Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Earth science

How glaciers grow

A state-of-the-art numerical model shows that the advance of glaciers in a cooling climate depends strongly on the pre-existing landscape, and that glacial erosion paves the way for greater glacial extent in the future. See Letter p.206

The growth of glaciers reflects the balance between the accumulation of snow and its loss through melting. Given the close association between altitude and temperature1, the elevation and morphology of the valley floor on which a glacier forms are key determinants of a glacier's size and longevity. Glacial erosion has carved the spectacular alpine landscapes of many mountain ranges, and these are characterized by extensive glacial valley floors at elevations close to the long-term snowline2. These landscapes contrast strongly with their precursors, which are generally steeper, narrower valleys sculpted by rivers (Fig. 1). On page 206 of this issue, Pedersen and Egholm3 use a state-of-the-art numerical model of glacier dynamics to quantify, for the first time, the stark contrast in glacier development between modern alpine mountain ranges and those that came before them. They compare the legacy of numerous glaciations during the Pleistocene epoch (between about 2.5 million and 10,000 years ago) with the landscapes that would have been present at the onset of these glaciations.

Figure 1: The morphological contrast between fluvial and glacial landscapes.


As documented by Pedersen and Egholm3, the width and downstream gradient of a valley floor determine how glaciers would develop should the climate cool sufficiently. a, A fluvial landscape characterized by a narrow, steep valley: Taroko Valley, Hualien, Taiwan. Many of the world's mountain ranges would have been dominated by fluvial erosion before the onset of glaciations about 2.5 million years ago. The first glaciers to develop in such fluvial landscapes will show an almost linear relationship between temperature change and subsequent ice volume. b, A glacial landscape: Yosemite Valley, California. Ice volume will increase dramatically if the temperature cools sufficiently for the glacier to occupy the broad, shallow valley floor.

It has long been recognized that glaciers and their surroundings share an intimate, coupled relationship4,5. Shaded aspects and large areas at high elevation promote glacier growth. In turn, glacial erosion strongly modifies the underlying landscape, widening valley floors and making hill slopes steeper. However, given the efficiency of glaciers at reworking and removing the evidence left by previous glacial cycles, direct insight into the relationship between glaciers and topography during the initial phases of late Cenozoic glaciation (between about 2.5 million and 1 million years ago) has remained frustratingly elusive.

Numerical modelling offers the opportunity to explore glacier growth during initial glaciation. Pedersen and Egholm demonstrate an almost linear relationship between the degree of climate cooling and ice volume when glaciers develop in fluvial landscapes. However, for landscapes previously occupied and sculpted by glaciers, the equivalent relationship is highly nonlinear. Once the climate cools sufficiently, glaciers that descend onto the wide, shallow valley floor carved by preceding glacial occupations expand markedly.

Pedersen and Egholm also explore the full evolution of modern glacial landscapes by driving a simplified version of their numerical model — that is, without fluvial or hill-slope erosion, or active tectonics — with temperature fluctuations representing the past 2 million years. The mid-Pleistocene transition (MPT) about 950,000 years ago marked a shift from 40,000-year glacial cycles accompanied by symmetrical periods of cooling and warming to the protracted cooling and rapid warming of the more recent 100,000-year glacial cycles. The authors find much more extensive and erosive glaciers after the MPT than during the previous 1 million years. This result is consistent with geological evidence for accelerated glacial erosion after the MPT6,7. However, as Pedersen and Egholm show, this accelerated erosion is not simply a consequence of the changing climate; faster glacial erosion post-MPT was preconditioned by the landscape modifications made by the preceding, smaller-scale glaciations.

Pedersen and Egholm3 have taken on a considerable challenge. Numerical modelling of landscape evolution resulting from glacial erosion is a complex, multi-faceted problem. As such, the first models of glacial landscape evolution have emerged only within the past 15 years8. Egholm and co-authors3,9 have led the field and achieved compellingly realistic results by introducing ice-dynamics formulations that are more appropriate to valley glaciers — which descend the relatively steep downstream gradient of the valley floor and interact with the valley sides —than earlier simplifications borrowed from ice-sheet models.

