Research ArticleCLIMATOLOGY

Mid-Pleistocene transition in glacial cycles explained by declining CO2 and regolith removal

See allHide authors and affiliations

Science Advances  03 Apr 2019:
Vol. 5, no. 4, eaav7337
DOI: 10.1126/sciadv.aav7337
  • Fig. 1 External drivers and best scenarios.

    (A) Summer solstice insolation at 65°N (S65N) (44). (B) Scenarios for the evolution of area of exposed crystalline bedrock (Arock) resulting from the gradual removal of regolith by glacial erosion over North America and Scandinavia (see Materials and Methods and fig. S3 for a spatially explicit illustration). (C) Volcanic CO2 (Volc C) outgassing scenarios (see Materials and Methods). (D) Root mean square error (rmse) and correlation (corr) between modeled and observed (3) benthic δ18O for simulations driven by the combination of different regolith removal and volcanic CO2 outgassing scenarios in (B) and (C). PD, present-day constant regolith and volcanic CO2 outgassing. The optimal scenarios, those minimizing root mean square error and maximizing correlation, are indicated by the white circles in (D) and by the thicker lines in (B) and (C).

  • Fig. 2 Transient modeling results.

    Results of model simulations driven by orbital forcing, optimal regolith removal scenario, and optimal volcanic outgassing scenario. In all panels, observations are shown in black and model results are shown as colored lines. (A) Benthic δ18O compared to the stack of (3). (B) Relative sea level compared to (49). (C) Calving from the Laurentide ice sheet into the North Atlantic compared to a proxy for ice-rafted debris at site U1313 (30). (D) Atmospheric CO2 concentration compared to ice core data (solid line) (50) and other proxies [circles: (16); squares: (18); *: (51); + and ×: (19); diamonds: (52); black box: (15); dotted lines: (17)]. (E) SST anomalies compared to the stack of (18). (F) Global annual surface air temperature compared to reconstructions (53). (G) Southern Ocean dust deposition compared to data (54).

  • Fig. 3 Benthic δ18O decomposition.

    (A) Contribution of sea level and deep ocean temperature to benthic δ18O variability through time, computed as moving SD with a window of 150 ka. The modeled total δ18O variability (gray lines) is compared to the variability of the stack from (3) (black). The modeled contribution of deep ocean temperature and sea level to δ18O variability is shown by the green and magenta lines, respectively. The decomposition in terms of sea level, zSL, and deep ocean temperature, Td, is derived using the following formula: δ18O = 4.0 − 0.22 Td − 0.01 zSL. (B) Ratio between deep ocean temperature and sea-level contribution to the total δ18O variability.

  • Fig. 4 Power and wavelet spectra of benthic δ18O.

    (A) Power spectra of modeled δ18O (blue) compared to power spectra of the δ18O stack of (3) (black) for different time intervals, as indicated above the panels. Wavelet spectra of (B) the benthic δ18O stack of (3) and (C) modeled δ18O. Black contours indicate the 5% significance level against red noise. The horizontal white lines represent the orbital periods of precession (∼23 ka), obliquity (∼41 ka), and eccentricity (∼100 ka). The model spectra are derived from the average δ18O computed over all ensemble members resulting from the time-splitting technique.

  • Fig. 5 Pre- and post-MPT ice sheets.

    Modeled maximum ice thickness in each grid cell (A) before and (B) after the MPT. The dotted lines in (B) indicate the reconstructed ice extent at the last glacial maximum.

  • Fig. 6 External drivers and model response.

    Modeled benthic δ18O compared to observations (black) (3) for (A) simulations driven only by variations in orbital configuration (green), (B) simulations driven by orbital forcing and optimal regolith removal scenario (red), and (C) simulations driven by orbital forcing and optimal volcanic outgassing scenario (blue).

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/4/eaav7337/DC1

    Fig. S1. Time-splitting technique.

    Fig. S2. Volcanic CO2 outgassing scenarios.

    Fig. S3. Regolith removal scenario.

    Fig. S4. Regolith scenarios.

    Fig. S5. Power spectra.

    Fig. S6. Comparison to additional observations and previous modeling results.

    Fig. S7. Transient simulations with present-day regolith and CO2 outgassing.

    Fig. S8. Transient simulations with present-day CO2 outgassing.

    Fig. S9. Transient simulations with present-day regolith.

    References (5558)

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. Time-splitting technique.
    • Fig. S2. Volcanic CO2 outgassing scenarios.
    • Fig. S3. Regolith removal scenario.
    • Fig. S4. Regolith scenarios.
    • Fig. S5. Power spectra.
    • Fig. S6. Comparison to additional observations and previous modeling results.
    • Fig. S7. Transient simulations with present-day regolith and CO2 outgassing.
    • Fig. S8. Transient simulations with present-day CO2 outgassing.
    • Fig. S9. Transient simulations with present-day regolith.
    • References (5558)

    Download PDF

    Files in this Data Supplement:

Navigate This Article