with the outer clear ice scattering exactly like bulk ice and no cables or module misalignment blocking the light.

Editorial extensions

If this is right

  • If the quoted precision transfers to the deployed detector, the hole-ice systematic in low-energy neutrino reconstruction would be substantially reduced.
  • The same fitted hole-ice parameters can be applied retroactively to the existing decade of IceCube data, since the Upgrade holes sit in the same ice.
  • IceCube-Gen2 can reuse the identical D-Egg LED calibration hardware and analysis without new design work.
  • With modules spaced 2.7 m apart along the string, the measurement yields a far denser map of bubble-column properties with depth than the current 17 m-spaced DOMs allow.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper leaves the precision bound set by its grid binning; a finer grid in the same likelihood setup is a natural extension and could yield tighter constraints than the quoted $\pm 20\%$ and $\pm 5$ cm.
  • The same ratio-observable idea could be applied to any two photosensors that view a common pulsed light source through a scattering medium, wherever the main calibration nuisance is the absolute source brightness.
  • A testable follow-up, not claimed in the paper, is to overlay the fitted bubble-column extent on camera images of the same hole ice once real D-Egg data exist; a mismatch would indicate the centered-uniform-column model needs refinement.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This ICRC 2019 proceedings paper from the IceCube Collaboration presents a Monte Carlo sensitivity study for measuring hole-ice properties with the D-Egg optical modules planned for the IceCube Upgrade. The proposed method uses four downward-facing calibration LEDs on an upper D-Egg and records the photoelectron yields at the upward-facing PMT of the lower D-Egg (forward PMT) and the upward-facing PMT of the upper D-Egg (backward PMT). The ratio of these yields is used as an observable to suppress the absolute LED-intensity uncertainty. Photon propagation is simulated with Geant4 over a grid of bubble-column effective scattering lengths lambda_e, bubble-column diameters D, and module offsets (r, phi). A binned Poisson likelihood over ratio histograms is constructed, and the authors report that likelihood fits recover the Monte Carlo truth and conclude that lambda_e can be determined to within +/-20% and D to within +/-5 cm. The paper explicitly states that the results are limited by the chosen binning and that systematic effects from bulk-ice scattering, D-Egg misalignment, and cabling obstruction are not included.

Significance. The proposed calibration concept is timely and relevant for the IceCube Upgrade, where hole-ice uncertainty is a known systematic for low-energy neutrino reconstruction. The use of the forward-to-backward PMT ratio is a sensible way to reduce LED-intensity calibration uncertainty, and the simulation incorporates realistic PMT quantum efficiency, collection efficiency, and charge resolution. The paper is honest in labeling the study as a Monte Carlo sensitivity study and in listing omitted systematics. The main value is as a baseline for future in-situ measurements. Its principal weakness is that the quoted precision is a bin-limited closure result rather than a calibrated measurement uncertainty, and the likelihood used for the ratio observable is not derived from a well-defined generative model.

