NGGPS Dynamical Core Evaluation
At the heart of a numerical weather prediction system is the dynamical core, which describes how the air moves and how the atmosphere behaves. The first step in building the Next Generation Global Predication System (NGGPS) is to select the best possible dynamical core to give us the most accurate representation of the atmosphere.
To provide impartial oversight to the process of selecting a new core from five existing US research models, a Dynamical core Test Group (DTG) was formed. The DTG, comprised of federal, academia, selected subject matter experts, and representatives from each candidate dynamical core, will conduct an overall assessment of test criteria and results to provide a recommendation to NWS leadership.
NOAA announces the selection of the dynamical core:
NOAA to develop new global weather model
Dr. Ricky Rood talks about Weather Prediction and NGGPS with WX GEEKS:
The Future of Weather Forecasting
Technical Overview
Phase 3: Implementation - in process
-
GFDL's Finite Volume v3 (FV3) was selected as the dynamical core for the new NGGPS atmospheric model. Phase 3 dynamical core implementation includes incorporating FV3 into NEMS, and developing advanced physics and data assimilation techniques to match or exceed the skill of operational GFS. In addition, NWS is working with federal partners, universities, and the community to create a fully accessible community model.
Making a Community Model
NGGPS FV3-based Unified Modeling System will be a community guided system. Additional information can be found on the Community Participation page.
Phase 2: Continued Dynamical Core Computational Performance Evaluation- Complete
-
Selecting a single dynamical core upon which to build a unified global coupled system, will be achieved through assessing 10 additional criteria described in the table below. As results from Phase 2 testing are generated, they will be made available here.
# Evaluation Criteria Results1 1 Relaxing shallow atmosphere approximation (deep atmosphere dynamics) Results 2 Accurate conservation of mass, tracers, entropy, and energy Results 3 Robust model solutions under a wide range of realistic atmospheric initial conditions using a common (GFS) physics package Results
Results (web)4 Computational performance with GFS physics Results 5 Demonstration of variable resolution and/or nesting capabilities, including physically realistic simulations of convection in the high-resolution region Results 6 Stable, conservative long integrations with realistic climate statistics Results 7 Code adaptable to NEMS/ESMF Requirements 8 Detailed dynamical core documentation, including documentation of vertical grid, numerical filters, time-integration scheme and variable resolution and/or nesting capabilities Documentation 9 Evaluation of performance in cycled data assimilation Results 10 Implementation Plan (including costs) Results 1See Dynamical Core Test Plan for description of testing methodology
FV3 (pdf)
(Finite Volume on the Cubed Sphere) from the Geophysical Fluid Dynamics Laboratory (GFDL)
MPAS
(Model for Prediction Across Scales) from the National Center for Atmospheric Research (NCAR)
Summary Documentation
Dynamical Core Phase 2 Test Report (pdf)
AVEC Report: NGGPS Phase 2 Benchmarks and Software Evaluation (Criteria #4, 5, 10) (pdf)
Phase 1: Dynamical Core Computational Performance Evaluation - Complete
-
Phase 1 testing built upon the High Impact Weather Predication Project (HIWPP) and evaluated six potential candidate dynamical cores:
- Non-hydrostatic Global Spectral Model (GSM)-EMC1
- Global Non-hydrostatic Mesoscale Model (NMM & NMM-UJ) - EMC
- Model for Prediction Across Scales (MPAS) - NCAR
- Non-hydrostatic Icosohedral Model (NIM) - ESRL
- Navy Environmental Prediction System Using the NUMA Core (NEPTUNE) - Navy
- Finite Volume Model version3 - (FV3) - GFDL
1The Non-hydrostatic GSM was an initial candidate in Phase 1 but did not participate in tests.
The DTG's assessment of the tests and results in Phase 1 resulted in the recommendation to reduce the candidate pool to two cores. Assessment of the test results of Phase 1 determined that going forward to Phase 2 with two dynamical cores is a low technical risk, and that no unique dynamical core quality will be lost.
The conclusion of Phase 1 testing is the decision to proceed to Phase 2 testing on schedule with two dynamical cores:
FV3 (pdf)
Finite Volume on the Cubed Sphere from the Geophysical Fluid Dynamics Laboratory (GFDL)
MPAS
Model for Prediction Across Scales from the National Center for Atmospheric Research (NCAR)
NIM
The icosahedral grid used by the Non-hydrostatic Icosahedral Model (NIM) developed at NOAA Earth System Research Laboratory (ESRL)
Summary Documentation
Executive Summary Report (pdf)
Idealized Tests leveraging High Impact Weather Prediction Project (HIWPP) (pdf)
3-km Three Day Simulations leveraging High Impact Weather Prediction Project (HIWPP) (pdf)