Fredrick H. M. Semazzi North Carolina State University
description
Transcript of Fredrick H. M. Semazzi North Carolina State University
Emerging Research Opportunities at the Climate Modeling
Laboratory
NC State University (Presentation at NIA Meeting: 9/04/03)
Fredrick H. M. SemazziNorth Carolina State University
Department of Marine, Earth and Atmospheric Sciences
&Department of Mathematics
Emerging Research Opportunities at the Climate Modeling
Laboratory
NC State University
For Details
http://climlab4.meas.ncsu.edu
•High Resolution Nested Regional Climate Prediction Models
•High Resolution Global Atmospheric Prediction Models
MAIN AREAS OF EMERGING RESEARCH OPPORTUNITIES
Equations of Motion
rFgvΩv
pDt
D
1
2
01
vDt
D
JDt
Dp
Dt
DTcv
MODEL NUMERICAL-DOMAIN
PERFORMANCE IN AN SIMULATING
CLIMATOLOGY (IRI)1970-1995 AVERAGE
• Observations• ENCHAM GCM• RCM-Low Resolution
Model• RCM-High Resolution
Model
OND ACTUAL RAINFALL (MM/DAY)
1970-95 AVERAGE
Comparison of models performance
Adaptation of RegCM2 for EA
0
50
100
150
200
250
TF KH SC TZ SA EA
Homogeneous climate sub-regions
rain
fall(
mm
)
Obs
Exp.1
Exp.2
Exp.3
Exp.4
Exp.5(Optimum)
Optimization of Regional Numerical Models Based on
Useable Prediction Skill
Useable Skill(Palmer et al, 1999)
Fig.2: Algorithm for computation of forecast value (V) & optimization
USER SECTOR PREDICTION MODELCLIMATE OBSERVATIONS
Define (E)Identity C & L
Set ParametersForecast E
Observe E Compute
o (E)
observed
Fst No Yes No Yes
Region 1
Region 2
Region 9ROC
Hit rate H / False alarm F Perfect Prediction
Model
Mper o CL
oL
CoHo
L
CFM
11
Climatology Prediction Model
MCli minCL
,o
HF
V pt MCli M
MCli Mper
Vopt max V
Savings$ MCli Mopt NL
See fig.3
Parameter updateand optimization
Global Atmospheric Model
==================Variable Resolution
Nonhydrostatic Global
Semi-implicit Semi-Lagrangian
Global Variable Resolution Grid
• No lateral boundary conditions
• Multiple scales
• Single code for multiple problems
• Flexible (easy to customize for different regions)
• Simplifies maintenance and optimization with only one code
Variable Resolution Grid
Nonhydrostatic Dymanics
• Increasing resolutions of atmospheric models
• Little additional computational cost
• Some atmospheric phenomena are nonhydrostatic (e.g. tropical cyclones)
BatesNASA GODDARD
GCM
Bates et al (1993)
NC STATE UNIVERSITY
GCM
Semazzi et al (2003)
Day 2 - 500 hPa
400 m resolution-courant#=3
Hydrostatic Non-Hydrostatic
2 km resolution
Hydrostatic Non-Hydrostatic
Future Work•Optimization of Regional Numerical Models Based on Useable Prediction Skill•Efficiency improvements to semi-implicit semi-Lagrangian (SISL) numerical scheme: solver, interpolation•Physical parameterization: heating, friction, convection, moisture, etc.•Parallel version
in collaboration with NASA and other organizations …