evaluation of gcm convection schemes via data assimilation: e.g. to study the madden-julian...
TRANSCRIPT
![Page 1: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/1.jpg)
Evaluation of GCM convection schemes via data assimilation:
e.g. to study the Madden-Julian
Oscillation in a model that doesn’t have one
Brian Mapes
RSMAS, University of Miami
with
Julio Bacmeister
(then NASA, now NCAR)
![Page 2: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/2.jpg)
Why assimilation-based science?
![Page 3: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/3.jpg)
Why assimilation-based science?
![Page 4: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/4.jpg)
Why assimilation-based science?
![Page 5: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/5.jpg)
New! MERRA reanalysis
OBS precip, u850 GEOS5
no MJO -- Good news!
Kim et al. 2009
![Page 6: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/6.jpg)
some analyzed state variable
Z at some point
time
free model solution: Żana= 0 (biased, weather unsynchronized, lacks MJO)
initialized free model use piecewise constant Żana(t) to make above equations exactly true in each 6h time intervalwhile visiting analyzed states exactly
“Replay” analyzed wx
ΔZ/Δt = Żmodel + Żana
ΔZ/Δt = (Żdyn + Żphys) + Żana
MERRA’s variables Z [T,u,v,qv]satisfy:
![Page 7: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/7.jpg)
time
any analyzed variable
Z at 6h intervals
Żana= (Ztarget– Z) /relax
model drift balanced by
nudge
ΔZ/Δt = Żmodel + Żana
ΔZ/Δt = (Żdyn + Żphys) + ŻanaPoor man’s version (& interpretive aid):
nudged trajectoryInterpolate analyses to GCM grid & time steps: ‘target’ state
![Page 8: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/8.jpg)
time
Misses analysis (in direction toward model attractor) by a skinch, but analysis is already biased that way
(analyzed MJO a bit weak)
miss analysis by a skinch ( 1/relax
![Page 9: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/9.jpg)
Żana= (Ztarget– Z) /relax
• Need to choose relax
• Any small value will converge to same results
•Strong forcing (incl. q & div) forces rainfall (M. Suarez), but can blow up model (B. Kirtman)
• Dodge trouble, and do science: discriminate mechanisms, by using different relax values for different variables (e.g. winds; div vs. rot; T, q)
ΔZ/Δt = Żmodel + Żana
ΔZ/Δt = (Żdyn + Żphys) + Żana
Poor man’s data assimilation: nudge to analyses
![Page 10: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/10.jpg)
Learning from analysis tendencies
(ΔZ/Δt)obs = (Żdyn + Żphys) + Żana
• If state is kept accurate (LS flow & gradients), then (ΔZ/Δt)obs and advective terms Żdyn will be accurate
• and thus
Żana ≅ -(error in Żphys)
✔✔
![Page 11: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/11.jpg)
Example 1: mean heating rate errorsdT/dtmoist dT/dtana
100
500
mb
1000
Strange “stripe” of moist-physics cooling at 700mb (melting at 10C, & re-evap)
High wavenumber in model T(p) profile disagrees w/obs. & so is fought by data assim = WRONG
(magnitudes much smaller)15-30 December, 1992 (COARE)
![Page 12: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/12.jpg)
Example 2: MJO-related physics errorsjust do more sophisticated Żana averaging
(MJO phase composites)
1. Case studies (JFMA90, DJFM92)of 3D (height-dependent) fields (dT/dtana , dq/dtana , etc)averaging Indian-Pacific sector longitudes together
1. 27-year compositeof various 2D (single level or vertical integral) datasetsas a function of longitude
![Page 13: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/13.jpg)
• Error lesson: model convection scheme acts too deep (drying instead of moistening) in the leading edge of the MJO.
![Page 14: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/14.jpg)
When MJO rain is over Indian Ocean, W. Pac. atmosphere is observed to be
moistening, but GCM doesn’t, so analysis tendency has to do it
![Page 15: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/15.jpg)
Equatorial section of MJO phase 2 dqdt_ana anomalies
![Page 16: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/16.jpg)
9 8 7 6 5 4 3 2 1 0 ‘back’ (W) ‘front’ (E)
Objective, unbiased-sample MJO mosaic of CloudSat radar echo objects
Riley and Mapes, in prep.
![Page 17: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/17.jpg)
Physics: lack of convective ”organization” ?
(a whole nuther talk)
org = 0.1 org =0.5New plume ensembleapproach(in prep)
![Page 18: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/18.jpg)
OK, a “better” scheme (candidates)• For schemes as mission-central as convection,
evaluation has to be comprehensive
• Żana is a powerful guide to errors!– Mean, MJO... but also diurnal, seasonal, ENSO,...
– simply save d()dt_ana, as well as state vars ()– send into existing diagnostic plotting codes– similar to (obs-model) analyses, but automatic
• (all data on same grid, etc.)
![Page 19: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/19.jpg)
How to get Żana datasets? Nudge GCMs to world’s great analyses
• Full blown raw-data assimilation is expen$$ive, & really...are we gonna beat EC, JMA, NCEP?
• Multiple GCMs nudged to multiple reanalyses– Bracket/ estimate/ remove 2-model (anal. model + eval.
GCM) error interactions
• Commonalities teach us about nature, since all exercises share global obs. & intensive assim.
• Differences play valuable secondary role of informing individual model improvement efforts
• (Shameless: CPT proposal in community’s hands now...)
![Page 20: Evaluation of GCM convection schemes via data assimilation: e.g. to study the Madden-Julian Oscillation in a model that doesn’t have one Brian Mapes RSMAS,](https://reader035.vdocuments.us/reader035/viewer/2022062515/56649f525503460f94c759ff/html5/thumbnails/20.jpg)