High Performance Computing and Four-Dimensional Data Assimilation
This talk at CRPC Annual Meeting describes the NASA GRand Challenge in
Data Assimilation. This application combines, using Kalman
filter-like methos, models and atmospheric measurements to provide the
best possible weather prediction. CRPC is represented by NPAC at Syracuse
which is developing parallel versions of the key components of the
application. We are developing both message-passing and High Performance
Fortran implementations. The work involves the team responsible for
current operational system at NASA Goddard, JPL particularly focussing on
new Optimal interpolation algorithms and NPAC. Optimal interpolation
terms the best way of combining data and models and is challenging
because the of the complex model governed by sets of partial differential
equations and large number of observations. The parallel implementation
has load balancing and other interesting HPCC features. The talk was
given by Miloje Makivic from NPAC who is leading the this work at NPAC.
The slides (Click on number for slide, or text for descriptions):
-
.1=..NASA grand challenge in atmospheric science - status of computer science issues.
-
.2=..Data assimilation and high performance computing.
-
.3=..General computational issues.
-
.4=..Block diagram of data assimilation system.
-
.5=..Comments on data assimilation software modules.
-
.6=..Basic optimal interpolation algorithms.
-
.7=..Local optimal interpolation features.
-
.8=..Global optimal interpolation features.
-
.9=..Eulerian Transport.
-
10=..Semi-lagrangian general circulation model - I.
-
11=..Semi-lagrangian general circulation model - II.
-
12=..Project milestones.
-
13=..Structure of grand challenge team.
InfoWarehouse Management team