Climatic controls on interannual mass balance of Arctic glaciers and ice caps

Ma, Xiaojun; King, Matt; Bamber, Jonathan; Liu, Bin; Sun, Wenke; Wang, Qiuyu

doi:10.5194/egusphere-2026-3461

Preprints

https://doi.org/10.5194/egusphere-2026-3461

Preprints

08 Jul 2026

| 08 Jul 2026

Climatic controls on interannual mass balance of Arctic glaciers and ice caps

Xiaojun Ma, Matt King, Jonathan Bamber, Bin Liu, Wenke Sun, and Qiuyu Wang

Abstract. Interannual variability in glacier and ice cap (GIC) mass balance can be large amplitude, masking the underlying decadal trends associated with external forcing. Here we apply Independent Component Analysis (ICA) to global GIC mass anomalies derived from the Gravity Recovery and Climate Experiment (GRACE) mission and GRACE Follow-On (GRACE-FO) satellite gravimetry missions over 2002–2024. We show that Arctic glacier regions dominate the leading interannual variability in the global gravimetry record, and account for more than two-thirds of recent global GIC mass loss. Two leading Arctic ICA modes explain ~75 % of interannual variance and are associated with North Atlantic Oscillation (NAO)- and Pacific Decadal Oscillation (PDO)-related multi-year climate variability. Multiple linear regression further shows that variability linked to these climate modes explains much of the interannual glacier-mass variability in Arctic glacier regions and significantly affects trend estimates. This effect is most pronounced in Alaska, where the uncorrected trend is about 25 % less negative than the value after accounting for this variability (−69 ± 21 Gt yr⁻¹ versus −92 ± 16 Gt yr⁻¹). These results suggest that persistent North Atlantic and Pacific circulation variability can substantially change regional glacier mass loss, with direct implications for interpreting recent Arctic glacier change and their secular trends driven by external forcing.

Received: 12 Jun 2026 – Discussion started: 08 Jul 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 1076 KB)

Supplement (43725 KB)

Download & links

Preprint (1076 KB)
Metadata XML
Supplement (43725 KB)
BibTeX
EndNote

Xiaojun Ma, Matt King, Jonathan Bamber, Bin Liu, Wenke Sun, and Qiuyu Wang

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-3461', Anonymous Referee #1, 22 Aug 2026
This manuscript uses three GRACE/GRACE-FO Mascon products from JPL, CSR, and GSFC spanning 2002–2024 to investigate interannual-to-decadal variability in glacier and ice-cap mass changes globally and across the Arctic. The authors apply ICA to extract the dominant spatiotemporal modes and relate them to the cumulative NAO index and cumulative PDO index. They then employ a multiple regression model incorporating a linear trend, annual and semiannual cycles, and the climate-mode indices to assess the influence of these climate modes on regional glacier mass variability and long-term trend estimates. However, several issues should be addressed before the manuscript can be considered for publication.
Major Comments
Equation (2) defines B as the mixing matrix. In the standard ICA formulation, if X denotes the observed signals, Y the independent source signals, and B the mixing matrix, the mixing model should be X = BY. However, Equation (2) in the manuscript is written as X = B⁻¹Y. Please clarify the definition of B and explain the reason for this formulation. If B actually denotes the unmixing/separation matrix (or its inverse), the notation and terminology should be revised consistently.

The relationship between NAO/PDO variability and Arctic glacier mass change has been widely recognized. In Sections 3.1 and 3.2, the ICA analysis demonstrates significant correlations between the dominant components of detrended glacier mass variability and the NAO/PDO indices. These results appear to establish the influence of NAO and PDO on interannual or low-frequency variability after removal of the linear trend, but they do not by themselves explain how these climate modes are related to the long-term mass-change trend. This distinction should be clarified, and additional evidence is needed to support any interpretation linking the ICA results to long-term trends.

The manuscript does not yet provide a sufficiently clear explanation of why NAO and PDO would affect the estimated long-term glacier mass-change trends. Is there an identifiable physical mechanism through which persistent phases or low-frequency variability of these climate modes project onto the finite GRACE/GRACE-FO observation period and thereby lead to an underestimation or overestimation of the long-term trend? A more explicit physical interpretation is needed to distinguish genuine climate-driven long-term changes from trend biases caused by decadal variability over a relatively short observational record.

I recommend comparing the corrected long-term trends with independent external observations, such as satellite altimetry or other available glacier mass-balance products. Such a comparison would provide an important validation of the proposed correction and help demonstrate whether accounting for the climate modes actually improves the trend estimates.

Minor Comments
For the “Spatial patterns” shown in Figures 1 and 2, please specify clearly whether they represent a particular time/epoch or a spatial pattern derived from or averaged over the entire study period.

The manuscript should be carefully checked for subject–verb agreement, wording, and terminology consistency. For example, in Section 4.1, “sustained meridional circulation are” should be changed to “sustained meridional circulation is,” or the subject should be rewritten in the plural. In addition, terms such as “Arctic ice mass variability” and “Arctic glacier mass variability” should be used consistently throughout the manuscript.

