Assimilation of Cloudy AMSU-A Microwave Radiances in 4D-Var Una O’Keeffe Thanks to Martin Sharpe and Stephen English IPWG Workshop, Melbourne October 2006 © Crown copyright 2004
Cloud liquid water incrementing operator Assimilation set up Overview Motivation AMSU-A 23GHz and 31Ghz Cloud liquid water incrementing operator Assimilation set up Assimilation results © Crown copyright 2004
Motivation Cloud liquid water has large impact on microwave radiances Currently low peaking AMSU-A channels are not assimilated if significant water is present Significant data gaps due to cloud AMSU-A window channels contain information on liquid water which is not currently exploited Step towards assimilation of AMSR high resolution cloud and precipitation-affected radiances © Crown copyright 2004
Information on cloud liquid water RTTOV8 with clw emission RTTOV8 without clw emission NOAA-16 Obs 23GHz 31GHz © Crown copyright 2004
O-B Stats for IRclear RTTOV – 31GHz © Crown copyright 2004
O-B Stats for MWcloudy RTTOV – 31GHz Mwcloudy (failed rain test) – with active cloud in rttov7 © Crown copyright 2004
Cloud Incrementing Operator Total moisture analysis variable used in 4D-Var Need cloud incrementing operator that relates liquid water and specific humidity to the total water control variable Cx+ = Cx + KCw’ Cx = model state (q,qcl,qcf,cf) Cw’ = analysis increment (T’,p’,qT’) K = incremental transform variable between control variable space and model parameter space (uses linearised physics). Sharpe,2005 © Crown copyright 2004
Currently formulated with full field total water 1D-Var Preprocessor Currently formulated with full field total water Up to 8% of solutions are rejected in 1D-Var with this approach Data volume in 3D-Var is not reduced but is biased away from cloudy areas, giving negative impact © Crown copyright 2004
Assimilation Experiment Set Up Configuration: 3DVar, Dec05 four week period 10 day run to generate clear air bias corrections Cloudy obs 23+31GHz assimilation trial assimilate NOAA-16 AMSU-A 23GHz and 31GHz extra-tropics sea only for all cloud conditions except for where rain flag is on © Crown copyright 2004
Analysis Increments © Crown copyright 2004
Analysis Increments © Crown copyright 2004
Impact on large scale fields fit to analysis NH | TROPICS | SH 50hPa height 850hPa humidity Most fields improved in SH 500hPa and 250hPa temp © Crown copyright 2004
Fit to observations 31GHz © Crown copyright 2004
Fit to observations 31GHz © Crown copyright 2004
Bias Correction of Cloudy Data…??? For this test, used N16 HIRS to define ‘clear air’ and bias corrected clear air data Operationally, also want to use N15, N17, N18 Options: Bias correct clear air data only – ignores large cloudy biases Bias correct all data – may degrade clear air assimilation © Crown copyright 2004
Current Status Testing different bias corrections Investigations of 1D-Var rejections indicated issue with high retrieved LWP on the first iteration causing failures. A fix is now in place Operational implementation planned for early 2007 Plans SSMI/SSMIS AMSR AMSU-B + ice incrementing operator © Crown copyright 2004
Some significant changes to lower level humidity cf analysis Summary Assimilation of cloudy AMSU-A 23GHz and 31GHz data gives consistent positive impacts in SH and tropics Some significant changes to lower level humidity cf analysis Cloud fields improved Unresolved issues with bias correction © Crown copyright 2004
Questions? © Crown copyright 2004