Downscaling tropical cyclones from global re-analysis and scenarios: Statistics of multi-decadal variability of TC activity in E Asia Hans von Storch,

Downscaling tropical cyclones from global re-analysis and scenarios:

Statistics of multi-decadal variability of TC activity in E Asia

Hans von Storch, Frauke Feser and Monika BarcikowskaInstitute for Coastal Research, GKSS Research Center, Germany

clisap-Klimacampus, University of Hamburg, Germany

• We have implemented the dynamical downscaling approach for E Asian marine weather.

• The key question is – will we master the description of typhoons?

• Simulation of 1948-today conditions• Simulation of possible future: An IPCC A1B scenario, 1959-

2100

Reference data: “best track”

• Provided by different weather services

• JMA

• CMA

• Joint Tropical Cyclone Warning Center JTWC

• Constructed on the fly

• Strong inhomogeneities in the course of tim (Ren, 2010) – mostly due to commencing and ending of reconnaissance flights, availability of satellites

JMA best track statistics

2010 2010

Typhoon 195313 (TESS)

1953-09-18 00:00

1953-09-27 18:00

Largest drop in BT core pressure:

93 hPa in 6 hours

Typhoon 195307 (NINA)

1953-08-08 06:00

1953-08-19 00:00Largest drop in core pressure August 11-1265 hPa in 6 hours

JMA Best Track

JMA Best Track

Barcikowska, unpublished

IBTrACS- International Best Track Archive for Climate Stewardship merging few different best track data sets,

JTWC is converted from 1 min to 10 min.

NOAA - Blended Sea Winds - global 0.25° grid, 6 –hrly ocean surface vector winds, based on wind speed retrievals from satellitesequipped with passive SSMI (special sensor microwave imager + active scatterometers )

IFREMER - Near Real Time Blended Surface Winds - 0.25° global oceans grid, 6 –hrly ocean surface vector winds, derived from satellite data blended to ECMWF analyses)

BTD=JMA BTD=JMA

First conclusion1) “Best track” data affected by inhomogeneities related to changing observational and analytical procedures in the course of time (see also Ren, 2010)

2) Different agencies produce different “best track” data.

3) Comparison with satellite-based data analyses reveal further differences. “Best track” data may provide an overestimate.

RCM simulations• CLM regional

atmospheric model• 50 km grid resolution• Two different sets of

parameters for turbulent latent heat flux (normal/high)

• “Reconstructions” – NCEP forcing, incl. spectral nudging (800 km), 1948-today

• “Scenario” – ECHAM5/MPIOM A1B1; also spectral nudging

All tracks in “reconstruction”

Barcikowska, unpublished

^Comparison of best track data (BTD = JMA), TC-reanalysis and CLM simulation

Quickscat based

Case study: wind speeds in NOAA re-analysis, RCM simulation (CLM) and global weather reanalysis (NCEP)

Barcikowska, unpublished

Note: different criteria employed1950 1960 1970 1980 1990 2000 2010

10

20

30

40

50

Number of detected typhoonsJMA Best TrackCCLM "normal"CCLM "high"

1994 - 36 JMA “best tracks”31 tracks in CLM

1998 - 16 JMA “best tracks” 25 tracks in CLM

Interannual variability

Second set of findings:

1) CCLM simulates typhoons.2) Number and interannual variability in CLM similar to „best track“

data set.3) Simulated typhoons too weak (increase of grid resolution to 18

km gives no significant improvement).4) Sensitivity to different formulations of latent heat flux minor

5) Long term trends in CCLM and in „best track“ markedly different.6) In CCLM, some intensification mainly since about 1980.7) In JMA-„best track“, mainly weakening since about 1980.

1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100

10

20

30

40

50

Number of detected typhoonsScenario A1BCCLM "normal"

1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100

900

910

920

930

940

950

960

970

980

hP

a

Minimum typhoon pressure in CCLM simulationsplus linear trends

Scenario A1BCLM "normal"

1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100

20

30

40

50

m/s

Maximum typhoon wind speedin CCLM simulationsplus linear trends

Reconstruction "normal"Scenario A1B

Third set of findings:

1) Number and intensity in reconstructions 1948-2008 increasing.

2) Number in scenario A1B slightly decreasing, while intensity almost stationary. Note – only one scenario.

3) Thus, scenarios can not be seen as “extending” description of past forward in time.

4) Possible explanations- reconstruction not homogenous (skill of NCEP re-analysis is improving), or- change in 1948-2008 not related to main driver in scenario simulation (increasing GHG concentrations)- scenario calculation does not describe influence of elevated GHG concentration properly