1. Graeme Stephens • Colorado State University
Satellite observations and the MJO
Graeme Stephens • Colorado State University

2. Graeme Stephens • Colorado State University
Relevant Comments
Ka band (8mm)
ARM MMCR
W-band (3 mm)
Cloudsat CPR
1. Millimeter-wave radars (MWR), traditionally used to study ‘large scale non-precipitating clouds’, in fact have much to offer in study of moist, precipitating convection
2. Observations of MWRs are a step beyond the artificial and all too common state where we deal with clouds and precipitation (both in models and obs) as separate chain of events of the water cycle.
3. In embracing this broader framework, we are perhaps beginning to uncover new ideas about cloud and preciptation structures of moist convection with implications to the diabatic heating.
Graeme Stephens • Colorado State University

3. Surface MWR Observations
Manus Island
Ka-band radar (ARM MMCR)
- Clothiaux et al (2000) combined product
- 10 s / 45 m resolution
Surface radiation / meteorology
Six year ( ) record of daily mean OLR, interpolated from AVHRR, 2.5 x 2.5 degree grid are used to identify the MJO events (34 events)
> 2.1 million radar profiles were composited
Graeme Stephens • Colorado State University

4. MWR Classification of convection
Two-parameter classification system used to identify precipitating convective regimes of ‘self-similar’ vertical cloud structures Stephens and Wood (2007).
ETH ≡ height of cloud-top echo
PTH ≡ depth of penetration of the dBZ echo
ETH
PTH
10 dBZ echo
P(Column max Ze < x)
No precip
Precip x
Graeme Stephens • Colorado State University

5. Graeme Stephens • Colorado State University
Classification
Deep convection
Shallow convection but from
Ground also deep cnvection
High cloud over low precipitating convection, ….- or deep cloud with shallow precip (stratiform)
MMCR
CloudSat
X1 - Clouds that did not meet precip criteria
Graeme Stephens • Colorado State University

6. Graeme Stephens • Colorado State University
TRMM equivalent
Shipborne MMCR
Graeme Stephens • Colorado State University

7. Graeme Stephens • Colorado State University
MJO active
MJO transition
The cloud structures of each respective storm regime associated with different synoptic forcings are ‘identical’ - what differs is the relative frequency of occurrence of each class that defines the convective envelope
JASMINE Monsoon
Crystal FACE
Graeme Stephens • Colorado State University

8. Graeme Stephens • Colorado State University
Class A and B are very common at Manus and contribute ~ 1/3 of rainfall.
Two examples of storm class A/B
Overall, ~ 40% of the observed precip during the MJO cycle derives from multi-layered systems.… implications for depth of atmosphere heating in this mode?
Graeme Stephens • Colorado State University

9. Graeme Stephens • Colorado State University
CloudSat and A-train results
Graeme Stephens • Colorado State University

10. Graeme Stephens • Colorado State University
Some relevant details
Nadir pointing, 94 GHz radar
3.3s pulse  480m vertical res, over- sampled at ~240m
1.4 km horizontal res.
Calibration better than 2 dBZ
Sensitivity ~ -28 dBZ (-31 dBZ)
Dynamic Range: 80 dB
1. Formation with the A-Train
Two main components of design - CPR and formation flying
500m
~1.4 km
demonstrated
post launch
Hardware continues to operate with nominal performance and the mission is funded through 2010
Graeme Stephens • Colorado State University

11. Graeme Stephens • Colorado State University
2006 Dec/Jan MJO 17 16 15
1400km 30 km
Graeme Stephens • Colorado State University

12. Graeme Stephens • Colorado State University
Level 2 products:
2B geoprof
2B geoprof-lidar
2B-cloudclass
2B-tau
2B-CWC
2B-flxhr
Ancillary products - MODIS and ECMWF met data
Auxiliary products
Aux-Mod06
Aux-ECMWF
Aux-SSF
Aux-TRMM
Aux-AMSRE
Cloud mask by bin, reflectivity, MODIS cloud mask
Cloud lidar fraction matched to CloudSat
Cloud type classification radar+
Optical depth using MODIS +geoprof
Cloud water, ice -radar+ only, radar+2B-tau
Fluxes & heating rate, products above + ECMWF+
Properties matched to CloudSat-
partially done
liquid + solid precipitation, incidence and amount is to be added
Graeme Stephens • Colorado State University

13. Graeme Stephens • Colorado State University
Joint Lidar-Radar product
532 Total Attenuated Backscatter
CPR Reflectivity
Jay Mace
Graeme Stephens • Colorado State University

