“High resolution ensemble analysis: linking correlations and spread to physical processes ” S. Dey, R. Plant, N. Roberts and S. Migliorini NWP 4: Probabilistic and ensemble forecasting at short and medium-range 13/09/2013
Feb 23, 2016
“High resolution ensemble analysis: linking correlations and spread to physical processes ”
S. Dey, R. Plant, N. Roberts and S. Migliorini
NWP 4: Probabilistic and ensemble forecasting at short and medium-range
13/09/2013
Overview
• Linking ensemble evolution with physical processes
• Understanding of convective events• Evaluating on believable scales
Objective: Investigate methods of evaluating high resolution ensembles
Background Case study Results
Background 1: spatial predictabilityPredictability limits“certain turbulent systems, possibly including the earth’s atmosphere, possess for practical purposes a finite range of predictability”
(Lorentz 1969)
Scale dependence – Faster error growth at smaller scales
(Hohenegger and Schär 2007, BAMS)– Need ensembles at convective scale
Upscale error growth: A forecast can be unpredictable at grid scale but predictable at larger scales.
– Should be evaluating on scales that are believable
Background 2: correlations
Bannister 2008, QJRMS
Auto-correlations
Autocross- correlations
(x…
,y…
,z…
)
(x…,y…,z…)
Data Assimilation: Background error covariance matrix (B)
• Sampling uncertainties• Localization
• Present method of analysing the ensemble using correlations. • Present one case study to show utility of techniques: future
work to test on more cases
Method 1: case study
• MOGREPS-UK domain, UK Met Office UM 7.7 • 11 members + control• 8th July 2011• 2.2km grid spacing
>2mm
>10mm
13:00- 14:00
Method 2: Analysis
2σ
1. Vertical auto- and autocross-correlations
2. Neighbourhood approach
Gaussian weighting of perturbationsWidth set by FSS scale
• Believable scale• Variable dependant• Spatially varying
Results 1: Gaussian width
Rain rate spatial scales Horizontal divergence spatial scales
0 4 8 12 16 Grid points
15:00 on 8th July 2013
0 4 8 12 16 Grid points
Results 2: rain rate correlations
Convective layer
09:00 12:00 15:00
18:00 Single point sampling
error
Results 3: auto-correlations• 12:00 on 8th July 2013• Horizontal divergence
Single column
Spatially augmented ensemble
Heig
ht [k
m]
Heig
ht [k
m]
Height [km]
Height [km]
Results 4: autocross-correlations
Convergence
Divergence
-ve correlatio
n
+ve correlatio
n
Single columnHe
ight
[km
]
Height [km]
Spatially augmented ensemble
Heig
ht [k
m]
Height [km]
Clou
d Fr
actio
n
Horizontal divergence
Conclusions
1. Extra information from convective scale ensemble using correlations.
2. Neighbourhood sampling for analysis on meaningful scales.
3. Reduce sampling error and increase confidence.
4. Application to one case: future work to look at multiple cases.
Thanks for listening. Questions?
Bannister, R. N., 2008: A review of forecast error covariance statistics in atmospheric variational data assimilation. i: Characteristics and measurements of forecast error covariances. Quart. J. Roy. Meteor. Soc., 134, 1951–1970
Hohenegger, C. and C. Schär, 2007: Atmospheric predictability at synoptic versus cloud-resolving scales. Bull. Amer. Meteor. Soc., 88 (7), 1783–1793.
Lorenz, E. N., 1969: The predictability of a flow which possesses many scales of motion. Tellus, 21 (3), 289–307.
Roberts, N., 2008: Assessing the spatial and temporal variation in the skill of precipitation forecasts from an NWP model. Meteorol. Appl., 15 (1), 163–169.
Roberts, N. M. and H. W. Lean, 2008: Scale-selective verification of rainfall accumulations from high-resolution forecasts of convective events. Mon. Wea. Rev., 136 (1), 78–97.