Progress Towards the Assimilation of Cloud-Affected Radiances at the GMAO

1. Progress Towards the Assimilation of Cloud-Affected Radiances at the GMAO Will McCarty1 and Jianjun Jin2 1 NASA Goddard Space Flight Center 2 Goddard Earth Sciences Technology and Research/USRA Global Modeling and Assimilation Office JCSDA Workshop June 5, 2013

2. Cloud Efforts at the GMAO • Microwave (J. Jin) • Preparations to implement all-sky microwave assimilation in a method consistent with: • NCEP-developed methodology • GMAO systems (namely, GEOS-5 model background fields and physics) • Infrared (McCarty) • Focusing on expansion of GSI towards assimilation of cloud-affected infrared radiance measurements using a graybody assumption

3. Cloudy Radiance Assimilation at the GMAO Microwave • In an effort to prepare for the launch of the Global Precipitation Mission (GPM), efforts are underway to investigate the assimilation of microwave imagery • Efforts to assimilate TRMM / Microwave Imager (TMI) brightness temperatures (product 1B11) are underway • Clear-sky assimilation with GSI/GEOS-5 successful • Efforts underway to expand TMI towards all-sky assimilation • Advance efforts towards all-sky microwave radiance assimilation • Expand NCEP methodology (M.-J. Kim et al.) that has focused on microwave sounding towards microwave imagery • Include modifications to utilize fields consistent with the GEOS-5 backgrounds in the state and control vectors (e.g., separate cloud liquid and ice fields)

4. TMI Clear-Sky Assimilation • Initially, clear-sky observations were assimilated • Not much was performed other than consistency in O-F calculation & data counts (left, 19.35 GHz Vertical Polarization • Not much further investigation as observations were expected to have minimal impact • Recent reanalysis sensitivity studies indicate a system-wide sensitivity to SSMI • Studies have shown that the upcoming MERRA2 system has a precipitation signal correlated to SSMI observations • Possibly related to GMAO implementation of qoption=2 & GEOS-5 sensitivity to q increments ~850 hPa 16 – 21 Mar 2012

5. Infrared Assimilation • In GMAO forward processing, infrared radiances are assimilated from IASI, AIRS, and HIRS • Heritage “multi”-spectral sounders like HIRS (~ 18 channels) and the GOES Sounder are being phased out • The US HIRS instruments replaced by CrIS from NPP onward (hyperspectral – 1297 ch total, 399 for DA) • The final European HIRS launched on MetOp-B. MetOp-C will only fly IASI (hyperspectral – 8461 ch, 616 for DA) • No Sounder in US GEO beginning w/ GOES-R • Hyperspectral sounding potentially in GEO in a number of future longitudes

6. Observation volumeJanuary 1977 to present Number of observations considered for assimilation Before thinning, QC IASI Reduction of observations heavily due to presence of clouds in observations Observations processed per 6h 1979 − 2011 AIRS Number of observations used for assimilation After thinning, QC Observations used per 6h 1979 − 2011 IASI AIRS

7. How are Clouds Handled in GSI • Cloud screening is a two-step process • Retrieve a cloud height • This is done via a minimum residual method (Eyre and Menzel 1989) • Compare cloud height against transmittance profile • If layer-to-top of atmosphere transmittance of a channel at the retrieved cloud height is greater than 2% reject the channel • For channels most-sensitive to the surface, this rejects ~80% of these data.

