Using ensemble data assimilation to investigate the initial

condition sensitivity of Western Pacific extratropical transitions

Ryan D. Torn

University of Washington

Satellite EvolutionET – 48 h ET ET + 48 h

Tok

age

(200

4)N

abi (

2005

)

Effect of Mid-latitude Flow

Harr and Elsberry 2000

Forecast Sensitivity to TC

Klein et al. 2002

Control Simulation TC displaced 250 km SW

Overview

• Want to determine the initial condition sensitivity of Western Pacific ET events

• Use EnKF data assimilation as a tool to answer the following questions about ET events:– What analysis features is the ET forecast most

sensitive to?– Are observations available in the area where the

forecast is most sensitive to the analysis?– Are these results generic or case dependent?

Forecast Sensitivity

cov( , )

var( )b

b

b x

x JJ

x

EnKF offers an alternative way to calculate the sensitivity of a forecast metric (J) to the analysis

using the ensemble of analyses and forecasts:

No tangent linear model necessary, only linear regression!

GFS Forecast of Tokage ET

Courtesy Pat Harr, Naval Postgraduate School

48 hour forecast initialized 12 UTC 19 October 2004

Experiment Setup• WRF model, 45

km resolution, 30 vertical levels

• 90 ensemble members

• observation assimilation every 6 hours– rawindsondes

– ACARS

– cloud track winds

– surface stations

– buoys, ships

– ~10,000 obs.

Tokage 00 hr ForecastSea-Level Pressure 500 hPa Height

Tokage 24 hr ForecastSea-Level Pressure 500 hPa Height

Tokage 48 hr ForecastSea-Level Pressure 500 hPa Height

Tokage Forecast

Tokage Track

Tokage min. SLP

Initialized 12 UTC 19 October

12 Hour Forecast SensitivitySea-level Pressure 500 hPa Height

48 Hour min. SLP Sensitivity500 hPa Height

• Shifting Siberian trough to the east

• Shifting Mongolian trough to the west

• Moving Tokage to the southwest

Min. SLP is increased by:

48 Hour RMS error sensitivity500 hPa Height

RMS error is decreased by:

• Shifting Siberian trough to the east

• Shifting Mongolian trough to the west

• Moving Tokage to the southwest

500 hPa Observations

•Lack of sonde observations in critical region

•Sondes were missing during this cycle

Nabi 00 hr ForecastSea-Level Pressure 500 hPa Height

Nabi 24 hr ForecastSea-Level Pressure 500 hPa Height

Nabi 48 hr ForecastSea-Level Pressure 500 hPa Height

Nabi ForecastInitialized 00 UTC 6 September 2005

48 Hour min. SLP Sensitivity500 hPa Height Sea-level Pressure

48 Hour RMS Error Sensitivity500 hPa Height

• Shifting Chinese trough to west

• Amplification of Siberian ridge

• Shifting downstream trough to the east

RMS error is decreased by:

500 hPa Observations

•Several sondes in the most sensitive regions

•Analysis more confident in trough position, thus less forecast variance.

Summary

• Extratropical Transitions can be a significant predictability problem for NWP models

• Described set of experiments to understand the sensitivity of the ET forecast to analysis features

• Tokage and Nabi results suggest that largest forecast sensitivities are associated with upper-level troughs upstream of TC. Stronger westerlies may lead to more sensitivity.

• Future work will include additional ET events and an assessment of observation impact.