The upper and/or lower limits of the summation in Equation (1) appear to be written incorrectly. Please check and correct them.
Citation: https://doi.org/10.5194/egusphere-2026-3461-RC1
RC2:
'Comment on egusphere-2026-3461', Anonymous Referee #2, 15 Sep 2026
------------------------
General comments
------------------------
The authors perform an Independent Component Analysis (ICA) to global GIC mass estimates from the GRACE and GRACE-FO satellite gravimetry missions over 2002–2024 in order to explain variability in ice mass associated with NAO and PDO. They present spatio-temporal variability across varying scales and further strengthen their arguments with regression analyses of atmospheric variables. Multiple regression highlights how accounting for NAO- and PDO-related variability impacts the estimated trend of gravimetric-derived mass estimates up to quarter/third of their magnitude.
This study contributes to the understanding of the impact of large-scale and regional climate variability on gravimetric estimates of GICs, focusing on the Arctic. They present their methods and results in a concise way. The manuscript is a good read and shows a mature stage. I have three significant concerns that should be addressed in the revision:
Most regression analysis in this study rely solely on direction and strength. The statistical metrics need to include significance testing.

I’d like to see how eventually the estimated ice mass variability and associated trends that are attributed to NAO- and PDO-related variability compares to other regional ice mass estimates across the Arctic (i.e. stemming from geodetic methods, such as shown in Fig. S2).

You discussed long-term trends and variability of NAO and PDO over the last decades. Given that you found how NAO and PDO explain variability in mass estimates and impact mass loss trends, please add a section in the discussion section that addresses future projections and trajectories of NAO and PDO variability and anomalies and mass trends in the Arctic.

Please find my specific comments below. All the best!

-------------------------
Specific comments
-------------------------
L99-100:           “We compared the GRACE-derived mass trends for the seven Arctic glacier regions with previously published regional estimates from earlier studies.” Needs a reference to Fig. S1.
L174-184:        List in this paragraph which atmospheric variables you did assess and justify your choice. Did you also include or tested separately for March-May (MAM) and September-November (SON)? If not, justify.
L294-296:        The abbreviations for the regions must appear way earlier in the script as those are used throughout all Figures of the Supplementary and are referred to already in the Methods section.
L303-310:        Here and throughout to study, I am missing an assessment of statistical significance. The statistical metrics shown solely rely on strength and direction.
L310:                   Color scale misses label. Also, uses cmw.e. while other part of the manuscript uses cm EWH. I’d recommend sticking to one of the SI units.
L315-316:        This paragraph needs to be moved to Section 3.2 and needs to refer to Fig. 2,3. The regional co-variation is only discussed later, and the argument is not sufficiently supported here when referring to Fig. 1a,b.
L325-345:        Color scales miss labels and bars need a different color signature as they do no relate to the continuous color scale of the map plot. In that sense, consider also using a positive and negative y-axis for Mass to avoid ambiguity. Also, the smoothing of the interannual mass anomaly variation appears here for the first time – this should already be included in the Methods section. Lowest and highest values indicating to maximum mass differences before and after 2012 should refer to the actual mass values and not the smoothed curve. Vertically dashed lines for the blue color must have same color as the font. Here too, consider moving away from the red/blue color coding to avoid ambiguity with the map plot color scale.
L346-347:        in the map plot in Fig. 3 it looks like some regions (i.e. ICE) are divided between the two zones. Please elaborate or justify.
L362-363:        To support the statistical measure of partial variance, the significance must also be assessed.
L369:                   Has the significance been tested for the conclusion about Scandinavia? If so, please state the significance/confidence level. What would be the argument then that the shape of the mass variability of SCA shown in Fig. 3 is so similar to ALA if insignificant? Please elaborate and justify.
L370-393:        Can you explain why for Svalbard and the Russian Arctic the GRACE-periodics overestimate mass change throughout mid/late-2000ish to late-2010ish and then understimate mass change since the early-2020s when compared to the GRACE-periodics-NAO-PDO (Fig. 4)? Different reference periods show the same trend even stronger (Fig. S22 g,i). Please elaborate and discuss.
L451-453:        Please use the same key number for your argument related to Alaska (i.e. compare to L22-23.
Supplem.:        Consider providing the supplementary data as Comma Separated Values (csv) files rather than txt files. The conversion should be quite straight forward.
Figure S1:         Uses cmw.e. while other sections in the manuscript use cm EWH. I’d recommend sticking to one of the SI units. Map plots need a scale or coordinate grid.
Figure S5:         [h] and [i] miss information in the figure caption what exactly is shown.
Figure S19:      Color scale misses label.
Citation: https://doi.org/10.5194/egusphere-2026-3461-RC2

Xiaojun Ma, Matt King, Jonathan Bamber, Bin Liu, Wenke Sun, and Qiuyu Wang

Supplement

https://doi.org/10.5194/egusphere-2026-3461-supplement

Xiaojun Ma, Matt King, Jonathan Bamber, Bin Liu, Wenke Sun, and Qiuyu Wang

Viewed

Total article views: 331 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
205	90	36	331	30	25	21

HTML: 205
PDF: 90
XML: 36
Total: 331
Supplement: 30
BibTeX: 25
EndNote: 21

Views and downloads (calculated since 08 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	104	39	15	158
Aug 2026	22	15	3	40
Sep 2026	79	36	18	133

Cumulative views and downloads (calculated since 08 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	104	39	15	158
Aug 2026	22	15	3	40
Sep 2026	79	36	18	133

Viewed (geographical distribution)

Total article views: 254 (including HTML, PDF, and XML) Thereof 254 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 30 Sep 2026

Short summary

Satellite records of glacier change are only about two decades long, so short-term climate swings can strongly affect estimated trends. Using satellite gravity data, we show that the largest year-to-year changes in global glacier mass come from the Arctic and are linked to the North Atlantic Oscillation and Pacific Decadal Oscillation. Accounting for these effects makes Alaska’s estimated ice loss about one-third larger, improving interpretation of glacier change and future sea-level rise.


Total:	0
HTML:	0
PDF:	0
XML:	0