Research ArticleCLIMATOLOGY

Contribution of the Greenland Ice Sheet to sea level over the next millennium

See allHide authors and affiliations

Science Advances  19 Jun 2019:
Vol. 5, no. 6, eaav9396
DOI: 10.1126/sciadv.aav9396
  • Fig. 1 Time series of air temperature anomalies, cumulative contribution to GMSL since 2008, and rate of GMSL rise due to mass changes of the Greenland Ice Sheet.

    (A) Ensemble minimum and maximum (thin lines) and mean (thick lines) of RCP 2.6, 4.5, and 8.5 temperature anomalies with respect to 2006–2015 derived from four GCM simulations that extend until 2300. Beyond 2300, the linear 2200–2300 trend was extrapolated to 2500, after which the 2500 value was kept constant (see Materials and Methods). The area between ensemble minimum and maximum is shaded. (B) Cumulative contribution to global mean sea level (GMSL) since 2008 (ΔGMSL). (C) Rates of GMSL in millimeter sea level equivalent (SLE) per year, M˙. (D) Contribution of ice discharge to M˙ given as D˙M˙=D˙/(D˙+R˙;)M˙, where D˙ and R˙ are ice discharge rate and surface runoff rate, respectively. (E) Ratio of ice discharge rate and the total of ice discharge rate and surface runoff rate, D˙%=D˙/(D˙+R˙). (B to E) Uncertainties are shaded between 16th and 84th percentile of the 500 ensemble members, the solid line is the median, and the thin dashed line is the control simulation. Some simulations under RCP 8.5 lose all ice, thus the 84th percentile of the cumulative contribution tapers out (B) and the rates decline (C). Rates in (C) to (E) are 11-year running means. The ensemble mean is used for the control simulation.

  • Fig. 2 Observed 2008 state and simulations of the Greenland Ice Sheet at year 3000.

    (A) Observed 2008 ice extent (53). (B to D) Likelihood (percentiles) of ice cover as percentage of the ensemble simulations with nonzero ice thickness. Likelihoods less than the 16th percentile are masked. (E) Multiyear composite of observed surface speeds (61). (F to H) Surface speeds from the control simulation. Basin names shown in (A) in clockwise order are southwest (SW), central-west (CW), northwest (NW), north (NO), northeast (NE), and southeast (SE). RCP 2.6 (B and F), RCP 4.5 (C and G), and RCP 8.5 (D and H). Topography in meters above sea level (m a.s.l.) [(A) to (H)].

  • Fig. 3 Evolution of ice sheet area and mass balance components for the control simulation for each RCP scenario.

    (A to C) Ice area evolution. (D to F) Partitioning of ice sheet wide mass balance rates into snow accumulation, runoff, and ice discharge into the ocean shown in Gt year−1 (D to F) and kg m−2 year−1 (left axis) and m year−1 ice equivalent (right axis) (G to I). We distinguish between runoff due to climate warming and runoff due to surface elevation lowering. (A) to (I) are plotted as 11-year running means.

  • Fig. 4 Retreat of two outlet glaciers in a similar climatic setting between 2015 and 2315 for the RCP 4.5 scenario.

    (A) Upernavik Isstrøm South (UIS) shows a gradual retreat of about 50 km over the next 200 years. (B) Store Gletscher (SG) is currently in a very stable position on a bedrock high. It takes almost a hundred years before substantial retreat happens. However, once the glacier loses contact with the bedrock high, retreat of 25 km occurs in less than a decade. The glacier retreats quickly until it is out of the water. Every line represents a year (A and B). (C) Location of the two outlet glaciers on the west coast, present-day observed surface speeds (61), and flow lines of Upernavik Isstrøm South and Store Gletscher (white dashed lines). Small inset shows area where the two glaciers are located.

  • Table 1 Contribution to GMSL in centimeters relative to 2008 for the years 2100, 2200, 2300, and 3000.

    Uncertainties for the ensemble analysis (ENS) at 1.8-km horizontal grid resolution are given as the 16th and 84th percentile range. In addition, the GMSL contribution from the control simulation (CTRL) at 900-m horizontal grid resolution is shown. To study the sensitivity of mass loss to grid resolution, we run additional simulations at 18 km (G18000), 9 km (G9000), 4.5 km (G4500), 3.6 km (G3600), 1.8 km (G1800), and 600 m (G600). NGIA is a simulation without glacio-isostatic adjustment, and NTLR is a simulation without a temperature lapse rate; both simulations were performed at 900 m. We also performed a simulation at 18 km that used the shallow-ice approximation (SIA18000). G600–18000, SIA18000, NGIA, and NTRL use the same parameters as CTRL.

    RCP 2.6RCP 4.5RCP 8.5
    210022002300300021002200230030002100220023003000
    ENS5–1911–3717–5559–1918–2320–5735–97197–41614–3352–15594–374538–728
    CTRL81625841128502431879174726
    NTRL6111639921361201563130573
    NGIA81625861128502491879175727
    G6006132176926462291675167727
    G180081523821128492431779174726
    G36006111868924432341676171727
    G4500591562723422311474170727
    G9000471260621412381477180728
    G18000471169723452891582196729
    SIA18000−2−5−7−251818147962154729
  • Table 2 Sobol indices computed from large ensemble of simulations.

    Values represent the percentage of variance in mass loss attributable to the variance in a given parameter. Large values imply that uncertainty in that parameter is responsible for a commensurately large uncertainty in mass loss. Small values imply that uncertainty in a given parameter has relatively little effect on uncertainty in total mass loss. Numbers for the variance in air temperature for year 3000 are in parentheses because they do not reflect the GCM intermodel variability but the choice of extrapolation.

    ParameterRCP 2.6RCP 4.5RCP 8.5
    210022002300300021002200 230030002100220023003000
    Climate (temperature and precipitation)
    ΔTair367246131616123845(33)
    ω000000000001
    37725613171612384534
    Surface processes (surface melt and refreezing)
    fi212632342328283728181110
    fs111315171110101011622
    ψ333332113110
    364150543740394842241413
    Ocean (submarine melt and calving)
    m˙xo122211111100
    m˙to011011111111
    hmin111111121221
    σmax111100000000
    444543343333
    Ice Dynamics (basal motion and internal deformation)
    q5134291538168325412
    E244423361101
    533833184018121026523

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/6/eaav9396/DC1

    Fig. S1. Coastward migration of basal cold ice competing with the inland migration of outlet glacier acceleration and thinning.

    Fig. S2. Time series of temperature anomalies for the four GCMs that extend until 2300.

    Fig. S3. Initial forcing and boundary conditions.

    Fig. S4. Ice discharge as a function of horizontal grid resolution.

    Table S1. Partitioning of mass fluxes for the control simulation.

    Table S2. Parameters and their distributions used in the ensemble analysis, source for distributions, and values for the control simulation CTRL.

    Movie S1. Evolution of the Greenland Ice Sheet over the next millennium.

  • Supplementary Materials

    The PDF file includes:

    • Fig. S1. Coastward migration of basal cold ice competing with the inland migration of outlet glacier acceleration and thinning.
    • Fig. S2. Time series of temperature anomalies for the four GCMs that extend until 2300.
    • Fig. S3. Initial forcing and boundary conditions.
    • Fig. S4. Ice discharge as a function of horizontal grid resolution.
    • Table S1. Partitioning of mass fluxes for the control simulation.
    • Table S2. Parameters and their distributions used in the ensemble analysis, source for distributions, and values for the control simulation CTRL.
    • Legend for movie S1

    Download PDF

    Other Supplementary Material for this manuscript includes the following:

    • Movie S1 (.mp4 format). Evolution of the Greenland Ice Sheet over the next millennium.

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article