8. Further Exploiting IR Data • To further exploit IR data, the next step is to include some characterization of clouds in the analysis

9. Clear IR Measurement = Surface + (Atmospheric Layers) Cloudy IR Measurement = Cloud Top + (Atmospheric Layers above cloud) Retrieved Cloud Height

10. Clouds in the Infrared • In the cloud height retrieval, a cloud fraction, N, is also solved • Under the graybody assumption, the partially cloudy observation can then be considered for a single, fractional cloud as: • In the GSI, we can then restructure the H operator to include the Cloud Height and Cloud fraction to allow for a partially cloudy forward operator (and also partially cloudy Jacobians) • In CRTM, cloud structure without scattering has the potential to provide needed cloud information Partially Cloudy IR Measurement = N * Cloudy IR Measurement + (1 – N) * Clear IR Measurement

11. Considering the O-Fs versus cloud fraction, it is seen that the O-Fs are closer, but the cold bias is, as expected, amplified for higher (colder) clouds • The accuracy of the calculated cloudy radiance is fundamentally dependent on the accurate retrieval of cloud height and fraction Obs minus Forecast (clear) Obs minus Forecast (cloudy)

12. Cloudy Infrared Radiance Assimilation within the GSI • Jacobians are adjusted to move sensitivity from below cloud to cloud surface • Single footprint assimilation shows that the system is drawing to the retrieved cloud top • Magnitude is inflated due to low observation errors. • Error in CTP will result in an erroneous O-F, which then can negatively impact the analysis • To compensate, CTP is allowed to vary in the minimization as a control variable Uncontaminated Including Cloud Cloud Top

13. Observation-Centered Control Variables • Current GSI implementations consider control variable only in terms of grids (2D & 3D) and channel-by-channel bias predictors • Bias prediction coefficients are of the dimension [5,number of channels] • each satellite channel on each instrument has its own set of predictors (i.e. MetOp_AMSU-A channel 8 will have the same set of five coefficients across every footprint globally • Observation-Centered control variables • consider a control variable at a footprint location over all channels measured at that point • Dimension dynamic -> any number of observation-centered control variables can be appended to the control vector

14. Observation-Centered Control Variables • Once developed, the functionality was expanded to CTP • Cloud Fraction still considered constant and set as the retrieved value • Jacobians • In addition to modified TB/T(p), TB/qv(p), etc., the minimization now incorporates the CTP Jacobian, TB/pcld. • TB/pcldcan be directly differentiated from the radiative transfer equation (i.e. the appendix of Li et al. 2001) • Background error for CTP

15. Background error for CTP • Background error for CTP (BCTP) was considered first in a single-footprint case: • Initial CTP – 624 hPa • Initial N – 0.968 • Consider behavior of two values of BCTP compared to clear-sky observations only and a static CTP (no variational CTP) • BCTP = 50 hPa and 5 hPa

16. Background error for CTP Clear Cloudy Static CTP Cloudy varCTP

17. Background error for CTP • Variational CTP acts as a “sink”, as a function of BCTP • As the bkg error is increases, the cloud signal is absorbed into the CTP variable • the solution approaches clear-only result • As bkg error is decreased, result approaches static CTP • Expected as CTP is tightly constrained to retrieved guess • This is only for a single footprint. How does the analysis respond to a full suite of observations • Since only CTP is varying, only consider cloudy IR if 1.0 > N > 0.9 -> higher confidence in cloud height for opaque clouds

18. CTP Increments B = 50 hPa B = 5 hPa • On a full analysis, large CTP background error values had a negative impact on the convergence of the minimization • Consistent with Tony McNally’s effort @ ECMWF • One potential issue involved with this is the use of a single B value • In this study, the real sensitivity we are adding is the temperature of the cloud top • The T/pc @ 250 hPa very different than T/pcld at 850 hPa

19. CTP Increments B = 50 hPa B = 5 hPa B = f(CTP) Error Model

20. Observation Selection Criteria • Cloud-affected AIRS observations are read as a separate data stream • Instead of type ‘airs’, type ‘airscld’ • Bias correction is consistent between clear and cloudy • Data counts “doubled” in that clear are selected by standard criteria, and cloudy are selected as coldest window channel in thinning box • Not ideal, as this method would be biased towards multilayer clouds, but simple • Above clouds, observations unaffected • double the number of observations in stratosphere • Observation errors of airscldobs will be relatively deweighted in thinning box to clear obs