14. Graeme Stephens • Colorado State University
CloudSat is the most sensitive detector of precipitation currently in
orbit today (sensitive down to O~1mm/day)
Using the surface reflectivity it is possible to determine the path attenuation and with information of reflectivity near the surface, we can identify the likelihood of precipitation at the surface thus identify where rain is occurring and its intensity - ‘easy over ocean’.
We can also determine precipitation intensity - this is more a more ‘controversial’ endeavor
Precipitation (enhanced products) - incidence and amount
Haynes et al. 2007
Graeme Stephens • Colorado State University

15. Graeme Stephens • Colorado State University
Frequency of deep clouds> 6km
Frequency of multi-layering
This is consistent with the ARM TWP obs - that the vertical structure of tropical cloud systems (including precipitating systems) are frequently (~ 40%) multi-layered
Graeme Stephens • Colorado State University

16. Graeme Stephens • Colorado State University
JJA
incidence
latitude
rate mm/hr
AMSR-E
CloudSat
accumulation
Incidence & accumulation similar in tropics but vastly different in mid-higher latitudes (accumulation results & subsequent val are works in progress)
Graeme Stephens • Colorado State University

17. Graeme Stephens • Colorado State University
CP-ETH histograms
This mode is only partially the deep stratiform mode of convective precipitation - it also represents the multi-layered modes of precipitation
MMF
Missing deep ice/ precipitation mode
NICAM global CSRM
JJA, 30N/S CloudSat
Cloud echo top height (ETH) against the precipitation ETH => ETH of -30 dBZ versus ETH of 10 dBZ
Luo et al., 2007
Graeme Stephens • Colorado State University

18. Tropical west Pacific histograms: 2006/12 – 2007/02
CloudSat
MetUM N320L50
Reasonable ice microphysics?
Evaporating ice – or T dependence in convective cloud ice fraction?
Lack of mid-level cloud
Lack of non-drizzling low cloud
Graeme Stephens • Colorado State University

19. Graeme Stephens • Colorado State University
Accumulated oceanic precipitation
20N-S
40N-S
CloudSat
CloudSat
Low mid high
Low mid high
Graeme Stephens • Colorado State University

20. Graeme Stephens • Colorado State University
Summary/Comments
MWRs, and CloudSat specifically, are powerful new tools for studying the properties of tropical convection. Diagnostic tools for analyzing these new observations are alos progressing.
These observations and related tools are steps in building some understanding of clouds and their relationship to tropical precipitation.
These observations, related analysis tools and the understanding produced will prove invaluable as modeling of the MJO inevitably moves the cloud-scale modeling through CSRM/CRMs
Graeme Stephens • Colorado State University

21. Graeme Stephens • Colorado State University

22. Graeme Stephens • Colorado State University
Classification
Deep convection
Shallow convection
High cloud over low precipitating convection, ….- or deep cloud with shallow precip
CTH MMCR – CTH GMS
+ Deep convection
Graeme Stephens • Colorado State University

23. latitude The dreary extra-tropics cloud fraction (lidar/radar) JJA
precip/cloud
fraction
Grey range ~ uncertainty
range
Precip to surface
probable
Precip to surface
certain
Results over oceans
Graeme Stephens • Colorado State University

24. Graeme Stephens • Colorado State University
CloudSat 30S-30N
Stephens and Wood, 2007
Graeme Stephens • Colorado State University

25. Conclusions Future Work and Work in Progress
Quantification of the latent heating due to these precipitating systems, better understanding of microphysics and mesoscale dynamics of multi-layer systems
Representation of these cloud layers in data sets from other sensors
CloudSat will essentially eliminate the ambiguity associated with attenuation and allow development of near-global climatologies of multi-layer cloud
Evaluation of cloud resolving model ability to simulate these observed vertical cloud structures
Graeme Stephens • Colorado State University

26. Graeme Stephens • Colorado State University
Identify MJO ‘events’ at Manus using a minimum OLR-based criteria
Analyze all cloud radar profiles, surface meteorology, and radiosonde data for 14 days on either side of this event
Group the profiles by their time relative to the event, d ≡ 0
Lag Lag Lag Lag Lag +2
-14 d d d d d d
d = 0
Group observed cloud profiles according to similar characteristics (cloud height, thickness, layer boundaries, surface precipitation, …)
Use this technique to identify the dominant types of cloud systems present in various phases of the MJO cycle
Graeme Stephens • Colorado State University