variational data assimilation and forecast error statistics

A B1 B2 C D E F G1 G2 G3 H I J K1 K2 K3 L

Variational data assimilationand forecast error statistics

Ross Bannister, 11th July 2011University of Reading, [email protected]

“All models are wrong …” (George Box)

“All models are wrong and all observations are inaccurate” (a data assimilator)


Distinction between ‘errors’ and ‘error statistics’When people say ‘errors’ they sometimes (but not always) mean ‘error statistics’

Error: The difference between some estimated/measured quantity and its true value. E.g. εest = xest – xtrue or εy = y – ytrue

Errors are unknown and unknowable quantities

Error Statistics: Some useful measure of the possible values that ε could have.

E.g. a PDF

Error statistics are knowable, although often difficult to determine – even in the Gaussian case.

Here, error statistics = second moment (ie assume PDFs are Gaussian and unbiased, <ε> = 0).

ε

PDF(ε)

E.g. second moment of ε, <ε2> (called a variance), or <ε2>1/2 = σ (standard deviation). If only the variance is known, then the PDF is approximated as a Gaussian.

P(ε) ~ exp – ε2/2<ε2>

σ= √<ε2>

ε


Plan

A. Erroneous quantities

B. Error statistics are important

C. ‘Observation’ and ‘state’ vectors

D. ‘Inner’ and ‘outer’ products

E. Error covariances

F. Bayes’ theorem

G. Flavours of variational data assimilation

H. Control variable transforms

I. The ‘BLUE’ formula

J. A single observation example

K. Forecast error covariance statistics

L. Hybrid data assimilation


A. Erroneous quantities in data assimilation

Data that are being fitted to

• Observations (real-world data)• Prior data for the system’s state• Prior data for any unknown parameters

Imperfections in the assimilation system

•Model error within the da system• Representivity

Data that have been fitted

• Data assimilation-fitted data (analysis, ie posteriori error statistics)

}

Information available about the system before observations are considered.

instrumentvityrepresenti

instrumentvityrepresentiresolvedtrue

exact

instrumentunresolvedtrue

resolvedtrue

exact

instrumentunresolvedtrue

resolvedtrue

exact

unresolvedtrue

resolvedtruetrue

instrumenttrueexact

modeltrueapprox

)(

)(

)(

)(

)(

)(

y

y

H

H

H

H

H

x

Hxxy

xxy

xxx

xy

xy


B.1 Error statistics are important1. Error statistics give a measure of confidence in data (eg here obs error stats)

No assim

Assim with large obs errors Assim with small obs errors


B.2 How are error statistics important?2. Error statistics of prior data imply relationships between variables

x1

x2

time

Background forecast (no assim)Analysis forecast (consistent with prior and ob errors)

x1 and x2 cannot be varied independently by the assimilation here because of the shape of the prior joint PDF.

• Known relationships between variables are often exploited to gain knowledge of the complicated nature of the prior error statistics (e.g. changes in pressure are associated with changes in wind in the mid-latitude atmosphere (geostrophic balance)).


C. ‘Observation’ and ‘state’ vectors

The structure of the state vector (for the example of meteorological fields u, v, θ, p, q are 3-D fields; λ, φ and ℓ are longitude, latitude and vertical level). There are n elements in total.

x =

The observation vector – comprising each observation made. There are p observations.

y =

extra parameters


D. ‘Inner’ and ‘outer’ products

The inner product (‘scalar’ product) between two vectors gives a scalar

The outer product between two vectors gives a matrix

When ε is a vector of errors, <εεT> is a (symmetric) error covariance matrix

1T2211

1

121

T ,,

) (

babababa nnn

n

n

bababa

b

bbaaa

nmnm

nmmm

n

nn

m bababa

bababababababbb

a

aa

T

21

22212

1211121

2

1

T , ,

) (

abbaab


E. Forms of (Gaussian) error covariancesThe one-variable case

2

2

2

2

2

2

2)(

exp2

1)(

2exp

2

1)(

xxxP

P

(unbiased) true

xx

xx

)()(

21exp

2

1)( 1T xxSxxS

x

nP

later used be willsymbols specific - symbol generic a is

2232

2121

S

S

(unbiased) true

xxxxε

0

1x2x

)(xP

x

)(xP σ= √<ε2>

<x>

The many variable case


F. Bayes’ theorem and the variational cost function

Bayes theorem links the following•PDF of the observations (given the truth)•PDF of the prior information (the background state)•PDF of the state (given the observations – this is the objective of data assimilation)

))(())((21)()(

21)(

)( minimize )|(ln minimize )|( maximize

)()(21))(())((

21ln)|(ln

)()(21))(())((

21exp

)()(21exp))(())((

21exp

)()|()|(

1Tb

1-f

Tb

b1-

fT

b1T

b1-

fT

b1T1

b1-

fT

b1T1

1

xhyRxhyxxPxxx

xyxyx

xxPxxxhyRxhyyx

xxPxxxhyRxhy

xxPxxxhyRxhy

xxyyx

J

JPP

AP

PPP

A

A

A

Lorenc A.C., Analysis methods for numerical weather prediction, Q.J.R.Meteor.Soc. 112 pp.1177-1194 (1986)


General form

Simplified form : assume model errors are uncorrelated in time (white noise)

G.1 Flavours of variational data assimilation

T

t

T

ttttttt

T

tttt

tttt

tttt

J

1 1'1''',

1T1

0

1T

b1-

fT

b

)])1'([)'(()()])1([)((21

)])([)(()])([)((21

))0()0(())0()0((21)(

xmxQxmx

xhyRxhy

xxPxxx

1. Weak constraint 4D-VAR

)(

)1()0(

Tx

xx

x

Q : model error covariance matrix

T

ttttt

ttttt

tttt1

11T

1

1'',

1

)])1([)(()])1([)((21

becomes last term

)(

xmxQxmx

QQ


G.2 Flavours of variational data assimilation2. Strong constraint 4D-VAR

Assume perfect model:

T

tttttt tt

J

00

1T0

b1-

fT

b

)])([)(()])([)((21

)()(21)(

xmhyRxmhy

xxPxxx

)0(only )0( determine toneed

)]1([)(0

1

xxx

xmxQ

tt tt

)0(xx

The ‘strong constraint’ approximation is valid if model error is negligible


G.3 Flavours of variational data assimilation2. Incremental 4D-VAR (quasi-linear)

ref

)()(

)(

)()(Let

)0( :remember

0

0ref

0

ref0

0

ref

xxxM

xMxm

xxm

xmxxxx

xx

t

t

t

tt

t

t

)(

model

ref

ref

model

ref

ref)()(

)()]([

)]()([

)]([)(

)()()(Let

tt

tt

t

t

tt

tt

tt

tt

ttt

xxyH

xHxh

xxh

xhyxxx

Linearisation of m Linearisation of h

Apply to strong constraint cost function

T

ttttttt

J

0

1T0

ref0

1-f

Tb

bref

b

)( ))]([)((21

)( )(21)(

Let

RxMHxmhy

Pxxx

xxx

T

tttt

tt

t

J

tt

0

1T0

1-f

Tb

ref0

)( ))((21

)( )(21)(

)]([)()(Let

RxMHy

Pxxx

xmhyy

The incremental form is exactly linear and is valid if non-linearity is ‘weak’


H. Control variable transforms

variable'control' theis here)]([)()( , , where

),( ))((21 )( )(

21)(

ref0

refbb

ref

0

1T0

1-Tb

xxmhyyxxxxxx

BRxMHyBxxx

tt

nnT

tttt

tt

tJ

Advantages of the new control variable:•No B -matrix to worry about•Better conditioning of the problem

In practice we design U and B follows (the implied B = U U T)The design of U is called ‘background error covariance modelling’

