Real-time forecasting for the Antarctic: An Real-time forecasting for the Antarctic: An evaluation of the Antarctic Mesoscale Prediction evaluation of the Antarctic Mesoscale Prediction

System (AMPS)System (AMPS)

David H. Bromwich1, Andrew J. Monaghan1, Kevin Manning2 and Jordan G. Powers2

1-Polar Meteorology Group, Byrd Polar Research Center, The Ohio State University,

Columbus, Ohio

2-Mesoscale and Microscale Meteorology Division, National Center for Atmospheric Research,

Boulder, Colorado

1. Evaluating AMPS: 1. Evaluating AMPS: OverviewOverview

• Statistical analysis of AMPS performed for 2-year period (Sep 2001-Aug2003)

• Surface:• 2-m temperature• Surface pressure• 3-m winds

• 500 hPa• Temperature• Geopotential Height• Winds• Dewpoint temperature

• Diurnal cycle at Relay Station

• Soundings at McMurdo

• Comparison of model skill at 3.3-km vs. 10-km

• Effects of implementing a new nudging upper boundary condition

1. Evaluating AMPS: 1. Evaluating AMPS: Surface Performance Surface Performance

(12-36 hr forecasts)(12-36 hr forecasts)

2-m

Tem

per

atu

re3-

m W

ind

Sp

eed

Su

rfac

e P

ress

ure• Surface pressure and near-

surface temperature at most sites have correlations of r>0.95 and r>0.75, respectively, and small biases.

• Surface wind speeds reflect the complex topography and generally have correlations between 0.5-0.6, and positive biases of 1-2 m s-1.

1. Evaluating AMPS: 1. Evaluating AMPS: Surface Performance Surface Performance (12-36 hr forecasts below 500 m)(12-36 hr forecasts below 500 m)

2-m

Tem

pC

lou

d %

Wn

d S

pd

Sfc

Pre

s

• Correlation coefficients generally consistent throughout annual cycle

• Temperature bias goes from negative in summer to positive in winter – likely because wind speed, which is slightly too fast, is eroding surface inversion in winter

1. Evaluating AMPS: 1. Evaluating AMPS: Diurnal Cycle at Relay Diurnal Cycle at Relay Station, January 2003 Station, January 2003

(12-36 hr forecasts)(12-36 hr forecasts)

2-m

Tem

pW

nd

Sp

dS

fc P

res

Wn

d D

irD

aily

C

ycle

• Diurnal cycle is reproduced with good skill

• Surface pressure bias likely due to uncertainties of reducing pressure at model elevation to station elevation

• the near-simultaneous timing of the peak wind speed with the maximum daily temperature suggests that the behavior of the near-surface winds is due to momentum transfer as the stable boundary layer heats up and higher momentum air aloft mixes downward. Opposite behavior found at coastal stations.

PMM5=solidAWS=dashed

1. Evaluating AMPS: 1. Evaluating AMPS: 500-hPa Performance 500-hPa Performance

(12-36 hr forecasts)(12-36 hr forecasts)

Tem

per

atu

re W

ind

Sp

eed

Geo

pot

enti

al H

gt D

ewp

oin

t T

emp

• High correlations for temperature, geopotential height, and wind speed.

• Biases are typically small, and either +/- (not systematic)

• Lower skill in dewpoint forecasts – may be due to low absolute humidity at 500 hPa, which makes small errors look big.

1. Evaluating AMPS: 1. Evaluating AMPS: McMurdo Soundings McMurdo Soundings

(12-36 hr forecasts)(12-36 hr forecasts)

Temperature

Wind Speed Dewpoint Temp

Geopotential Hgt

• Correlations generally high throughout atmosphere.

• Temperature biases are small and become generally warmer and slightly positive near the tropopause (~300 hPa).

