Parallel Ice Sheet Model

Parallel Ice Sheet Model
Name	Parallel Ice Sheet Model
Author	Scientific Computing and Imaging Institute, University of Washington
Developer	Los Alamos National Laboratory, NASA, University of Washington, University of Bristol
Released	1990s
Programming language	Fortran (programming language), C (programming language), C++
Operating system	Linux, Unix, Microsoft Windows
Platform	High-performance computing, Supercomputer
License	GNU General Public License

Parallel Ice Sheet Model is an open-source, higher-order, thermomechanical ice-sheet model used for simulating the evolution of continental and marine ice masses. It couples ice dynamics, thermodynamics, and subglacial processes to simulate ice-flow response to climate forcings relevant to studies of sea-level rise and paleoclimate. The model has been developed and applied by researchers across national laboratories, universities, and intergovernmental research programs.

Introduction

The Parallel Ice Sheet Model arose to address challenges in simulating ice dynamics at the scale of the Greenland ice sheet, Antarctic ice sheet, and ice sheets that existed during the Last Glacial Maximum. It integrates higher-order stress balance approximations, thermomechanical coupling, and subglacial hydrology to capture processes also studied in Intergovernmental Panel on Climate Change assessments and National Academies of Sciences, Engineering, and Medicine reports. The software is frequently used alongside climate models developed at organizations such as National Aeronautics and Space Administration, National Oceanic and Atmospheric Administration, and institutions including Columbia University and University of Bristol.

History and development

PISM traces its conceptual roots to earlier ice dynamics frameworks developed at Los Alamos National Laboratory and academic centers during the 1990s research surge following influential results from the International Geosphere-Biosphere Programme and paleoclimate reconstructions by teams linked to Lamont–Doherty Earth Observatory. Major code development was advanced through collaborations among University of Washington, Alfred Wegener Institute, University of Oregon, University of Calgary, and computational science groups at the Argonne National Laboratory. The project integrated numerical schemes drawn from finite-difference and finite-element traditions used in projects at Lawrence Livermore National Laboratory and was shaped by community modeling practices exemplified by Community Earth System Model efforts.

Model architecture and algorithms

PISM's architecture combines a modular framework for coupling distinct physics components with parallelization strategies optimized for platforms such as NERSC and Oak Ridge National Laboratory supercomputers. Core numerical methods include a higher-order approximation to the Stokes flow equations, shallow-ice and shallow-shelf approximations, and enthalpy-based thermodynamics inspired by work in ETH Zurich and University of Oslo research groups. Solvers leverage libraries and tools common in computational geoscience, including PETSc, multigrid preconditioners developed at Argonne National Laboratory, and domain decomposition techniques from Sandia National Laboratories. The codebase uses interfaces compatible with data formats popularized by NetCDF and HDF5.

Physical processes represented

PISM represents ice deformation governed by Glen's flow law derived from laboratory results associated with researchers at University of Strasbourg and Alfred Wegener Institute, basal sliding parameterizations influenced by field studies at Byrd Station and Siple Coast, thermomechanical coupling reflecting geothermal flux observations from Dome C and Dome F, and subglacial hydrology modules that build on cavity and sheet models developed in studies tied to University of Cambridge and University of Edinburgh. The model includes isostatic bedrock adjustment formulations related to studies of glacial isostatic adjustment by teams at University of Toronto and University of Colorado Boulder. Calving and ice–ocean interactions in marine-terminating sectors draw on parameterizations used in investigations around Pine Island Glacier and Thwaites Glacier.

Computational implementation and performance

PISM is implemented for distributed-memory parallelism using Message Passing Interface and is routinely deployed on leadership-class systems such as Blue Waters and Cray supercomputers. Performance scaling studies have been reported in venues connected with SC Conference and collaborations with computing centers like NERSC and Oak Ridge National Laboratory. The implementation emphasizes scalability for high-resolution regional simulations similar in ambition to coupled frameworks used by groups at Princeton University and University of Washington for long-term projections tied to IPCC scenarios. Workflow integration supports data assimilation approaches comparable to methods used at European Centre for Medium-Range Weather Forecasts.

Applications and case studies

Researchers use PISM for reconstructing ice-sheet behavior during the Last Glacial Maximum, projecting Greenland ice sheet mass loss under Representative Concentration Pathways studied in IPCC reports, and exploring dynamic responses of Antarctic ice sheet sectors such as Thwaites Glacier and Pine Island Glacier. The model has been applied in assessments by teams at NASA Goddard Space Flight Center, University of Bristol, and University of Washington and in interdisciplinary studies involving US Geological Survey and National Snow and Ice Data Center. PISM supports paleoceanographic reconstructions related to Younger Dryas events and contributes to hazard assessments informing policies debated in forums like the United Nations Framework Convention on Climate Change.

Validation, limitations, and uncertainties

Validation efforts compare PISM outputs against observations from satellite missions such as GRACE, ICESat, and CryoSat-2, as well as field campaigns at sites including Camp Century and Fletcher Ice Rise. Limitations include parameter sensitivity associated with basal rheology studies from University of Minnesota and uncertainties in subglacial hydrology constrained by measurements near Russell Glacier. Model limitations mirror challenges faced by coupled systems developed at Lawrence Berkeley National Laboratory and highlight needs for improved coupling to ocean models used by groups at Scripps Institution of Oceanography and Woods Hole Oceanographic Institution. Ongoing community development, testing, and intercomparison projects link PISM efforts to initiatives such as the Ice Sheet Model Intercomparison Project.

Category:Ice sheet models Category:Climate modeling software