Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Advances towards an efficientmulti-instrument assimilation of IR cloud-
cleared radiances in a global data assimilation and forecast framework
Oreste Reale, Niama Boukachaba, Erica McGrath-Spangler, Manisha Ganeshan(GESTAR/USRA and NASA/GMAO)
Chris Barnet (STC) Will McCarty, Ron Gelaro (NASA/GMAO)
1
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Outline• Past work – Demonstrating the superiority of cloud-cleared radiances
(CCRs) from AIRS and that the operational use of AIRS clear-sky radiances was suboptimal
• Porting the cloud-clearing algorithm to NASA Center for Climate Simulation (NCCS) and depriving it of external dependencies
• Making it efficient – parallelized for the NCCS HEC environment• Producing a new set of CCRs and assimilating them in the GEOS• Experiments: 2017 Boreal TC season; Selected Case Study: Harvey• Ongoing Work: assimilating CCRs from CrIS• Future work: IASI
2
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Ten years (2007-2017) of work by this team with assimilation of cloud-cleared retrievals did not have an impact on operational forecast centers
because of the choice of assimilating only radiancesIn 2018, this team produced an article demonstrating the positive impact of assimilating cloud-cleared AIRS radiances in a global framework on both global skill and tropical cyclone representation.
Reale, O., E. McGrath-Spangler, W. McCarty, D. Holdaway, R, Gelaro, 2018: Impact of adaptively thinned AIRS cloud-cleared radiances on tropical cyclone representation in a global data assimilation and forecast system. Weather and Forecasting, 33, 908-931.
Cloud-cleared AIRS radiances are superior compared to clear-sky
radiances, but need to be thinned more aggressively because of their
higher information content.
The best way to use AIRS CCRs is through an adaptive strategy that
assimilates more data around Tropical Cyclones, and less globally.
3
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Progress towards an operational use of CCRs• In spite of the overwhelming evidence that cloud-cleared radiances are an
immensely superior data type compared to clear-sky, cloud-cleared infrared radiances are still not been operationally used because of: 1) latency; and 2) external dependencies (ECMWF data; neural network) which are perceived by operational centers as not controllable
• With the goal of raising awareness and interest towards cloud-cleared AIRS products, and hopefully applying the methodology to CrIS and IASI, the cloud-clearing algorithm developed by Joel Susskind and his team was ported to NCCS and a first attempt was made to customize it (thanks to Lena Iredell, Lou Kouvarisand John Blaisdell).
• During 2019, the algorithm has been fully parallelized with a 75 gain factor in speed
• Produced AIRS CCRs to cover the entire boreal 2017 TC season
• Vast set of experiments in the new hybrid 4DEnVAR GEOS for the entire period
4
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Features of the parallelized Cloud-clearing algorithm
• Speed: Using Portable Distributed Scripts (PODS) available onNCCS platforms, on Intel Xeon Haswell processor nodes (each nodehas 28 cores of 2.6 GHz each and 128 Gb of available memory).With 8 nodes (224 processors), 7410 granules (31 days) areprocessed in about 5 hours instead of 370.5 hours.
• First Guess: based on internally produced GEOS fields• Selection of channels: based on customer needs, in our case based
on GMAO selection• Portability: with minimal changes, any operational center can
generate their own CCRs and adapt them to their own model
5
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
New Experiments: 2017 Boreal TC season
GEOS-5 DAS hybrid 4DEnVar version 5.17
Assimilation from 31 Jul – 20 Oct 2017 of all observations assimilated operationally
10 day forecasts from the following experiments:
• RAD: AIRS clear-sky radiances, regularly-spaced thinning (180km)• CLD3: AIRS cloud-cleared radiances from the GES DISC, regularly-spaced thinning (300km)• CCR: internally-produced AIRS cloud-cleared radiances, same thinning as CLD3
6
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
• The assimilation of AIRS CCRs produced with the customized algorithm has better forecast skill than the assimilation of AIRS CCRs obtained from the DISC beyond day 3, and slightly better skill than the assimilation of clear-sky radiances beyond day 7
• This is a completely GMAO-customized CCR product on NCCS tailored for the GEOS. It demonstrates that any Agency can produce CCRS internally without any external dependency and controlling latency
500 hPa Anomaly Correlation – Global10 Aug to 24 Sep 2017–still running-
10 days before Aug10th discarded for spinup
Using CLD3 as reference
Using RAD as reference
7
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
• The assimilation of AIRS CCRs produced with the customized algorithm has significantly better forecast skill than the assimilation of AIRS CCRs obtained from the DISC and slightly better skill than the assimilation of clear-sky radiances in the tropics on any forecast range
• This is consistent with the overall better representation of convective systems
• The improvement caused by assimilation of internally-produced AIRS CCRs is very robust
850 hPa Temperature RMS - Tropics
Using RAD as reference
Using CLD3 as reference
8