• Winds are captured with the best skill in the free atmosphere – decreases near surface due to topography

0.2

0.4

0.6

0.8

1

1.2

1.4

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

a) 10-km b) 3.3-km c) 3.3-km /10-km

1. Evaluating AMPS: 1. Evaluating AMPS: 10-km vs 3.3-km domain 10-km vs 3.3-km domain

(24-36 hr forecasts)(24-36 hr forecasts)

Ratio of 3.3-km standard deviation to 10-km standard deviation (normalized)

Spectral Density vs frequency for wind speed at 2 AWS sites near McMurdo

Conclusion – The 3.3-km domain is more sensitive and more accurate than the 10-km domain

Observed=thick gray3.3-km = solid black10-km = dashed black

2. Implementing the Nudging UBC in AMPS: 2. Implementing the Nudging UBC in AMPS: OverviewOverview

• AMPS switched from its old UBC (rigid lid at 100 hPa) to its new UBC (nudging UBC at 50 hPa) in May 2003

• Here we compare three months of statistics from two winter periods:1. JJA 2002 (“Before” new UBC)2. JJA 2003 (“After” new UBC)

• The Correlations, RMSEs, and biases of the following fields are examined:• Column Temperature• Column Wind Speed• 150-hPa Temperature• 150-hPa Wind Speed• Surface Pressure

• The statistics are calculated for the AMPS 30-km grid versus radiosonde and surface observations, and unless otherwise noted, are valid for the 36-60 hr forecasts.

2. Implementing the Nudging UBC in AMPS: 2. Implementing the Nudging UBC in AMPS: Temperature and Wind Speed Soundings Temperature and Wind Speed Soundings

(36-60 hr, averaged over all stations)(36-60 hr, averaged over all stations)

•In general, correlations and RMS errors are improved throughout model column

•Temperature biases reduced (become cooler) in top layers of model

•Winds become stronger throughout column

2. Implementing the Nudging UBC in AMPS: 2. Implementing the Nudging UBC in AMPS: Surface Pressure Surface Pressure (36-60 hr)(36-60 hr)

•Not much change in correlations – performance very high before/after new UBC

•RMS errors are improved at all stations

•Biases reduced at 15 of 16 stations

Before After

2. Implementing the Nudging UBC in AMPS: 2. Implementing the Nudging UBC in AMPS: Performance vs. Forecast Hour Surface PressurePerformance vs. Forecast Hour Surface Pressure

(averaged over all stations)(averaged over all stations)

•RMS errors improved by about 1 hPa throughout forecast

•Improvement to bias in later forecast hours

Summary:Summary:

•Surface pressure and near-surface temperature at most sites have correlations of r>0.95 and r>0.75, respectively, and small biases.

•Surface wind speeds reflect the complex topography and generally have correlations between 0.5-0.6, and positive biases of 1-2 m s-1.

•In the free atmosphere, r>0.95 (geopotential height), r>0.9 (temperature), and r>0.8 (wind speed) at most sites.

•The 3.3-km domain is more sensitive to spatial and temporal changes in the winds than the 10-km domain, which represents an overall improvement in forecast skill especially on the windward side of the island where the Williams Field and Pegasus runways are situated, and in the lee of Ross Island, an important area of mesoscale cyclogenesis.

•New Nudging UBC increases in the skill of temperature and wind speed forecasts throughout the atmospheric column. Surface pressure bias and RMSE are reduced as well

•The new UBC reduces unrealistically high vertical motions throughout the column

3. AMPS Case Study: 3. AMPS Case Study: Experimental DesignExperimental Design

• Ran 2 AMPS forecasts for 12Z October 02, 20031. Old-100 (Rigid Lid, Top = 100 hPa)2. New-50 (Nudging UBC, Top = 50 hPa)

• Look at differences in cross sections of vertical motion on AMPS 30-km domain

3. AMPS Case Study: 3. AMPS Case Study: Synoptic SituationSynoptic Situation

Hour 60 Hour 66 Hour 72

• An upper level ridge approaches West Antarctica.

• Upward(red)/Downward(blue) motion becomes enhanced

3. AMPS Case Study: 3. AMPS Case Study: Vertical MotionVertical Motion

Hour 60 Hour 66 Hour 72

Rothera

PossessionIsland

Mt Siple

Byrd

Theresa

Elizabeth

Uranus Glacier

Ski-Hi

Halley

Novolazarevskaja

Relay StationMawson

Mirnyj

Casey

Dumont D’Urville

TerraNovaBay

Dome CElaine

Gill

Marilyn

McMurdo

WhitlockManuela Ferrell

South Pole

Wed

dell

Sea

Ross Sea

Bellin

gsh

ausen

Sea

Amundsen SeaRoss

IceShelf

Ronne

-Fi

lchn

er

Ice

Shelf

AmeryIceShelf

South Pacific Ocean

So

uth

Ind

ian

Oce

an

South Atlantic Ocean

Unused slides to follow:Unused slides to follow: