mean systematic error of 500 hpa geopotential height fields lowres highres
TRANSCRIPT
How random numbers improve weather and climate predictions Expected and unexpected effects of stochastic
parameterizations
NCAR day of networking and discovery, April, 17, 2015
Judith Berner
With contributions from:
Dani Coleman, Hannah Christensen, Kate Fossell,
Soyoung Ha, Josh Hacker, Glen Romine, Craig
Schwartz, Chris Snyder Felipe Tagle
Mean systematic error of 500 hPa geopotential height fields
SKEBSLOWRES
• Reduction of z500 bias in all simulations with model-refinement
Berner et al., 2012
HIGHRES
Potential of stochastic parameterizations to reduce model error
Weak noise
Multi-modalUnimodal
Ball in double-potential well
Strong noise
Potential of stochastic parameterizations to reduce model error
Weak noise
Multi-modalUnimodal
Potential
Strong noise
Potential of stochastic parameterizations to reduce model error
Weak noise
Multi-modalUnimodal
Potential
Strong noise Stochastic parameterizations can change the mean and variance of a PDF Impacts variability Impacts mean bias
Outline
The stochastic parameterization schemes
Climate application: Impact in coupled and uncoupled simulations with the Earth System Model CESM
Weather application: Improving reliability and reducing analysis error in cycled and uncycled forecasts with the weather model WRF
Local tendency for variable X
Dynamical tendencies => Resolved scales
Physical tendencies => Unresolved scales
Rationale: Especially as resolution increases, the equilibrium assumption is no longer valid and fluctuations of the subgrid-scale state should be sampled (Buizza et al. 1999, Palmer et al. 2009, Berner et al. 2014)
Stochastically perturbed tendency scheme (SPPT)
Perturbs accumulated U,V,T,Q tendencies from physical parameterizations packages
Same pattern for all tendencies to minimize introduction of imbalances
Rationale: A fraction of the subgrid-scale energy is scattered upscale and acts as random streamfunction and temperature forcing for the resolved-scale flow (Shutts 2005, Berner et. al 08,09). Here simplified version with constant dissipation rate: can be considered as additive noise with spatial and temporal correlations.
Stochastic-kinetic energy backscatter scheme (SKEBS)
Stochastic Forcing Pattern
Local tendency for variable X =U,V,T
Dynamical tendencies => Resolved scales
Physical tendencies => Unresolved scales
Additive stochastic perturbation tendencies => Unresolved scales
Outline
The stochastic parameterization schemes
Climate application: Impact in coupled and uncoupled simulations with the Earth System Model CESM
Weather application: Improving reliability and reducing analysis error in cycled and uncycled forecasts with the weather model WRF
NCEP
SKEBS
CNTL
SPPT
21% 50%
35% 37%
Northern Annular Mode (MAM)1st EOF of sea level pressure over Northern Hemispheric Extratropics
CAM4 AMIP simulations (prescribed SSTs), 1900-2004
Stochastic parameterization improves pattern and reduces explained variance
Degenerate response: SKEBS and SPPT have same effect
NCEP
SKEBS
CNTL
SPPT
21% 50%
35% 37%
Northern Annular Mode (MAM)1st EOF of sea level pressure over Northern Hemispheric Extratropics
CAM4 AMIP simulations (prescribed SSTs), 1900-2004
Stochastic parameterization improves pattern and reduces explained variance
Degenerate response: SKEBS and SPPT have same effect
Sketch: CAM4 behavior
Potential with stochastic perturbations
Potential without stochastic perturbations
Including a stochastic parameterization does not lead to large changes in the pattern of modes of variability, but to decreased explained variances
This is consistent with a flattening of a potential well
Stochastic parameterizations can also lead to a depending of a potential well.
Sketch: CAM4 behavior
Potential with stochastic perturbations
Potential without stochastic perturbations
Including a stochastic parameterization does not lead to large changes in the pattern of modes of variability, but to decreased explained variances
This is consistent with a shallowing of a potential well
Stochastic parameterizations can also lead to a depending of a potential well.
Impact of SPPT on sea surface temperature (SST) variability
Coupled simulations with CAM4, 1880-2004
Too much variability in SSTs in Tropical Pacific
SPPT reduces bias in SST variability in Tropical Pacific
How can a stochastic parameterization reduce variability?
Impact of SPPT on sea surface temperature (SST) variability
How can perturbations to the atmosphere improve the ocean?
SPPT reduces variability in u850 variability over the Western Pacific
Probability density function of daily temperatures over North America (JJA)
AMIP simulations with CAM4
Too much variability in daily temperatures in summer compared to reanalysis
General extreme value distributions fitted to annual monthly temperature maxima and minima
CAM4 has to high return values for both, TMAX and TMIN
Overestimation of extreme temperatures
SKEBS and deteriorates 20yr return values for TMAX, but slightly improves values for TMIN
Tagle et al. 2015
Impact of SKEBS on precipitation bias
Coupled control run shows significant bias due to split inter-tropical convergence zone
SKEBS reduced bias in precipitation
Outline
The stochastic parameterization schemes
Climate application: Impact in coupled and uncoupled simulations with the Earth System Model CESM
Weather application: Improving reliability and reducing analysis error in cycled and uncycled forecasts with the weather model WRF
Representing initial uncertainty by an ensemble of states
Forecast uncertainty in weather models: Initial condition uncertainty Model uncertainty Boundary condition uncertainty
Represent initial forecast uncertainty by ensemble of states
Reliable forecast system: Spread should grow like ensemble mean error Predictable states with small error
should have small spread Unpredictable states with large error
should have large spread
\
analysis
spread
RMS error
ensemble mean
Spread and error near the surface
Ensemble is underdispersive (= not enough spread) Unreliable and over-confident Depending on cost-loss ratio potentially large
socio-economic impact (e.g. should roads be salted)
Solid lines: rms error of ensemble mean
Dashed: spread
Brier skill score near the surface
Brier skill measures probabilistic skill in regard to a reference (here CNTL). Verified event: μ<x<μ+σ
Berner et al., et al 2015
Reliability diagram for rain-thresholds, averaged over
forecast hours 18–36 using a 50-km neighborhood
Romine et al., 2014
V-10m T-2m
Including a model-error representation reduces the RMS error of the surface analysis (also prior) in 10m wind and Temperature at 2m
WRF-DART: Verification of surface analysis against independent
observations
CNTLSKEBSPHYS
Ha et al. 2015
Sketch: WRF behavior
Potential with stochastic perturbations
Potential without stochastic perturbations
Verifying observation
Including a stochastic parameterization increased ensemble spread
In cycled forecasts is reduces the mean analysis error
Debate in the field: A priori vs a posteriori
Model
Model uncertainty added a posteriori:
Process uncertainty added a priori during model development:
Forecast uncertainty
Stochasticity
Conclusions
Random numbers can improve weather and climate predictions by impacting variability and mean in expected (increase variability) and
unexpected (decrease variability) ways
Thank you!
Berner, J, K. Fossell, S.-Y. Ha, J. P. Hacker, C. Snyder 2015: “Increasing the skill of probabilistic forecasts: Understanding performance improvements from model-error representations, Mon. Wea. Rev., 143, 1295-1320
Berner, J., S.-Y. Ha, J. P. Hacker, A. Fournier, C. Snyder, 2011: “Model uncertainty in a mesoscale ensemble prediction system: Stochastic versus multi-physics representations” , Mon. Wea. Rev, 139, 1972-1995
Romine, G. S., C. S. Schwartz, J. Berner, K. R. Smith, C. Snyder, J. L. Anderson, and M. L. Weisman, 2014: “Representing forecast error in a convection-permitting ensemble system”, Mon. Wea. Rev, 142, 12, 4519–4541
Ha, S.-Y., J. Berner, C. Snyder, 2015: “Model-error representation in mesoscale WRF-DART cycling”, under review at Mon. Wea. Rev.