Slide 1 Retrospective analysis of ozone at ECMWF Rossana Dragani ECMWF Acknowledgements to: D. Tan, A. Inness, E. Hólm, and D. Dee R. Dragani, SPARC/IOC/IGACO,

Slide 1

Retrospective analysis of ozone at ECMWF

Rossana Dragani ECMWF

Acknowledgements to:D. Tan, A. Inness, E. Hólm, and D. Dee

R. Dragani, SPARC/IOC/IGACO, Geneva, Jan 2011 © ECMWF 2011

Slide 2

Overview

Reanalysis activity at ECMWF

Science application for reanalysis data

Suitability to assess long-term variability

Ozone analyses in ERA-40 and ERA-Interim

Conclusions and remarks


Slide 3

Reanalysis at ECMWF:

Reanalysis is based on analysis methods developed to provide initial states for numerical weather prediction.

It applies a fixed, modern data assimilation system to multi-year sets of observations of different types, resulting in more uniform analysis quality.


ERA-15: 1979 – 1993ERA-40: 1957 – mid-2002ERA-Interim: 1989 onwards

ORA-S3: 1959 onwardsMACC: 2003 – 2010

ERA-CLIM:European Reanalysis of Global Climate Observations

An EU project to help prepare the next ECMWF reanalysis

ERA-20C: 1900 onwards

Slide 4

The ERA data base: www.ecmwf.int

ERA-Interim reanalysis is continuing in near real-timeProducts are updated monthly


Slide 5

Science applications that rely on reanalysis data

“Observations” for verification and diagnosis- Development of forecast model, climate model; calibration of seasonal

forecasting systems; use of data assimilation increments for identifying model errors

Input data for model applications- for smaller-scales (globalregional; regionallocal), ocean circulation,

chemical transport, …

Study of short-term atmospheric processes and influences- Polar vortex dynamics,…

Providing climatologies Assessment of the observing system

- providing feedback on observational quality, bias corrections and a basis for homogenization studies; contributing to data reprocessing activities

Study of long-term climate variability and trends


Slide 6

ERA-40 + ERA-Interim Mid 1957 to the present

Analyses are physically coherent and consistent with observations

A realistic model can propagate information in poorly observed areas as well as moved forward in time.


Suitability to assess long-term changes:

Geer, Planet Earth Autumn 2004, www.nerc.ac.uk

MIPAS O3 retrievals:

23 Sep 2002 © ESA, 2002.

Ozone fc: D+6 from

17 Sep 2002

Ozone analysis: 12Z 23 Sep 2002

Slide 7

Reanalysis uses a modern, stable and invariant forecasting /assimilation system to reprocess (re-analyse) past observations:

Remove artificial changes introduced by model upgrades

Changes in the observing system and their error characteristics can also produce shifts and spurious trends.

- ERA-Interim used VarBC for all radiances

- ERA-Clim will also use VarBC for all L2 products (O3, WV)


Suitability to assess long-term changes:From archived weather analyses: From a retrospective analysis:

Slide 8

Ozone reanalyses at ECMWF: Ozone has been reanalysed in ERA-40 and ERA-Interim

- MACC also reanalysed ozone for the period 2003-2010.

Long record: ERA-40 until Dec 1988 + ERA-Interim from Jan 1989.

Main differences between ERA-40 (Cy23r4) and ERA-Interim (Cy31r2):

- Assimilation scheme: ERA-40 used 3D-Var; ERA-Interim used 4D-Var

- Data usage

- Variarional Bias Correction (VarBC) for radiances

- Upgrades in the ozone model

- Horizontal resolution: T159 (125km, ERA-40) T255 (80 km, ERA-Interim)

ERA-40 ozone analyses validated by Dethof and Hólm (2004, QJ). ERA-Interim ozone analyses validated by Dragani (2011, QJ submitted).


