Evaluating the NCEP GFS and NAM Model Clouds against Satellite Retrievals

1
Evaluating the NCEP GFS and NAM Model Clouds
against Satellite Retrievals

2

• Comparisons of High, Mid, Low Cloud Amounts

Zhang et al. 2005
3
(No Transcript)
4

• NCEP Global Forecast System
• (grid 003)
• Global Latitude/Longitude 1 deg Resolution
• Control time chosen 00Z
• Forecast times chosen 03, 06, 09, 12, 15,
  18, 21, 24Z
• Variables extracted
• high, middle, and low cloud cover
• cloud-top and cloud-base pressures
• - converted to km using
  relation
• 44307.693
  1-(pressure/1013.25)0.190284/1000
• Data availability (daily) off-line Feb.
  15, 2005 to May 31, 2007
• 
  on-line June 1, 2007 to current date
• http//nomads.ncdc.noaa.gov/cgi-bin/ncdc-ui/def
  ine-collection.pl?model_sysgfs-himodel_namegfs
  grid_name3

5
NCEP North American Mesoscale (NAM) model
Product nam.tcycz.awip32forecast_time.tm00,
where cyc00 and forecast_time00,03,06,09
, ...,24 Domain solid line below
Dates Analyzed January/April/July/Oct
ober/ 2,6,10,14,18,22,26,30
of years 2005,2006            
          
6

• CloudSat/Calipso
• As part of the A-train constellation, CloudSat
  is the
• first satellite-based millimeter-wavelength
  cloud radar
• - 94-GHz nadir-looking
• - 4 km (along-track) by 1.4 km
  (cross-track) footprint
• - 0.5 km vertical resolution between
  the surface and 25 km
• Goal is to obtain cloud profile information,
  liquid ice
• water content profiles and precipitation
  information to
• aid in the quantitative evaluation ofglobal
  atmospheric
• circulation models
• In operation collecting data since May 2006

Credit Alex McClung
7
thick clouds drizzle
thin-cloud profiles Ice/water phase
CERES TOA fluxes MODIS cloud height, re, ?
MLS
8
Zonal cloud distributions from CALIPSO
cloud fraction
aerosol extinction
9
Merged CALIPSO-CloudSat product now
available(Mace et al., JGR 2008)
CloudSat
CALIOP
Merged
10
Chang and Li (2005) Algorithm

• Capable of retrieving single-layer and
  thin-over-thick overlapped clouds and optical
  properties using multi-channels of MODIS
  satellite data
• takes full advantage of multi-spectral channels
  
• available from the Moderate Resolution Imaging
  
• Spectroradiometer (MODIS) on the Terra and
  Aqua
• platforms
• combines the MODIS CO2-slicing method with
• traditional IR and VIS techniques to overcome
  some
• limitations due to single-layer cloud
  assumptions used
• by conventional satellite cloud retrieval
  methods

11

• The Retrieval Scheme (Chang and Li, 2005, JAS)

• The basic principle
• Dual-layer clouds can be detected by using CO2
  slicing channel and IR channels.
• Location of high cloud is inferred from CO2
  slicing channel.
• Location of low cloud is determined from IR
  channels.
• Retrieve high-cloud and low cloud optical depth
  from a combination of visible and IR channels
  through an iterative procedure

(1) (2) (3) (4)
12
GLOBAL CLOUD COVER
Global Frequency of Cloud Layering from GLAS
13
Vertical Structure from CloudSat
14
(No Transcript)
15
ISCCP Cloud Classification
16

• Comparison between clouds from the Chang and Li
  (left) and the ISCCP-like (right) Methods

High cloud
Mid cloud
Low cloud
17

• Cirrus-Overlapping-Low Cloud Amount (High/Low)

