reanalysis Nathan Eizenberg , Chun-Hsu Su, Peter Steinle, Drte Jakob, - - PowerPoint PPT Presentation

Evaluation of Australian regional atmospheric reanalysis

Nathan Eizenberg, Chun-Hsu Su, Peter Steinle, Dörte Jakob, Paul Fox-Hughes

Bureau of Meteorology http://www.bom.gov.au/research/projects/reanalysis/

BARRA – Bureau of Meteorology Atmospheric high-

resolution Regional Reanalysis for Australia (1990-2017)

BARRA-R (12 km forecast model, 36 km 4dVAR) BARRA-xx (1.5 km)

• Dynamical downscaling of BARRA-R

BARRA – Bureau of Meteorology Atmospheric high- resolution Regional Reanalysis for Australia (1990-2017)

Domains of BARRA-R (assimilation & hindcast system) and BARRA-xx (dynamical downscaling systems). BARRA-R system is based on UK Met Office UERRA reanalysis system.

BARRA-PH BARRA-AD BARRA-SY BARRA-TA

Nested in ERA-Interim OSTIA SST/SICE boundary

Heatwave in Sydney, New South Wales

"NSW recorded a statewide average maximum temperature of 35.34 °C during January, 3.66 °C above the historical average. This is the third warmest January since records began in 1910, and the warmest for NSW since 1939." – Bureau of Meteorology Monthly Climate Summary

Daily maximum temperature on 18 January 2013. The locations of weather stations are overlaid over the AWAP (Jones et al., 2007) grid.

Hottest at 46.5°C Penrith Lakes 5 km 80 km 12 km 1.5 km

Cyclone Yasi, 3 February 2011

"Yasi is one of the most powerful cyclones to have affected Queensland since records commenced." – Bureau of Meteorology

6-hourly rainfall accumulation from TRMM/TMPA (satellite, first row), BARRA-R (second row), and ERA-Interim (last row), central pressure track in red

25 km 12 km 80 km

"Yasi is one of the most powerful cyclones to have affected Queensland since records commenced." – Bureau of Meteorology

Cyclone Yasi, 3 February 2011

10 m wind speed forecast at 18 February 2011, 18 UTC from ERA-Interim and BARRA-R. Last plot shows the instantaneous 10 m wind gust (10 min) from BARRA-R.

Source: Bureau of Meteorology

Screen temperature (2011)

Evaluation of T+6h forecast valid at 6 and 18 UTC; thus using the observations outside the analysis window. Boxplots showing the distribution of metrics across stations in the BARRA-SY domain.

𝐹[𝑛]−𝐹[𝑝]

10 m wind speed (2011)

𝐹[𝑛]−𝐹[𝑝]

Evaluation of T+6h forecast valid at 6 and 18 UTC; thus using the observations outside the analysis window. Boxplots showing the distribution of metrics across stations in the BARRA-SY domain.

Mean annual precipitation, fraction of light rain days (1-10 mm), and fraction of very wet days (> 50 mm)

Precipitation (2010-15)

5 km 25 km 80 km 12 km

Fractions of very wet days from AWAP, BARRA-R, BARRA-SY, and ERA-Interim.

Precipitation (2010-15)

AWAP BARRA-R BARRA-SY ERA-Interim

Frequency histogram of daily precipitation. Comparison against AWAP.

Precipitation (2010-15)

BARRA-R BARRA-SY ERA-Interim

Pressure level wind (2010-2013)

Comparisons of BARRA-R analysis, ERA-Interim analysis and BARRA-R forecast (T+6h forecast) against GRASP pilot balloon and radiosonde observations valid at 0 and 12 UTC (Ramella Pralungo, et al. 2014). Boxplots showing the distribution of metrics across sites.

RMSD [m/s]

Conclusions

• Completed 10-year reanalyses BARRA-R, BARRA-SY, BARRA-

TA, BARRA-PH

• BARRA show realistic details at 12 km and 1.5 km resolution

at surface or near-surface

• BARRA appears more realistic during high impact weather
• BARRA shows better agreement with point-scale
• bservations than ERA-Interim, even after areal-to-point

interpolation

• Downscaled reanalyses (BARRA-xx) improve upon BARRA-R
• Negative biases in wind speed persist – under investigation

and to be addressed via post-processing

• Difficult to discern differences between BARRA-R and ERA-

Interim for upper air temperature and wind.

Drew Whitehouse@NCI/ANU