The Lamont–Doherty Earth Observatory (LDEO) is a research unit of The Earth Institute at Columbia University that focuses on the earth sciences. It is located on a 157-acre (64 ha) campus in Palisades, New York, 18 miles (29 km) north of Manhattan on the Hudson River.
The Greenland Ice Sheet (GrIS) is a major contributor to sea-level rise, yet its long-term response to sustained warmth remains uncertain 1,2,3 . The Holocene represents the most recent interval during which the GrIS experienced centuries of temperatures above preindustrial levels, providing a natural benchmark for assessing its sensitivity to prolonged warming 4,5 . However, reproducing the geologically constrained evolution of the GrIS throughout the Holocene has remained a persistent challenge for ice-sheet models 6,7 . Here we present GrIS simulations forced by paleoclimate reconstructions and evaluated against extensive geological constraints to reconstruct its evolution from the Holocene to 2300 CE. By simultaneously capturing multiple independent records of past ice-sheet change, our ensemble provides a robust foundation for future projections. Under current warming trajectories, the GrIS shrinks below its Holocene minimum extent and volume by 2300 CE, while mass loss rates exceed Holocene maxima and contribute 26–86 cm of sea-level equivalent. Continued warming drives a nonlinear expansion of the ablation zone beyond its Holocene range, whereas substantial emissions reductions limit the sea-level contribution to approximately 10 cm. These results indicate that future warming may push the Greenland Ice Sheet beyond the range of states experienced during the Holocene.
Robust information on the spatial distribution of global carbon fluxes is required to project the future trajectory of carbon-climate feedback effects and atmospheric CO2 concentrations. Estimates of the latitudinal partitioning of carbon fluxes from top-down atmospheric CO2 inverse models currently diverge widely, because of methodological limitations or systematic biases in models or observations. We use airborne CO2 observations from the NASA Atmospheric Tomography Mission to evaluate and refine inverse model estimates from the Orbiting Carbon Observatory version 10 Model Intercomparison Project of total CO2 exchange for the two-year period of June 2016-May 2018. Applying emergent concentration-flux relationships as constraints reduces zonal total flux uncertainties by 46 to 56% relative to the full v10 MIP ensemble and by 17 to 28% relative to the subset excluding satellite observations over ocean. Subtracting independent estimates of fossil-fuel emissions and air-sea gas exchange results in residual land fluxes with a large northern extratropical sink, a small southern extratropical sink, and a small tropical source. The airborne-derived tropical land source disagrees with a large tropical land sink from process-based terrestrial models combined with estimates of land use emissions and river fluxes, representing an important challenge for our understanding of the global carbon cycle. The large implied northern extratropical sink can be explained either by underestimated land uptake by process models or a combination of process model bias and overestimated fossil fuel emissions.
Abstract SO2 flux and gas composition were measured during the 2021 flank eruption of Pacaya Volcano (Guatemala). The SO2 flux, measured with Tropomi, gradually rose from ∼2 to ∼30 kg/s over a 5 month period before the eruption. Field UV camera data showed that >95% of the SO2 was emitted from the summit crater. The Multigas composition measurements evidenced a depletion in CO2 and a significant enrichment in H2O over SO2 in the plume from the flank eruptive fissure compared to the central crater plume, and punctual amounts of H2S. We explain these variations by a combination of (a) Fractional degassing between a vertical conduit and an almost horizontal dyke diverging at a calculated pressure of 10–20 MPa, (b) SO2 scrubbing by a local and temporary hydrothermal system, and (c) Inclusion of groundwater in the magma dyke, even if visually the activity looked purely magmatic (lava spattering and effusion).
Reconstructions of past Antarctic Ice Sheet thinning are ground-truthed using geologic records from terrestrial cosmogenic nuclide measurements[1]. However, rates of cosmogenic nuclide thinning have historically been oversimplified with linear approximations; and how past rates compare to projected future change remains poorly understood, as does the degree to which local thinning profiles cap- ture regional trends[2]. We reconstruct Antarctic Ice Sheet thinning since Last Glacial Maximum using 766 quality-controlled cosmogenic nuclide measurements at 147 sites across Antarctica, then compare past thinning to modern and projected rates at adjacent ice streams using the ICESat-2 observational record[3] and ISMIP-6 projections[4, 5]. When aggregated, cosmogenic nuclide profiles capture regional trends that align with global climate fluctuations and yield rate distributions whose time dependence offers quantitative insight into driving processes. Under sustained high emissions, maximum postglacial rates are 61%(59–63%, 95% credible interval) likely to exceed modern rates, 53%(46–60%) likely to exceed rates by 2100, but only 16%(11–22%) likely to exceed rates by 2300. Our results indicate that Antarctica may soon thin faster than it has since Last Glacial Maximum—a finding with implications for which Antarctic areas will next become ice free.
Modeling the mechanical deformation of geological materials across the vast range of relevant thermodynamic conditions and dynamic processes remains a fundamental challenge in Earth sciences. Conventional rheological models often rely on isolated disciplinary approaches and hard-coded iterative solvers that struggle with strongly coupled, non-linear constitutive laws. Furthermore, inverting macroscopic laboratory data to infer sub-grid micro-physics is fundamentally ill-posed, leading to structural non-uniqueness and models that may violate thermodynamic laws during transient loading. To address these limitations, we present TIP-INN (Thermodynamics of Irreversible Processes Informed Neural Network), a thermodynamically consistent machine learning framework grounded in the Generalized Standard Materials (GSM) formalism. TIP-INN discovers constitutive laws directly from time-series stress-strain data by learning two strictly convex scalar thermodynamic potentials-the Helmholtz free energy and the dual dissipation potential-architecturally guaranteeing non-negative mechanical dissipation. By utilizing a continuous, differentiable neural integrator to solve internal state variable (ISV) evolution as an initial-value problem, TIP-INN enables stable rapid training across complex rheologies. We resolve the inherent non-uniqueness of the constitutive inverse problem through two regularizing strategies: (1) utilizing diverse loading protocols (e.g., cyclic and stress-relaxation paths) to map the full topological state space of the potentials, and (2) a multi-modal loss formulation that directly assimilates auxiliary seismo-acoustic data. Thus, TIP-INN provides a mathematically rigorous bridge between continuum damage mechanics and laboratory seismology. Finally, to ensure physical interpretability, the learned potentials are distilled via a Sparse Identification of Nonlinear Potentials (SINP) procedure into closed-form analytical expressions enabling efficient implementation in geodynamic codes.