major comments (3)
  1. [Section 4, Figure 7] The central quantitative claim that lambda_e can be determined to within +/-20% and D to within +/-5 cm is not supported by a proper uncertainty quantification. The likelihood scan uses pseudo-data generated from the same Geant4 model and the same binning used to build the template histograms, so recovering the MC truth is expected closure behavior. The paper states that "the size of the 3sigma region is smaller than the current bin size," which means the quoted precision is essentially the grid spacing of the scan, not a calibrated confidence interval. To make the stated precision claim valid, the authors should either perform an ensemble of pseudo-experiments and demonstrate coverage of the quoted intervals, or explicitly rephrase the claim as a bin-resolution-limited sensitivity estimate.
  2. [Section 4, Eq. (4.1)] The binned Poisson likelihood in Eq. (4.1) is applied to histograms of the ratio of NPE at the forward and backward PMTs, but the ratio of two Poisson-distributed counts is not itself Poisson distributed, and the manuscript does not define what "number of events" (n_i) means for a ratio histogram. Without a generative model for how individual entries in the ratio histogram are produced and counted, the Poisson likelihood is only a heuristic. The authors should either define the ratio-bin count model explicitly or use a likelihood based on the actual sampling distribution of the ratio, e.g., through a Monte Carlo-calibrated likelihood or an unbinned likelihood.
  3. [Section 4, final paragraph; Section 5] The abstract and summary present the +/-20% and +/-5 cm numbers as measurement capabilities without the caveats that the paper itself lists in the final paragraph of Section 4: bulk-ice scattering, D-Egg alignment uncertainty, and cabling obstruction are not included. Since these effects are directly relevant to in-situ deployment, the precision claims should be qualified throughout the paper as idealized, simulation-only sensitivities. Otherwise the summary overstates what the study actually demonstrates.
minor comments (4)
  1. [Section 3.1] The text says the bulk ice "does not scatter photons," while Section 1 states that the bulk ice has an approximately 20 m scattering length. The simulation approximation should be stated explicitly as neglecting bulk-ice scattering for the sensitivity study, not as a property of real ice.
  2. [Section 4, Eq. (4.1)] The notation in Eq. (4.1) is under-specified: the LED index l, the bin index i, the meaning of Nbin = 10, and the construction of the ratio histograms from NPE values should be defined in the text. The bold Greek symbols theta and nu also render awkwardly and should be replaced with standard vector notation.
  3. [Figure 7] The color scale label "log(L/L)" appears to be missing a subscript or normalization; it should be something like "log(L/L_max)" or "-2 Delta log L," and the caption should state how the white star marks the best-fit bin and how the 3sigma contour is defined.
  4. [Section 2] The sentence stating that these results "plan to be applied to the entire IceCube data collected over 10 years" needs a brief justification or a qualifier, since D-Eggs are only being deployed in the Upgrade and transferring their calibration to legacy IceCube data will require additional assumptions about hole-ice properties over time and depth.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: this is an explicitly labeled Monte Carlo sensitivity study whose parameter-recovery check is a closure test, not a disguised external prediction.

full rationale

The paper's central quantitative claim is a likelihood recovery of Monte Carlo truth in a Geant4 simulation (Section 4, Figure 7). Both the pseudo-data and the likelihood templates are generated by the same simulation; the abstract says 'the minimization recovers best fit values close to the Monte Carlo truth' and Section 5 explicitly calls the work 'a Monte Carlo-based sensitivity study.' This is a closure test rather than a validation against nature, but it is not a circular derivation: the fitted quantities (λe, D, r, φ) are not defined in terms of the recovered likelihood, and no parameter is silently fit to a subset and then reported as an independent prediction. The paper candidly limits the result: 'These results are limited by the chosen binning scheme' and 'Currently, these results do not include systematic errors due to bulk ice scattering, potential uncertainty in the D-Egg alignment, and obstruction of the signal due to cabling.' The ±20% and ±5 cm statements are phrased as potential precision from the simulation, not as measured in-situ values. The cited PMT quantum/collection efficiencies are laboratory measurements, and the Henyey-Greenstein scattering is a stated modeling assumption, not a self-citation used to force the conclusion. No uniqueness claim or load-bearing self-citation appears. The statistical concern that the 3σ region has no coverage calibration and that the ratio-Poisson likelihood is approximate is a correctness/robustness risk, not a circularity of the kind where a prediction reduces to its input by construction.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on a simplified optical model of hole ice, a chosen scattering law, and several statistical assumptions. The simulation parameters are not fitted to real data, but the quoted precision is tied to the chosen grid. No new physical entities are introduced.