33333

3 243210 OEQOOOO

O ClccTTcccdt

d

ERA-InterimERA-40

Slide 9

Ozone in ERA-40:


Dethof & Hólm (2004)

Hohenpeissenberg (1983-1987)

JFM JJA

Neumayer (1992-1995)

SO

SondeERA-40

Hilo (1992-1995)

JFM JJA

Slide 10

Ozone in ERA-Interim:


TCO validated against

OMI TCO (TOMS-like);

TCO from Dobson Spectrometers (WOUDC);

TCO climatology created as a 5-year running mean from the NASA merged satellite.

3D O3 analyses (Jan 89-Dec 08) validated against

WOUDC sondes; Satellite data.

V6.2

V5 V2.2

V19

V4V6POAM:

HALOE:

MLS:

SAGE:

Slide 11

Validation of the 3D O3 reanalyses:


30 hPa

Matching criteria:

t 3 hrs.

ERA-Interim analyses were interpolated at

the obs locations.

GOME

SAGE HALOEMLS POAM

Sat-AnSat

Slide 12

Summary of the comparisons: 1989-2008

Lat Lev (hPa) SAGE HALOE MLS POAM

30 - 90 N

5 ±5% [0,+5]% [0,+5]% [-25,+5]%

10 [-10,+5]% [-5,+1]% [-5,+3]% [-25,+10]%

30 [0,+20]% [-2,+20]% [0,+20]% [-10,+20]%

65 [-20,+10]% [-2,+10]% [-20,+10]% [-20,+10]%

30S - 30N

5 ±5% ±5% ±5%

10 ±10% ±5% [-5,+8]%

30 [0,+20]% [0,+20]% [0,+20]%

65 [-20,+30]% [-5,+20]% [+5,+25]%

30 - 90 S

5 [-8,+5]% ±2% [-8,+5]% [-40,+1]%

10 [-10,+5]% [-8,+1]% [-10,+3]% [-40,-5]%

30 [0,+20]% [0,+20]% [0,+20]% [-20,+10]%

65 [-20,+20]% [-2,+20]% [0,+20]% [-20,+20]%


Slide 13

ERA-40 vs. ERA-Interim: Jan 1989 - Aug 200230

-90

N30

S-3

0 N

30-9

0 S

HALOE UARS MLS

403333

ERASATInterimERASAT OORMSOORMSR. Dragani, SPARC/IOC/IGACO, Geneva, Jan 2011 © ECMWF 2011

Slide 14

ERA-40 vs. ERA-Interim:Comparisons with ozone sondes


Pre-GOME(1989-1995)

GOME(1996-2001)

60 – 90 SSep-Oct

30N – 30SJan-Feb-Mar

ERA-InterimERA-40

203 soundings 942 soundings

238 soundings227 soundings

Slide 15

Conclusions

ERA-40 O3 analyses showed general good agreement with

observations, but some problems were seen e.g. at mid-latitudes in winter, and at high latitudes in the SH in spring.

ERA-Interim O3 analyses compare well with observations, and show

departures from SAGE, MLS and HALOE ≤10% in the middle stratosphere, and ≤ 20% in the lower stratosphere.

ERA-Interim better than ERA-40 in the UTLS region before 1996, and over the whole stratosphere afterwards (GOME assimilation).

Planned improvements on ozone:

- Use the Variational Bias Correction (VarBC) scheme with L2 data.

- Revise/improve the assimilation of ozone profiles (e.g. SBUV data)

- Assimilate ozone information from different data type/instruments (ozone-sensitive radiances, L2 ozone data not yet used, e.g. those used for validation).


Slide 16

Remarks:

What can reanalysis deliver?

- Consistent with observations

- Physically coherent

- Complete, with no gaps

- Comprehensive

- Accurate variability and trends

- Meaningful information about uncertainties

Progress towards climate quality requires open access to all input data.

Progress is iterative and needs regular re-processing.


Slide 1 Retrospective analysis of ozone at ECMWF Rossana Dragani ECMWF Acknowledgements to: D. Tan, A. Inness, E. Hólm, and D. Dee R. Dragani, SPARC/IOC/IGACO,

Documents