THE NCEP CLIMATE FORECAST SYSTEM VERSION 2
SURANJANA SAHA
THE ENVIRONMENTAL MODELING CENTER
NCEP/NWS/NOAA
NEMS/GFS Modeling Summer School 2013
Presentation on Climate
8:30-9:00am, Wednesday July 31, 2013
1
CFSv2 consists of the following components:
1. Analysis Systems :
Atmospheric (CDAS)-GSI
Ocean-Ice (GODAS) and
Land (GLDAS)
2. Atmospheric Model :
Operational CFS
Noah Land Model
3. Ocean Model :
MOM4 Ocean Model
Sea Ice Model
The models used by the CFSv2 are:
1. Atmosphere at horizontal resolution of spectral T126
(~100 km) and vertical resolution of 64 sigma-pressure
hybrid levels
2. Interactive ocean with 40 levels in the vertical, to a
depth of 4737 m, and horizontal resolution of 0.25
degree at the tropics, tapering to a global resolution of
0.5 degree northwards and southwards of 10N and 10S
respectively
3. Interactive 3 layer sea-ice model
4. Interactive land model with 4 soil levels
The NCEP Climate Forecast System Version 2 (CFSv2)
was implemented into operations on March 30, 2011.
A new Reanalysis of the atmosphere, ocean, seaice and land was made over
a 32-year period (1979-2010) to provide consistent initial conditions for:
A complete Reforecast of the new CFSv2 over the 29-year period (19822010), in order to provide stable calibration and skill estimates of the new
system, for operational seasonal prediction at NCEP
What is a Reanalysis ?
• analysis made after the fact (not ongoing in real time)
• with an unchanging model to generate the model guess
(MG)
• with an unchanging data assimilation method (DA)
• no data cut-off windows and therefore more quality
controlled observations (usually after a lot of data mining)
5
Motivation to make a Reanalysis ?
• To create a homogeneous and consistent climate record
Examples: R1/CDAS1: NCEP/NCAR Reanalysis (1948-present) Kalnay et al.,
Kistler et al
R2/CDAS2 : NCEP/DOE Reanalysis (1979-present) Kanamitsu et al
ERA40, ERA-Interim, MERRA, JRA25, NARR, etc….
• To create a large set of initial states for Reforecasts
(hindcasts, retrospective forecasts..) to calibrate real time
extended range predictions (error bias correction).
6
ONE DAY OF ANALYSIS
12Z GSI
18Z GSI
0Z GSI
6Z GSI
0Z GODAS
6Z GODAS
0Z GLDAS
12Z GODAS
18Z GODAS
9-hr coupled T382L64 forecast guess (GFS + MOM4 + Noah)
7
CFSR
T382 horizontal resolution (~38 Km)
Sigma-pressure hybrid vertical coordinate with 64 levels with top pressure ~0.266
hPa
Simplified Arakawa-Schubert convection with momentum mixing
Tiedtke (1983) shallow convection modified to have zero diffusion above the low
level inversions
Prognostic ozone with climatological production and destruction terms computed
from 2D chemistry models
Prognostic cloud condensate from which cloud cover is diagnosed
Orographic gravity wave drag based on Kim and Arakawa(1995) approach and
sub-grid scale mountain blocking following Lott and Miller (1997)
8
Courtesy: Shrinivas Moorthi
CFSR (contd)
AER RRTM IR radiation with maximum/random cloud overlap and observed global
mean CO2
AER RRTM SW radiation with maximum/random overlap and observed global mean
CO2, aerosols including volcanic origin plus rare gases.
Non-local vertical diffusion in the PBL with local-K in the free atmosphere with
exponentially decaying background diffusion coefficient
Eighth order horizontal diffusion
Specific enthalpy as a prognostic variable. More accurate thermodynamic equation.
Noah 4 layer land surface model
Coupled to GFDL MOM4 and a 3 layer sea-ice model
9
Courtesy: Shrinivas Moorthi
CFSR data dump volumes, 1978-2009, in GB/month
10
Courtesy: Jack Woollen
Another innovative feature of the CFSR GSI is the use of the
historical concentrations of carbon dioxide when the historical
TOVS instruments were retrofit into the CRTM.
Satellite Platform
Mission Mean
(ppmv)b
TIROS-N
337.10
NOAA-6
340.02
NOAA-7
342.96
NOAA-8
343.67
NOAA-9
355.01
NOAA-10
351.99
NOAA-11
363.03
NOAA-12
365.15
GEOS-8
367.54
GEOS-0
362.90
GEOS-10
370.27
NOAA-14 to NOAA-18
380.00
IASI METOP-A
389.00
NOAA-19
391.00
Courtesy: http://gaw.kishou.go.jp11
The linear trends are 0.66, 1.02 and 0.94K per 31 years for R1, CFSR
and GHCN_CAMS respectively. (Keep in mind that straight lines
may not be perfectly portraying climate change trends).
12
Courtesy: Huug van den Dool
SST-Precipitation Relationship in CFSR
Precipitation-SST lag correlation in tropical Western Pacific
13
Response of Prec. To SST increase : warming too quick in R1 and R2
simultaneous positive correlation in R1 and R2
Courtesy: Jiande Wang
5-day T126L64 forecast anomaly correlations
14
Courtesy: Bob Kistler
THE OCEAN, SEA ICE AND COUPLER
15
The global number of temperature observations assimilated per month by the
ocean component of the CFSR as a function of depth for the years
1980-2009.
16
Courtesy: Dave Behringer
The global distribution of all temperature profiles assimilated by the ocean
component of the CFSR for the year 1985. The distribution is
dominated by XBT profiles collected along shipping routes.
17
Courtesy: Dave Behringer
The global distribution of all temperature profiles assimilated by the ocean
component of the CFSR for the year 2008. The Argo array (blue)
provides a nearly uniform global distribution of temperature profiles
18
Courtesy: Dave Behringer
The Diurnal Cycle of SST in CFSR
The diurnal cycle of SST in the TAO data (black line) and CFSR (blue line) in
the Equatorial Pacific for DJF (top three panels) and JJA (bottom three panels).
165E
165E
170W
110W
170W
110W
T
DJF
DJF
JJA
JJA
19
Courtesy: Sudhir Nadiga
Zonal and meridional surface velocities for CFSR (top left and top right) and
differences between CFSR and drifters from the
Surface Velocity Program of TOGA (bottom panels).
20
Courtesy: Sudhir Nadiga
The first two EOFs of the SSH variability for the CFSR (left) and for TOPEX
satellite altimeter data (right) for the period: 1993-2008.
The time series amplitude factors are plotted in the bottom panel.
21
Courtesy: Sudhir Nadiga
Monthly mean sea ice concentration
for the Arctic from CFSR
(6-hr forecasts)
22
Courtesy: Xingren Wu
Monthly mean Sea ice extent (106 km2)
for the Arctic (top) and Antarctic (bottom) from CFSR (6-hr forecasts).
5-year running mean is added to detect long term trends.
23
Courtesy: Xingren Wu
CFS grid architecture in the coupler. ATM is MOM4 atmospheric model (dummy
for CFS), SBL is the surface boundary layer where the exchange grid is
located, LAND is MOM4 land model (dummy for CFS), ICE is MOM4 sea ice
model and OCN is MOM4 ocean model.
24
Courtesy: Jun Wang
Data Flow
Fast loop: if Δa= Δc= Δi, coupled at every time step
Slow loop: Δo
ATM (dummy)
GFS
Coupler
Δa
Δc
LAND (dummy)
Sea-ice
Ocean
Δi
Δo
• Fast loop: can be coupled at every time step
• Slow loop:
•
a. passing variables accumulated in fast loop
•
b. can be coupled at each ocean time step
25
Courtesy: Jun Wang
Passing variables
• Atmosphere to sea-ice:
•
- downward short- and long-wave radiations,
•
- tbot, qbot, ubot, vbot, pbot, zbot,
•
- snowfall, psurf, coszen
• Atmosphere to ocean:
•
- net downward short- and long-radiations,
•
- sensible and latent heat fluxes,
•
- wind stresses and precipitation
• Sea-ice/ocean to atmosphere
– surface temperature,
– sea-ice fraction and thickness, and snow depth
26
Courtesy: Jun Wang
THE SURFACE
27
2-meter volumetric soil moisture climatology of CFSR
for May averaged over 1980-2008.
28
Courtesy: Jesse Meng
2-meter volumetric soil moisture climatology of CFSR
for Nov averaged over 1980-2008.
29
Courtesy: Jesse Meng
The CFSR soil moisture climatology is consistent with GR2
and NARR on regional scale. The anomaly agrees with the
Illinois observations, correlation coefficient = 0.61.
Global Soil Moisture Fields in the NCEP CFSR
Soil Moisture
[%]
[mm]
Anomaly
[%]
CONUS
[mm]
Illinois
R(GR2,OBS)=0.48
GR2
NARR
CFSR
OBS
R(NARR,OBS)=0.67 R(CFSR,OBS)=0.61
30
Courtesy: Jesse Meng
Global average of monthly-mean
Precipitation (a), Evaporation (b) and E-P (c).
31
Courtesy: Wanqiu Wang
Monthly mean hourly surface pressure with the daily mean
subtracted for the month of March 1998
32
Courtesy: Huug van den Dool
Fig. 3 Correlation of intraseasonal precipitation with CMORPH. (a) R1, (b) R2,
and (c) CFSR. Contours are shaded starting at 0.3 with 0.1 interval.
33
Courtesy: Jiande Wang et al
Fig. 4. (a) Standard deviation of intraseasonal rainfall anomalies from CMORPH. (b) differences
in standard deviation of intraseasonal rainfall anomalies between R1 and CMORPH. (c) As in (b)
except for R2. (d) As in (b) except for CFSR. Contours are shaded at an interval of 2 mm/day in
(a) and 1 mm/day in (b), (c) and (d) with values between -1 and 1 plotted as white.
34
Courtesy: Jiande Wang et al
The NCEP Climate Forecast System
Reanalysis
Suranjana Saha, Shrinivas Moorthi, Hua-Lu Pan,
Xingren Wu, Jiande Wang, Sudhir Nadiga, Patrick Tripp,
Robert Kistler, John Woollen, David Behringer, Haixia
Liu, Diane Stokes, Robert Grumbine, George Gayno,
Jun Wang, Yu-Tai Hou, Hui-ya Chuang, Hann-Ming H.
Juang, Joe Sela, Mark Iredell, Russ Treadon, Daryl
Kleist, Paul Van Delst, Dennis Keyser, John Derber,
Michael Ek, Jesse Meng, Helin Wei, Rongqian Yang,
Stephen Lord, Huug van den Dool, Arun Kumar, Wanqiu
Wang, Craig Long, Muthuvel Chelliah, Yan Xue, Boyin
Huang, Jae-Kyung Schemm, Wesley Ebisuzaki, Roger
Lin, Pingping Xie, Mingyue Chen, Shuntai Zhou, Wayne
Higgins, Cheng-Zhi Zou, Quanhua Liu, Yong Chen,
Yong Han, Lidia Cucurull, Richard W. Reynolds, Glenn
Rutledge, Mitch Goldberg
Bulletin of the American Meteorological Society
Volume 91, Issue 8, pp 1015-1057.
doi: 10.1175/2010BAMS3001.1
35
AND NOW……..
THE SECOND ‘R’ IN
36
Differences between the model used here and in CFSR are mainly
in the physical parameterizations of the atmospheric model and
some tuning parameters in the land surface model and are as
follows:
•We use virtual temperature as the prognostic variable, in place of
enthalpy that was used in major portions of CFSR. This decision
was made with an eye on unifying the GFS (which uses virtual
temperature) and CFS, as well as the fact that the operational CDAS
with CFSv2 currently uses virtual temperature.
•We also disabled two simple modifications made in CFSR to
improve the prediction of marine stratus (Moorthi et al., 2010, Saha
et al., 2010, Sun et al., 2010). This was done because including
these changes resulted in excessive low marine clouds, which led to
increased cold sea surface temperatures over the equatorial oceans
in long integrations of the coupled model.