free parameters (4)
  • Henyey-Greenstein mean cosine = 0.95
    Chosen value for Mie scattering in the bubble column; the paper does not test the sensitivity of the fit to this value.
  • LED intensity spread = 30% (1 sigma Gaussian)
    Assumed in Section 4; the ratio observable reduces its impact, but the value is an input to the simulation.
  • PMT charge resolution = 20% (1 sigma Gaussian)
    Assumed from laboratory measurement [10] and used in the likelihood; the exact lab setup and systematics are not described.
  • Likelihood grid and binning = Nbin=10; D range 20 to 60 cm; lambda_e range 10^1 to 10^2 cm in Figure 7
    The claimed precisions of +/-20% and +/-5 cm are effectively the grid/bin widths; finer sampling would change the quoted numbers.
assumptions (5)
  • domain assumption Bulk ice is non-scattering in the simulation.
    Section 3.1 states that the bulk ice volume is generated without photon scattering; real bulk ice has a finite scattering length of about 20 m, and the clear outer region of the hole is assumed identical to bulk ice.
  • domain assumption Drill hole geometry is a 60 cm cylinder with a centered cylindrical bubble column of uniform optical properties.
    The paper fixes the hole diameter, places the bubble column in the center, and characterizes it by only D and effective scattering length; off-center or non-uniform bubble structures are not modeled, only the module offset r and phi.
  • domain assumption Henyey-Greenstein scattering with mean cosine 0.95 describes bubble-column Mie scattering.
    Equation (3.1) relates lambda_e and lambda_s using this phase function choice; no measured phase function for the bubble column is used.
  • domain assumption The NPE ratio is governed by Poisson statistics, a 30% Gaussian LED-intensity spread, and a 20% Gaussian charge resolution.
    The Section 4 bullet list introduces these assumptions; they set the width of the likelihood and therefore the recovered sensitivity.
  • domain assumption No systematic effects from bulk ice scattering, D-Egg misalignment, or cabling obstruction are included.
    The paper states this limitation at the end of Section 4; if these effects are large, the recovered contours will change.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Calibration LEDs in the IceCube Upgrade D-Egg Modules." pith.science (2026). https://pith.science/paper/JVR6UCUE

@misc{pith2026190810780,
  author       = {Pith},
  title        = {Pith review of: Calibration LEDs in the IceCube Upgrade D-Egg Modules},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JVR6UCUE}},
  note         = {Machine review of arXiv:1908.10780}
}
read the original abstract

The IceCube Upgrade, planned for deployment in the 2022/2023 South Pole Summer, will involve deployment of a greater density of optical modules (vertically spaced ~3 m). Improvements in the calibration of optical sensors and an enhanced understanding of the optical properties of the deep glacial ice, due to the more closely-spaced modules, are projected to have a large impact on neutrino reconstruction. A new optical sensor module called the "Dual optical sensors in an Ellipsoid glass for Gen2" (D-Egg), is planned for installation in both the IceCube Upgrade and IceCube-Gen2, and has both an upward and downward facing 8" high quantum efficiency PMT. The D-Egg modules will make use of a downward-facing LED calibration system to measure the optical properties of the refrozen drill holes ("hole ice"). An inner section of the hole ice contains some impurities, which modify the optical properties of the ice; this area is known as the "bubble column". The measurement of this "hole ice" is critical both for the upgraded IceCube detector as well as the current generation IceCube, as refrozen ice contributes significant systematic uncertainties to the reconstruction of low energy neutrinos. A simulation was performed, where the size and optical properties of the bubble column were varied. A log likelihood function is constructed from the geometry of the D-Eggs and properties of the hole ice. The minimization recovers best fit values close to the Monte Carlo truth.

Figures

Figures reproduced from arXiv: 1908.10780 by the authors.

Figure 1
Figure 1. Camera image looking down￾ward into the hole ice. The bubble column is found on the center right. The IceCube Neutrino Observatory [1] cur￾rently consists of 5160 digital optical modules (DOMs) inserted deep into the glacial ice at the South Pole. Each DOM has a diameter of 33 cm, a single downward-facing 10” photomultiplier tube (PMT), and detects the Cherenkov light produced by charged particles passing through th… view at source ↗
Figure 2
Figure 2. a) Prototype D-Egg. b) Section drawing of the lower structure. The flasher holder is placed on the lower optical elastomer. Eight LEDs shooting horizontally are implanted along the side and four LEDs illuminating vertically at the bottom. 3. Monte Carlo simulations 3.1 Photon Propagation Photon propagation has been simulated using Geant4 [8]. To study the photon propagation in the bubble column, a large volume of bu… view at source ↗
Figure 3
Figure 3. a) Depiction of the simulation: two D-Egg modules are located (partially) inside the bubble column. Photons from the LED originating from the upper D-Egg module are Mie scattered in the bubble column. b) Position parameters r and φ. The stars represent the four downward-facing LEDs. φ = 0 when LED1 is the furthest from the center. c) Angular distribution of the LED with a viewing angle of 120◦ [11]. 10 20 30 40 50 6… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Simulation results when r = 0, for NPE averaged over 200 runs. 5 [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Simulation results when LED1 flashes, λe = 100 cm and D = 15 cm, for NPE averaged over 200 runs. 0 50 100 150 200 250 300 350 Á [deg] 0 2 4 6 8 10 12 14 16 r [c m] IceCube Work in Progress 200 runs of 2e7 photons 0 40 80 120 160 200 240 280 320 NPE NPE at the forward P…
Figure 6
Figure 6. Figure 6: Simulation results when LED1 flashes, λe = 100 cm and D = 40 cm, for NPE averaged over 200 runs. • The range of LED intensities follows a Gaussian distribution with a standard deviation of 30%. • NPE follows a Poisson distribution with the mean NPE determined from simu…
Figure 7
Figure 7. Figure 7: Log likelihood ratio for each MC truth. Best fit parameter bin is marked with a white star. 8 [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 8 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION blank.sep after.quote 'output.state := FUNCTION fin.entry output.state after.quoted.block = 'skip 'add.period if write newline FUNCTION new.block output.state before.all = 'skip output.state after.quote = after.quoted.block 'output.state := after.block 'output.state := if if FUNCTION new.sentence out...

  2. [2]

    IceCube Collaboration, M. G. Aartsen et al., JINST 12 (2017) P03012

    2017

  3. [3]

    IceCube Collaboration, M. G. Aartsen et al., Nucl. Instrum. Meth. A711 (2013) 73--89

    work page 2013

  4. [4]

    Rongen, EPJ Web Conf

    IceCube Collaboration, M. Rongen, EPJ Web Conf. 116 (2016) 06011

    work page 2016

  5. [5]

    IceCube Collaboration, M. G. Aartsen et al., Phys. Rev. 12 (2019)

    work page 2019

  6. [6]

    IceCube Collaboration, PoS(ICRC2019)1031 (these proceedings)

  7. [7]

    Shimizu, PoS(ICRC2017)1051 (2018)

    IceCube-Gen2 Collaboration and S. Shimizu, PoS(ICRC2017)1051 (2018)

    work page 2018

  8. [8]

    IceCube Collaboration, M. G. Aartsen et al., http://arxiv.org/abs/1412.5106 arXiv:1412.5106

    arXiv

Show all 12 references
  1. [9]

    Allison et al., Nucl

    J. Allison et al., Nucl. Instrum. Meth. A835 (2016) 186--225

    2016

  2. [10]

    L. G. Henyey and J. L. Greenstein, Astrophysical Journal 93 (1941) 70--83

    1941

  3. [11]

    Makino, EPJ Web Conf

    IceCube-Gen2 Collaboration, Y. Makino, EPJ Web Conf. 207 (2019) 06005

    2019

  4. [12]

    http://www.kingbright.com/attachments/file/psearch/000/00/20160808bak/KAA-3528RGBS-11(Ver.3B).pdf

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.