Documentation for PISM, the Parallel Ice Sheet Model

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

applications:201209 [2012/09/03 05:22]
Ed Bueler created from future_applications:201209
applications:201209 [2012/09/03 05:59] (current)
Ed Bueler
Line 12: Line 12:
 | **venue**: | [[http://​www.igsoc.org/​journal/​|Journal of Glaciology]] | | **venue**: | [[http://​www.igsoc.org/​journal/​|Journal of Glaciology]] |
  
-Inverse methods are used to estimate model parameters from observations,​ here estimating ​basal shear stress from the surface velocity of an ice sheet.  ​These methods start with an initial estimate of the model parameters and then they update ​the parameters ​to match observations in an iterative process. ​ Large-scale spatial features are adjusted first. ​ A stopping criterion prevents the overfitting of data.  In this paper, iterative inverse methods are applied to the shallow-shelf approximation. ​ A new incomplete Gauss–Newton method is introduced and compared to the steepest descent and nonlinear conjugate gradient methods. Two different stopping criteria, the discrepancy principle and a recent-improvement threshold, are compared. The IGN method shows faster convergence than the others. ​ Though PISM is not mentioned by this paper, and its [[https://​github.com/​damaxwell/​siple|experiments were done in python]], code supporting these inversion methods is already present in the [[https://​github.com/​pism/​pism/​tree/​dev|PISM dev branch]].+Inverse methods are used to estimate model parameters from observations,​ here basal shear stress from the surface velocity of an ice sheet.  ​One starts ​with an initial estimate of the model parameters and then updates them to improve ​the match to observations in an iterative process. ​ Large-scale spatial features are adjusted first. ​ A stopping criterion prevents the overfitting of data.  In this paper, iterative inverse methods are applied to the shallow-shelf approximation ​forward model.  A new incomplete Gauss–Newton method is introduced and compared to the steepest descent and nonlinear conjugate gradient methods. Two different stopping criteria, the discrepancy principle and a recent-improvement threshold, are compared. The IGN method shows faster convergence than the others. ​ Though PISM is not mentioned by this paper, and [[https://​github.com/​damaxwell/​siple|the experiments were done in python]], code supporting these inversion methods is already present in the [[https://​github.com/​pism/​pism/​tree/​dev|PISM dev branch]].
  
 </​div>​ </​div>​
 <WRAP clear></​WRAP>​ <WRAP clear></​WRAP>​
applications/201209.txt · Last modified: 2012/09/03 05:59 by Ed Bueler
© 2019 by PISM | webmaster