37
A new parameterization of gravity wave drag induced by
cumulus convection based on the approach of Chun and
Baik (1998) (Johansson, 2009, personal communication).
The occurrence of deep cumulus convection is associated
with the generation of vertically propagating gravity waves.
While the generated gravity waves usually have eastward or
westward propagating components, in our implementation
only the component with zero horizontal phase speed is
considered.
This scheme approximates the impact of stationary gravity
waves generated by deep convection. The base stress
generated by convection is parameterized as a function of
total column convective heating and applied at the cloud
top.
Above the cloud top the vertically propagating gravity
waves are dissipated following the same dissipation
algorithm used in the orographic gravity wave formulation.
38
In CFSR, a standard cloud treatment is employed in both the
RRTM longwave and shortwave parameterizations, that layers of
homogeneous clouds are assumed in fractionally covered model
grids. In the new CFS model, an advanced cloud-radiation
interaction scheme is applied to the RRTM to address the
unresolved variability of layered cloud. In McICA, a random
column cloud generator samples the model layered cloud into subcolumns and pairs each column with a pseudo-monochromatic
calculation in the radiative transfer model.
In calculating cloud optical thickness, all the cloud condensate in
a grid box is assumed to be in the cloudy region. So the in-cloud
condensate mixing ratio is computed by the ratio of grid mean
condensate mixing ratio and cloud fraction when the latter is
greater than zero.
The CO2 mixing ratio used in these retrospective forecasts
includes a climatological seasonal cycle superimposed on the
observed estimate at the initial time.
39
The Noah land surface model (Ek et al., 2003) used in CFSv2 was
first implemented in the GFS for operational medium-range
weather forecast (Mitchell et al., 2005) and then in the CFSR (Saha
et al., 2010).
Within CFSv2, Noah is employed in both the coupled landatmosphere-ocean model to provide land-surface prediction of
surface fluxes (surface boundary conditions), and in the Global
Land Data Assimilation System (GLDAS) to provide the land
surface analysis and evolving land states.
While assessing the predicted low-level temperature, and land
surface energy and water budgets in the CFSRR reforecast
experiments, two changes to CFSv2/Noah were made:
1. To address a low-level warm bias (notable in mid-latitudes), the
CFSv2/Noah vegetation parameters and rooting depths were
refined to increase evapotranspiration, which, along with a
change to the radiation scheme (RRTM in GFS and CFSR, and
now McICA in CFSv2), helped to improve the predicted 2-meter
air temperature over land.
2. To accommodate a change in soil moisture climatology from
GFS to CFSv2, Noah land surface runoff parameters were
nominally adjusted to favorably increase the predicted runoff. 40
Hindcast Configuration for CFSv2
•
•
•
9-month hindcasts were initiated from every 5th day and run from all 4 cycles of that day,
beginning from Jan 1 of each year, over a 29 year period from 1982-2010 This is required to
calibrate the operational CPC longer-term seasonal predictions (ENSO, etc)
There is also a single 1 season (123-day) hindcast run, initiated from every 0 UTC cycle between
these five days, over the 12 year period from 1999-2010. This is required to calibrate the
operational CPC first season predictions for hydrological forecasts (precip, evaporation, runoff,
streamflow, etc)
In addition, there are three 45-day (1-month) hindcast runs from every 6, 12 and 18 UTC cycles,
over the 12-year period from 1999-2010. This is required for the operational CPC week3-week6
predictions of tropical circulations (MJO, PNA, etc)
Jan 1
Jan 2
Jan 3
Jan 4
Jan 5
Jan 6
0 6 12 18
0 6 12 18
0 6 12 18
0 6 12 18
0 6 12 18
0 6 12 18
9 month run
1 season run
45 day run
41
Operational Configuration of CFSv2
•
•
•
•
There are 4 control runs per day from the 0, 6, 12 and 18 UTC cycles of the CFS realtime data assimilation system, out to 9 months.
In addition to the control run of 9 months at the 0 UTC cycle, there are 3 additional
runs, out to one season. These 3 runs per cycle are initialized with a simple
perturbation method.
