Impact of EOS MLS ozone data on medium-extended range ensemble forecasts Jacob C. H. Cheung 1, Joanna D. Haigh 1, David R. Jackson 2 1 Imperial College.

Impact of EOS MLS ozone data on medium-extended range ensemble forecasts

Jacob C. H. Cheung 1, Joanna D. Haigh1, David R. Jackson2

1Imperial College London 2Met Office

Overview

• Motivation

• Methods

• Experimental period selected

• Impact on tropospheric forecasts

• Is there significant improvement?

Motivation

Comparison of Forecast RMS differences: Forecast against analysis for extended global index

-2

-1.5

-1

-0.5

0

0.5

1

1.5

NH Tropics SH

Location

FC

RM

S (

%)

dif

fere

nces

SPARC climatology

ECMWF Ozone Field

Assimilation of SBUV only

Assimilation of SBUV andEOSMLS

Source: Mathison et al. 2007

Increase in forecast skill

Decrease in forecast skill

Motivation

• Full tropospheric response to stratospheric thermal forcing is a two-stage process

• Improving representation of ozone will possibly improve medium-extended forecasts

Source: Simpson et al. 2009

Forecast range considered by Mathison et al.

Aim

Is the representation of stratospheric ozone important in medium-extended range tropospheric forecasts?

Methods

• Met Office Global and Regional Ensemble Prediction System (MOGREPS)

• Resolution: N216L85• 31-day free running forecast• 24 ensemble member

Run Id LiShine (Control) MLS (Experiment)

Dataset Li and Shine 95 EOS MLS

Ozone 5-year monthly mean zonal mean

Monthly mean zonal mean correspond to the chosen forecast date

Case Studies Northern winter; Southern winter, spring

Northern spring (March 2011)

Experimental period selected – March 2011

Source: NASA/Goddard

Ozone profiles

Results – Stratospheric temperature

MLS-LiShine

MLS forecast errorLiShine forecast error

- General reduction in temperature forecast errors with MLS ozone

- Temperature anomaly between runs is significant in stratosphere

- Not much change in temperature in troposphere

Results – Tropospheric zonal wind

MLS-LiShine

MLS forecast errorLiShine forecast error

- Tropospheric zonal wind anomaly between runs is weak compared to individual forecast errors

- Response is statistically significant in some area

Results - SLP

[hPa] [hPa]

MLS-LiShine

NH SH

Temperature RMSE (10hPa)

NH

TR

SH

Horizontal wind RMSE (250hPa) GPH RMSE (500hPa)

NH

TR

SH

Horizontal wind RMSE (250hPa) GPH RMSE (500hPa)

NH

TR

SH

Summary

- Performed a case study in which the MLS ozone profile is much superior that of LiShine

- Zonal wind and temperature response is sensitive to the ozone climatology in current NWP systems (in agreement with other ST coupling studies)

- Tropospheric forecast errors are dominated by ensemble spread in medium-extended range forecasts

-> In our experiments, using monthly mean zonal mean EOS MLS ozone data does not significantly improve medium-extended tropospheric forecasts

Questions?

Horizontal wind RMSE (50hPa) Temperature RMSE (50hPa)

NH

TR

SH