21. Observation Counts • Figure shows the assimilated observation counts for AIRS Ch. 123 (12 m) • More clear-sky observation accepted in CLD experiment • 85% of additional clear-sky observations correspond to less rejections due to cloud screening (black vs. orange) • Additional 184% observations are assimilated (sum of red vs. orange) • This is for a window channel, where overall rejection rates are large AIRS Clear Sky (CTL) AIRS Clear Sky (CLD) AIRS Cloudy Sky (CLD) 25 Mar - 16 Apr 2012

22. Observation Counts • Though the coldest footprint is chosen, the distribution peaks at ~700 hPa • IR-derived CTP distribution of the atmosphere is typically bimodal, w/ a lack of mid-tropospheric clouds (700-400 hPa) • Distribution tendency towards high clouds largely affected by low fraction/transmissive cirrus AIRS Clear Sky (CTL) AIRS Clear Sky (CLD) AIRS Cloudy Sky (CLD) 25 Mar - 16 Apr 2012

23. Variance of the Analysis Differences Std. Dev (T(CLD) – T(CTL)) @ 850 hPa Used Cloudy Obs for AIRS Ch. 123 (12 m) • The analysis differences show the most variation where the additional cloudy observations are added • Shown by the distribution of assimilated Cloud-affected AIRS observations on the left • Areas driven by common, high weight observations (i.e. sondes over North America and Europe) show little variation Temperature (K) 1-16 Apr 2012

24. Observation Characteristics AIRS Clear Sky (CTL) AIRS Clear Sky (CLD) • Analysis of the clear-sky AIRS observations show that the BC is larger in CLD • More Accurate low-levels? Or forcing O-F to be less negative -> less “cloudy” -> more observations accepted? AIRS Clear Sky (CTL) Mean: 0.03 K AIRS Clear Sky (CLD) Mean: 0.25 K

25. Mean Temperature Analysis DifferenceCLD - CTL |T(CLD) – T(CTL)| @ 850 hPa |T(CLD) – T(CTL)| @ 700 hPa Temperature (K) Temperature (K) 1-16 Apr 2012

26. Mean Temperature Analysis DifferenceCLD - CTL |T(CLD) – T(CTL)| @ 500 hPa |T(CLD) – T(CTL)| @ 300 hPa Temperature (K) Temperature (K) 1-16 Apr 2012

27. 4-Day Forecast Verification |T(CTL, t=96h) – T(CTL,t=0)| - |T(CTL, t=96h) – T(CTL,t=0)| 600hPa 850 hPa • Red -> CLD forecast is closer to verifying analysis than CTL (CLD is improved) • Blue -> CLD forecast is further from verifying analysis than CTL (CLD is degraded) • *VERY* limited sample 1-15 April 2012, 00Z only

28. 4-Day Forecast Verification |Z(CTL, t=96h) – Z(CTL,t=0)| - |Z(CTL, t=96h) – Z(CTL,t=0)| 300 hPa 500 hPa • Red -> CLD forecast is closer to verifying analysis than CTL (CLD is improved) • Blue -> CLD forecast is further from verifying analysis than CTL (CLD is degraded) • *VERY* limited sample 1-15 April 2012, 00Z only

29. Final Remarks • This effort is hampered by the initial guess of CTP • Some effort has been in place to improve this, but the minres algorithm in its current form isn’t accurate enough • Interpolation between layers is likely necessary, but initial implementation caused bias correction to run amok (could have been a bug) • Inclusion of co-located imager data could likely be used to improve QC • Sub-gridscale info can help constrain obs that don’t violate the graybody single-layer cloud assumption • “Sink” control variables are nearly ready to be handed off (summer goal) • Effort to implement in a general manner so that it can be readily expanded to other variables

30. Final Remarks • Plans to expand beyond the graybody assumption to a more in-depth all-sky methodology are on the horizon • Particularly focusing on the analysis of thin cirrus • B. Kahn’s efforts @ JPL have shown some promise in the retrieval of thin ice cloud properties