T

tttt

T

tttt

tJ

tJ

0

1T0

Tb

0

1T0

1-TTb

bb

)( ))((21 )()(

21)(

variable'control' theis here

)( ))((21 )( )(

21)(

,Let

RUMHy

RUMHyUBU

UxUx

IUBUU -1Tsuch that Design


I. The cost function and the ‘BLUE’ formula

)form' covariance' (the ))(()(

get we)()( formulaWoodbury -Morris-Sherman theUsing

)form'n informatio' (the ))(()(

0))(())((

0))()(()(

)(function cost ofmin at 0))()(()(

))()(())()((21)()(

21

))(())((21)()(

21)(

b1T

fT

fba

Tf

1TTf

1T1-f

b1T11T1-

fba

b1T

ba1T1-

f

bab1T

ba1-

f

a

bb1T

b1-

f

bb1T

bbb1-

fT

b

1Tb

1-f

Tb

xhyHHPRHPxx

HHPRRHHPHRHP

xhyRHHRHPxx

xhyRHxxHRHP

xxHxhyRHxxP

xxxxHxhyRHxxP

xxHxhyRxxHxhyxxPxx

xhyRxhyxxPxxx

x

x

JJ

J


J. A single observation exampleAnalysis increment of the assimilation of a direct observation of one variable.

),()(

),'()'()'(

))(()],([

0010

))((

0010

)0 0 1 0(

0010

))(()(

f2

bfba

b1

f2

f

b

1

f2

f

b1T

fT

fba

rrPry

rrPrr

ryrrP

ry

y

y

y

xxx

xP

xPP

xhyHHPRHPxx

Obs of atmospheric pressure →


K.1 Forecast error covariance statistics

• In data assimilation prior information often comes from a forecast.

•Forecast error covariance statistics (Pf) specify how the forecast might be in error

εf = xf – xtrue, Pf = <εf εfT>.

•How could Pf be estimated for use in data assimilation?

• Analysis of innovations (*).• Differences of varying length forecasts (‘NMC method’).• Monte-Carlo method (*).• Forecast time lags.

•Problems with the above methods.

•A climatological average forecast error covariance matrix is called ‘B’.

332313

322212

312111

f or

BP


K.2 Analysis of innovations We don’t know the truth, but we do have observations of the truth with known error statistics.

Definition of observation error : y = ytrue + εy = h(xtrue) + εy

Definition of forecast error : xtrue = xf – εf

Eliminate xtrue : y = h(xf – εf) + εy ≈ h(xf ) - Hεf + εy

‘Innovation’ : y - h(xf ) ≈ εy - Hεf

LHS (known), RHS(unknown)

Take pairs of in-situ obs whose errors are uncorrelated (for variable v1, posn r and v2, r+Δr) y(v1,r) - xf (v1,r) ≈ εy(v1,r) - εf(v1,r) y(v2,r +Δr) - xf (v2,r +Δr) ≈ εy(v2,r +Δr) - εf(v2,r +Δr)

Covariances<[y(v1,r) - xf (v1,r)] [y(v2,r +Δr) - xf (v2,r +Δr)]> = <[εy(v1,r) - εf(v1,r)] [εy(v2,r +Δr) - εf(v2,r +Δr)]>

= <εy(v1,r) εy(v2,r +Δr)> - <εy(v1,r) εf(v2,r +Δr)> - <εf(v1,r ) εy(v2,r +Δr)> + <εf(v1,r) εf(v2,r +Δr)> ↑ ↑ ↑ ↑ Obs error covariance Zero (obs and forecast errors Forecast error covariance between (v1, r) and (v2, r+Δr) uncorrelated) between (v1, r) and (v2, r+Δr) zero unless v1=v2 and Δr=0 (one particular matrix element of Pf or B)<> average over available observations and sample population of forecasts


K.3 Monte-Carlo method (ensembles) •N members of an ensemble of analyses.•Leads to N members of an ensemble of forecasts.•The ensemble must capture the errors contributing to forecast errors.

• Initial condition errors (forecast/observation/assimilation errors from previous data assimilation).•Model formulation errors (finite resolution, unknown

parameters, …).•Unknown forcing.

•Can be used to estimate the forecast error covariance matrix, e.g.•Pf ≈ < (x-<x>) (x-<x>) T > = 1/(N-1) ∑i=1,N (xi - <x>) (xi - <x>)T

•Problem: for some applications N << n.•n elements of the state vector (in Meteorology can be 107).•N ensemble members (typically 102).•Consequence – when Pf acts on a vector, the result is forced to

lie in the sub-space spanned by the N ensemble members.

Ensembles

t

x

N

iii

N

iii NN 1

T

1

Tf )(

11

11 xxxP


L. Hybrid data assimilationVariational data assimilation (with B-matrix)•B is full-rank•B is static

Ensemble data assimilation

•Sample Pf is flow dependent•Sample Pf is low rank

xxxxU

UU

RUUMHy

xxx

UxUxUxUx

N

NnNn

T

ttttt

J

1E

V

EVE

V

0

1TEEVV0

TEb

ETVb

V

EV

Eb

EEb

EEEVb

VVb

VVV

)onassimilati data al variationlincrementa standardin as(

, , ,

)( ])[)((21

)()(21 )()(

21)(

, , , Suppose

Comments and Questions

!?

variational data assimilation and forecast error statistics

Documents