In addition to the control run of 9 months at the 6, 12 and 18 UTC cycles, there are 3
additional runs, out to 45 days. These 3 runs per cycle are initialized with a simple
perturbation method.
There are a total of 16 CFS runs every day, of which 4 runs go out to 9 months, 3 runs
go out to 1 season and 9 runs go out to 45 days.
0 UTC
6 UTC
9 month run (4)
12 UTC
1 season run (3)
18 UTC
45 day run (9) 42
9-MONTH HINDCASTS
28 years: 1982-2009; all 12 months.
CFSv1 : 15 members per month, total of 180 initial states per year
CFSv2: 24 members per month (28 for November), total of 292
initial states per year.
Sample size: 5040 for CFSv1; 8176 forCFSv2.
43
•
•
•
•
•
•
•
•
Definitions and Data
AC of ensemble averaged monthly means
GHCN-CAMS (validation for Tmp2m)
CMAP (validation for Prate)
OIv2 (validation for SST)
1982-2009 (28 years)
Common 2.5 degree grid
Variables/areas studied: US T, US P, global and
Nino34 SST, global and Nino34 Prate.
Two climos used for all variables within tropics
30S-30N: 1982-1998 and 1999-2009
Elsewhere: 1982-2009
44
More skill globally for
CFSv2
45
More skill in the
western Pacific for
CFSv2
46
More skill west of the
dateline and over the
Atlantic for CFSv2
47
Evaluation of anomaly correlation as
a function of target month (horizontal
axis) and forecast lead (vertical axis).
On the left is CFSv1, on the right
CFSv2.
Top row shows monthly 2-meter
temperature over NH land
Middle row shows monthly
precipitation over NH land
Bottom row shows the SST in the
Nino3.4 area.
The scale is the same for all 6 panels.
48
The annual mean systematic
error in three parameters (SST,
T2m and Prate) at lead 3
evaluated as the difference
between the predicted and
observed climatology for the
full period 1982-2009.
Column on the left (right) is
for CFSv1 (CFSv2).
The header in each panel
contains the root-mean-square
difference, as well as the
spatial mean difference.
Units are K for SST and T2m
and mm/day for prate.
Contours and colors as
indicated by the bar
underneath.
49
The Brier Skill Score (BSS) of
prediction of the probability of
terciles of monthly T2m at
lead 1month.
On the left (right) the upper
(lower) tercile.
Upper row is CFSv1 15
members.
Middle row is CFSv2 15
member
Lower row is CFSv2 all 24
members.
All start months combined.
Period is 1982-2009.
Below each map is the map
integrated BSS value.
50
45-DAY HINDCASTS
11 years: 1999-2009; all 12 months.
CFSv1 : 15 members per month, total of 180 initial states per year
CFSv2: every cycle of every day of the year, total of 1460 initial
states per year.
Sample size: 1980 for CFSv1; 16060 for CFSv2.
51
The bivariate anomaly
correlation (BAC)x100 of
CFS in predicting the MJO
for period 1999-2009, as
expressed by the Wheeler
and Hendon (WH) index
(two EOFs of combined
zonal wind and OLR).
On the left is CFSv2 and on
the right is CFSv1.
Both are subjected to
Systematic Error Correction.
The black lines indicate the
0.5 level of BAC.
Qin Zhang and Huug van den Dool, CPC
52
Before Model Bias Correction
After Model Bias Correction
53
Qin Zhang and Huug van den Dool, CPC
Before Model Bias Correction
After Model Bias Correction
54
Qin Zhang and Huug van den Dool, CPC
The NCEP Climate Forecast System Version 2 (2013)
Suranjana Saha, Shrinivas Moorthi, Xingren Wu, Jiande Wang,
Sudhir Nadiga, Patrick Tripp, David Behringer, Yu-Tai Hou, Hui-ya
Chuang, Mark Iredell, Michael Ek, Jesse Meng, Rongqian Yang,
Malaquias Pena Mendez, Huug van den Dool, Qin Zhang, Wanqiu
Wang, Mingyue Chen, Emily Becker.
(Journal of Climate, under review, revised.)
55
THANK YOU
CFS Website : http://cfs.ncep.noaa.gov
Email : cfs@noaa.gov
56