Ice sheet modeling of Greenland, Antarctica helps predict sea-level upward thrust

The Greenland and Antarctic ice sheets will make a dominant contribution to twenty first century sea-stage upward thrust if contemporary climate tendencies maintain. however, predicting the anticipated loss of ice sheet mass is tough due to the complexity of modeling ice sheet behavior.

To better understand this loss, a crew of Sandia country wide Laboratories researchers has been improving the reliability and efficiency of computational models that describe ice sheet behavior and dynamics. The team includes researchers Irina Demeshko, Mike Eldred, John Jakeman, Mauro Perego, Andy Salinger, Irina Tezaur and Ray Tuminaro.

This research is part of a 5-12 months challenge known as Predicting Ice Sheet and weather Evolution at excessive Scales (PISCEES), funded by means of the U.S. department of energy's (DOE) scientific Discovery via superior Computing (SciDAC) software. PISCEES is a multi-lab, multi-university undertaking that includes researchers from Sandia, Los Alamos, Lawrence Berkeley and all rightRidge country wide laboratories, the Massachusetts Institute of generation, Florida kingdom college, the college of Bristol, the university of Texas Austin, the university of South Carolina and big apple university.

Sandia's largest contribution to PISCEES has been an analysis tool, a land-ice solver referred to as Albany/FELIX (Finite factors for Land Ice eXperiments). The tool is primarily based on equations that simulate ice waft over the Greenland and Antarctic ice sheets and is being coupled to Earth models through the improved climate for energy (ACME) assignment.

"One of the goals of PISCEES is to create a land-ice solver this is scalable, speedy and strong on continental scales," stated computational scientist Irina Tezaur, a lead developer of Albany/FELIX. no longer only did the new solver want to be dependable and green, however it turned into vital that the team broaden a solver equipped with subsequent-era and superior evaluation capabilities.

Tezaur said the crew subsequent wishes to run the solver on new and emerging computers. additionally they want which will calibrate models and quantify uncertainties in predicted sea-level rise.

"The records we get from climate scientists are usually measurements from the pinnacle floor of the ice," she said. "To initialize an ice sheet simulation, we want facts approximately what's occurring interior and at the bottom of the ice. figuring out interior and bedrock ice properties is what we name version calibration, and requires the solution of an inverse trouble. a variety of our work has been in developing and enforcing optimization algorithms that are able to clear up these inverse issues robustly and efficaciously."

Tezaur stresses that the achievement of PISCEES is due in big component to strong collaborations between glaciologists, weather modelers, computational scientists and mathematicians.

"Glaciologists and weather scientist collaborators on PISCEES provide us with statistics sets to go into our version, even as computational scientists give you the right algorithms to apply and put into effect them correctly," she explains.

Enhancing computational modeling

The PISCEES mission started out in 2012 because there was no strong land-ice model as a constructing block for earth system models that calculate sea-stage rise predictions to assist the DOE's weather missions.

Sandia's Albany/FELIX solver includes advanced abilties and affords extended robustness, scalability and pace, Tezaur stated.

"In just 3 years of work, we've got created a subsequent-technology land-ice version that is established, scalable and robust and transportable to new and emerging structure machines," Tezaur said. "these models are equipped with superior analysis skills."

The Albany/FELIX solver became written the use of the so-called "component-based" software development approach, an method devised through Sandia computational scientists wherein new software codes are written the use of mature modular libraries. New solvers created the use of this method are "born" scalable: fast, sturdy and able to superior evaluation considering that they may be based on a group of algorithms developed and examined through area experts.

The components comprising Albany/FELIX are the Trilinos libraries, a group of open-supply applications developed by way of Sandia. in addition to recommending and executing the issue-based code development strategy, Sandia researchers have advanced methods for improving the robustness of the nonlinear solver. 

The Albany/FELIX code has demonstrated scalability up to at least one billion unknowns and tens of thousands of cores thanks to parallel scalable iterative linear solvers and newly advanced preconditioning techniques by way of Tuminaro. Adjoint-based totally deterministic inversion algorithms and software developed and carried out by using Perego have enabled rigorous model calibration. In collaboration with professionals from the quest SciDAC institute, Eldred and Jakeman, a framework for ahead and inverse uncertainty quantification (UQ) has been evolved. in the end, Albany/FELIX has been made portable to new structure machines thanks in large component to the efforts of laptop scientist, Irina Demeshko.

Verification and validation are vital

Even as code performance is crucial to the fulfillment of Albany/FELIX, similarly crucial are verification and validation, two tactics for evaluating a model and its code. Verification ascertains that a code is bug-loose. In contrast, validation seeks to check that the physical manner defined via a model is consistent with what is seen within the real global.

Sandia has carried out a radical verification of the Albany/FELIX solver the usage of the method of synthetic solutions, code-to-code comparisons on canonical land-ice benchmarks and by using appearing convergence studies on realistic Greenland and Antarctica landscapes with actual records, Tezaur said.

Tezaur explains that, in standard, validation is plenty tougher to do than verification. The diploma to which ice sheet models had been validated by means of observations within reason restricted, due in part to the limited period of the satellite tv for pc observation technology and the lengthy adjustment time scales of ice sheets. The PISCEES team has checked that its solver predicts ice sheet quantities which include floor velocities, surface mass stability and that these quantities are consistent with past and current observations.

"We're in the manner of doing a validation take a look at for the Greenland Ice Sheet for the period 1991-2012," Tezaur stated. "The ice sheet model output might be in comparison to ice floor elevation and ice sheet mass exchange observations from ICESat and IceSat 2, the Ice, Cloud and land Elevation satellite tv for pc."

Early consequences show promise for assessing the overall performance of different model configurations. A verification and validation take a look at suite, called the Land Ice Validation and Verification package, is being evolved via PISCEES collaborators at very wellRidge, and goes hand-in-hand with the efforts at Sandia.