January 2001
April 2001
October 2001
July 2001
Annual mean 27
Chang and Li (2004, JCL)
18
Avg Box 2.0 degrees Latitude, 2007 July
Calipso
GFS_Model
Our_retrieval
19
ISCCP 25.2
CALIPSO, single-layer low cloud 27.5
Our retrieval, single-layer low cloud 27.13
20
Our retrieval
GFS Model
Calipso
21
Our retrieval
GFS Model
Calipso
22
Our retrieval
GFS Model
Calipso
23
Our retrieval
GFS Model
Calipso
24
Our retrieval
GFS Model
Calipso
25
ISCCP 25.2
CALIPSO, single-layer low cloud 27.5
Our retrieval, single-layer low cloud 27.13
26
Our retrieval
GFS Model
Calipso
27
Our retrieval
GFS Model
Calipso
28
Our retrieval
GFS Model
Calipso
29
Our retrieval
GFS Model
Calipso
30
Our retrieval
GFS Model
Calipso
31
Our retrieval
GFS Model
Calipso
32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
Avg Box 2.0 Degrees Latitude, 4 Months (2006 Jul,
Oct, 2007 Jan, Apr)
Our_retrieval
Calipso
GFS_Model
37
Avg Box 2.0 Degrees Latitude, 4 Months (2006 Jan,
Apr, Jul, Oct )
NAM MODEL
Our Retrieval
38
Avg Box 6.0 Degrees Longitude, 4 Months (2006
Jan, Apr, Jul, Oct )
NAM MODEL
Our Retrieval
39
Resolution 32 km
Resolution 1 º
40
Comparison with Cloud Retrievals from CloudSat
41
Ongoing and Future Studies
Objective try to link model deficiencies
identified by detectable physical quantities with
atmospheric processes
Assess detailed cloud properties from the
models against similar parameters derived from
CALIPSO, CloudSat, MODIS, AMSR-E, etc. -
Cloud fraction, phase, type - Cloud heights
effective, top, base - Cloud thickness -
Cloud particle sizes - Ice and liquid water
paths IWP
42
CloudSat Global Cloud Fractions
43
The seasonal variations stratiform Ac occurrence
Wang et al.
44
Water, ice, and mixed-phase cloud distribution
annual mean
Wang et al
45
Satellite Estimates of Liquid Water Path
CERES/MODIS LWP
ISCCP LWP
SSM/I LWP
CloudSat non-precip LWP
CloudSat precip LWP
CloudSat total LWP
46
(No Transcript)
47
(No Transcript)
48
Measurable range by 35GHz radar
Measurable range by 14GHz radar
tropical rain
mid- high- latitude rain
Frequency
strong rain
weak rain
Rainrate
new measurable range by addition of 35GHz radar
(T. Iguchi CRL)
49
20N/S
JJA
20N-50N
20S-50S
50
Major findings

• Cloud products from three satellites sensors
  (MODIS-CL, CALIPSO and CLOUDSAT) bear great
  resemblance
• MODIS-CL is most compatible with CALIPSO w.r.t.
  detection of cloud tops
• In general, the GFS produce sound total cloud
  patterns on the global scale, and NAM does an
  even better job in NA
• The GFS model tends to generate less high clouds,
  more middle clouds and substantially less low
  clouds than C-C clouds, while NAM clouds agree
  better in all levels
• Both GFS and NAM produces far less cirrus cloud
  in the tropics,
• Both produce too much low clouds over land and
  too little over oceans
• Many other regional features are yet to be
  explored .. .

51
Future plan

• We shall continue to validate MODIS clouds
  against CC clouds and ground-based observation
• We shall continue to evaluate the GFS model to
  find any dependence of the discrepancies on
  atmospheric and surface environments and weather
  regimes
• We shall gain further insights into the causes of
  the discrepancies by looking into cloud
  microphysics, thermodynamic conditions,
• We shall match and analyze more satellite
  datasets to fully understand the problems,
  especially rainfall.
• Finally, we are looking forward to closer and
  fruitful collaboration with NOAA scientists.

52
Thanks my future neighbors !