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Scorecard-based diagnostics reveal that the improvement brought by the assimilation of GMAO internally produced CCRs affects all levels of the atmosphere and is particularly remarkable in the tropics
CCR vs CLD3 GEOS Scorecard10 Aug – 24 Sep 2017
Comparison between the experiment assimilating internally produced CCRs against the CCRs obtained from the DISC
9
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Analysis of Hurricane Harvey (2017)
10
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Harvey’s rapid intensification as seen in the 3 experiments
Vertical cross sections (above): Wind speed (m/s shaded), Temperature (°C, black), Temp. Anomaly (°C, red)
Below: 850 hPa winds (m/s shaded), slp (hPa, contours)11
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Harvey’s forecast
Assimilation of internally produced AIRS CCRs improves the forecast for both track and intensity beyond day 3
12
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Continuing work on Arctic Dynamic Sensitivity to assimilation of AIRS cloud-cleared radiances (McGrath-Spangler et al. 2020, tbs)
Previous results are confirmed in the hybrid 4DEnVar environment
13
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space AdministrationContinuing work on convective mesocyclones outside the tropics (Ganeshan et al. 2020, tbs)
14
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Conclusions • Long progress since first experiments with retrievals in 2007
• In 2018 demonstrated value of AIRS cloud-cleared radiances
• Still resistance from the operational community because of latency and dependencies
• Ported the AST algorithm on NCCS, deprived if of neural network, ECMWF, allowed for customized
selection of channels; as well as parallelized it
• New set of experiments focused on the 2017 TC season (Aug-Oct)
• Obtained better results from assimilating customized CCRs than DISC CCRs or clear-sky radiances
• Started assimilating CrIS cloud-cleared radiances in hybrid 4DEnVar system
15
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
Ongoing and future work• Complete new sets of experiments to establish optimal density of global AIRS cloud-cleared
radiances in the hybrid 4DEnVar GEOS• Continue adaptive thinning experiments in hybrid 4DEnVar to further improve TC forecast skill• Perform experiments assimilating internally-generated CrIS cloud-cleared radiances• Continue experiments on the impact of cloud-cleared radiances in the Arctic region and on
convectively driven mesocyclones at high latitudes such as Polar Lows• Start producing internally generated cloud-cleared radiances for IASI• Submit article on Arctic Dynamic Sensitivity to assimilation of AIRS CCRs• Submit article on impact of AIRS CCRs on Polar Lows and extratropical convective mesocyclones
16
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration AcknowledgementsTsengdar Lee for current support through grant 80NSSC18K0927 “Using AIRS and CrISdata to understand processes affecting TC structure in a Global Data Assimilation and Forecasting Framework (2018-2021)” (PI: O. Reale)Ramesh Kakar for past support through previous grants NNX11AK05G and NNX14AK19G “Using AIRS Data to Understand Processes Affecting Tropical Cyclone Structure and Extreme Precipitation in a Global Data Assimilation and Forecasting Framework”(2011-2014, 2014-2018), PI: O. RealeTsengdar Lee for generous allocations of NASA High End Computing resources (NCCS)Louis Kouvaris and Lena Iredell, with help from John Blaisdell, for porting to NCCS the CC algorithm originally developed by Joel Susskind’s teamAIRS team at JPL and the Sounder Research Team at NASA GSFCAmal El Akkraoui, Matt Thompson and Ben Auer for help with the GEOSJules Kouatchou for help with parallelization of AST algorithm on NCCSGES DISC for their outstanding service to the community
17
Global Modeling and Assimilation Officegmao.gsfc.nasa.govGMAO
National Aeronautics and Space Administration
AIRS-related peer-reviewed articles published by this team
Reale, O., J. Susskind, R. Rosenberg, E. Brin, E. Liu, L. P. Riishojgaard, J. Terry, J. C. Jusem, 2008: Improving forecast skill by assimilation of quality-controlled AIRS temperature retrievals under partially cloudy
conditions. Geophysical Research Letters, 35, L08809, doi:10.1029/2007GL033002.
Reale, O., W. K. Lau, J. Susskind, E. Brin, E. Liu, L. P. Riishojgaard, M. Fuentes, R. Rosenberg, 2009: AIRS Impact on the Analysis and Forecast Track of Tropical Cyclone Nargis in a global data assimilation and
forecasting system. Geophysical Research Letters, 36, L06812, doi:10.1029/2008GL037122.
Reale, O., W. K. Lau, K.-M. Kim, E. Brin, 2009: Atlantic tropical cyclogenetic processes during SOP-3 NAMMA in the GEOS-5 global data assimilation and forecast system. Journal of the Atmospheric Sciences, 66, 3563-
3578.
Reale, O., K. M. Lau, J. Susskind, and R. Rosenberg, 2012: AIRS impact on analysis and forecast of an extreme rainfall event (Indus River Valley, Pakistan, 2010) with a global data assimilation and forecast system,
J. Geophys. Res., 117, D08103, doi:10.1029/2011JD017093.
Reale, O., E. McGrath-Spangler, W. McCarty, D. Holdaway, R, Gelaro, 2018: Impact of adaptively thinned AIRS cloud-cleared radiances on tropical cyclone representation in a global data assimilation and forecast system.
Weather and Forecasting, 33, 908-931.
18