Many considerations are involved in any attempt to numerically model glaciers and glacial erosion. The amount of snow that falls is strongly influenced by local climate, and is sensitive to changes in atmospheric moisture content, temperature and prevailing wind direction. Falling snow can be redistributed by wind10 and avalanching before it reaches the glacier surface. Glacial-ice melt is influenced by temperature change, shading and rock fall from the valley sides. Glacial ice deforms under its own weight, and can also slide on its bed if liquid water is present. However, at this point, this generic model still only describes ice accumulating, deforming and melting. How would this moving ice (and water) erode the landscape?

As was recently noted11, models of glacial landscape evolution, including Pedersen and Egholm's, still rely on simple, empirical relationships between glacier sliding velocity and glacial erosion rate, supported by modest field-data sets. There is little clear connection between these numerical relationships and the quarrying of large bedrock blocks from the glacier bed, which in most circumstances represents the primary process of glacial erosion. The liquid water in the subglacial hydrological network is also known to have a key role in glacial erosion, but determining the appropriate numerical formulation for the role of subglacial water is in its infancy12. Further progress will require numerical models that are robustly supported by a combination of careful observations of modern glaciers13 and thermochronological evidence for how glacial landscapes have evolved over the longer timescales7 that are explored numerically here. It will also require driving models with more-realistic climate representations10,14 than those currently considered, and will probably depend on continuing advances in computer power.


  1. 1

    Stone, P. H. & Carlson, J. H. J. Atmos. Sci. 36, 415–423 (1979).

    ADS  Article  Google Scholar 

  2. 2

    Brozović, N., Burbank, D. W. & Meigs, A. J. Science 276, 571–574 (1997).

    Article  Google Scholar 

  3. 3

    Pedersen, V. K. & Egholm, D. L. Nature 493, 206–210 (2013).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Johnson, W. D. J. Geol. 12, 569–578 (1904).

    ADS  Article  Google Scholar 

  5. 5

    Gilbert, G. K. J. Geol. 12, 579–588 (1904).

    ADS  Article  Google Scholar 

  6. 6

    Haeuselmann, P., Granger, D. E., Jeannin, P.-Y. & Lauritzen, S.-E. Geology 35, 143–146 (2007).

    ADS  Article  Google Scholar 

  7. 7

    Valla, P. G., Shuster, D. L. & van der Beek, P. A. Nature Geosci. 4, 688–692 (2011).

    ADS  CAS  Article  Google Scholar 

  8. 8

    Braun, J., Zwartz, D. & Tomkin, J. H. Ann. Glaciol. 28, 282–290 (1999).

    ADS  Article  Google Scholar 

  9. 9

    Egholm, D. L., Knudsen, M. F., Clark, C. D. & Lesemann, J. E. J. Geophys. Res. 116, F02012 (2011).

    ADS  Article  Google Scholar 

  10. 10

    Anders, A. M., Roe, G. H., Montgomery, D. R. & Hallet, B. Geology 36, 479–482 (2008).

    ADS  Article  Google Scholar 

  11. 11

    Iverson, N. R. Geology 40, 679–682 (2012).

    ADS  Article  Google Scholar 

  12. 12

    Herman, F., Beaud, F., Champagnac, J.-D., Lemieux, J.-M. & Sternai, P. Earth Planet. Sci. Lett. 310, 498–508 (2011).

    ADS  CAS  Article  Google Scholar 

  13. 13

    Riihimaki, C. A., MacGregor, K. R., Anderson, R. S., Anderson, S. P. & Loso, M. G. J. Geophys. Res. 110, F03003 (2005).

    ADS  Article  Google Scholar 

  14. 14

    Rowan, A. V., Plummer, M. A., Brocklehurst, S. H., Jones, M. A. & Schultz, D. M. Geology (2012).

Download references

Author information



Corresponding author

Correspondence to Simon H. Brocklehurst.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Brocklehurst, S. How glaciers grow. Nature 493, 173–174 (2013).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing