Deglaciation-enhanced mantle CO2 fluxes at Yellowstone imply positive climate feedback

Abstract

Mantle melt generation in response to glacial unloading has been linked to enhanced magmatic volatile release in Iceland and global eruptive records. It is unclear whether this process is important in systems lacking evidence of enhanced eruptions. The deglaciation of the Yellowstone ice cap did not observably enhance volcanism, yet Yellowstone emits large volumes of CO₂ due to melt crystallization at depth. Here we model mantle melting and CO₂ release during the deglaciation of Yellowstone (using Iceland as a benchmark). We find mantle melting is enhanced 19-fold during deglaciation, generating an additional 250-620 km³. These melts segregate an additional 18-79 Gt of CO₂ from the mantle, representing a ~3-15% increase in the global volcanic CO₂ flux (if degassed immediately). We suggest deglaciation-enhanced mantle melting is important in continental settings with partially molten mantle - including Greenland and West Antarctica - potentially implying positive feedbacks between deglaciation and climate warming.

Fig. 1. Background mantle temperatures and melt fractions (i.e., “degree-of-melting”), prior to unloading.

Temperatures beneath (a) Iceland and (b) Yellowstone are plotted in red-blue. The thick black line is the lithosphere-asthenosphere boundary (LAB), at T = 1300 °C. The green parabola represents the ice volume at its maximum (10-fold vertical exaggeration). Black arrows indicate imposed plate motions. Melt fractions in blue-green plotted for (c) Iceland and (d) Yellowstone.

Fig. 2. Modeled melt production due to deglaciation of Iceland ice sheet.

The ice sheet is represented by the green parabola at a given time step and by the dashed black line at its maximum extent. Rates of pressure change are colored teal-brown (a, c) and rates of melt fraction change are colored blue-orange (b, d). Top row shows a model time step prior to any glacial loading/unloading, while bottom row shows a time step 500 years following deglaciation onset. Red arrows show mantle flow (arrow size is scaled to velocity magnitude); the thick black line is the lithosphere-asthenosphere boundary (LAB), at T = 1300 °C.

Fig. 3. Evolution of melt production rate and CO₂ flux during deglaciation.

a Ice volumes used as model forcings for Iceland (blue) and Yellowstone (red) during the deglaciation intervals (shaded). Melt production rates (black lines) for (b) Iceland and (c) Yellowstone; background rates from time steps prior to loading/unloading are plotted in orange. Note the different scales (2–D for Iceland; 3–D for Yellowstone). refer to Figure S9b for 3–D Yellowstone ice volumes. CO₂ fluxes for (d) Iceland and (e) Yellowstone assuming mantle source CO₂ concentrations of 150 and 600 ppm are plotted as dashed and solid lines, respectively. Estimates of modern melt production rates and magmatic CO₂ fluxes for Iceland^, and Yellowstone^,, are denoted by purple bars. “WB03” refers to Werner & Brantley, “M18” refers to McMillan et al., “HL14” refers to Hurwitz & Lowenstern.

Fig. 4. Modeled melt production due to deglaciation of Yellowstone ice cap.

The ice cap is represented by the green parabola at a given time step and by the dashed black line at its maximum extent. Rates of pressure change are colored teal-brown (a, c) and rates of melt fraction change are colored blue-orange (b, d). The top row shows a model time step prior to any glacial loading/unloading, while the bottom row shows a time step 1000 years following deglaciation onset. Red arrows show mantle flow (at shallow depths only for ease of visualization), the thick black line is the lithosphere-asthenosphere boundary (LAB), at T = 1300 °C.

Fig. 5. Influence of rheologic and loading parameters on change in pressure at depth.

(yellow-green colors) after one deglaciation, in a viscoelastic half-space. A pressure change of 10⁶ Pa in partially molten mantle scales to 3 km³/km² of additional melt, under a melt productivity (dF/dP) of 0.15/GPa and 20-km melt column. Pressure changes as a function (a) shear modulus (G) and overlying viscosity, at a constant depth of 70 km and load radius of 50 km. The influence of rheology is small, and is bounded in the limit of rigid/soft lithosphere (white stars correspond to the parameter analysis in Section S3.4). b For constant rheological parameters (G = 10 GPa, η = 10²⁴ Pa s), the load radius controls the magnitude of pressure changes as a function of depth (beneath the load center). Rectangles represent different settings (red for Yellowstone, blue for Iceland, black for West Antarctica). Note that pressure changes are directly proportional to load height, here kept fixed at 1.25 km.