The Land Model and Land Assimilation of the

1
The Land Model and Land Assimilation of the CFS
Reanalysis and Reforecast (CFSRR)
Assistance from other EMC members Suru Saha,
Shrinivas Moorthi, Cathy Thiaw
Land Assimilation Collaborators NASA GSFC
Hydrological Sciences Branch
CFSRR Advisory Board Meeting 07-08 November 2007
2
ONE DAY OF REANALYSIS Note daily GLDAS (spans
prior 24-hrs)
12Z GSI
18Z GSI
0Z GSI
6Z GSI
0Z GLDAS
12Z GODAS
18Z GODAS
0Z GODAS
6Z GODAS
9-hr coupled T382L64 forecast guess (GFS MOM4
Noah)
1 Jan 0Z
2 Jan 0 Z
3 Jan 0Z
4 Jan 0Z
5 Jan 0Z
2-day T382L64 coupled forecast ( GFS MOM4
Noah )
3
Outline

• Next-generation CFS
• Analysis physics upgrades Atmos, Ocean, Land,
  Sea-Ice
• History, summary, and assessment of Noah LSM
• Noah LSM features compared to forerunner OSU LSM
• Noah LSM Impact in coupled GFS and Regional
  Reanalysis
• Impact in N. American Regional Reanalysis (NARR)
• Impact in medium-range Ops GFS upgrades of May
  2005
• GLDAS Global Land Data Assimilation System
• Configuration and results from lower-resolution
  27-year execution
• CFS Reforecast Experiments Land Component tests
• Land Models Two models (Noah LSM, OSU LSM)
• Land initial states Two sources (Global Reanal
  2, GLDAS)
• Conclusions and Pending Issues

4

• New CFS implementation
• Analysis Systems Operational DAS Atmosphe
  ric (GSI) Ocean (GODAS) and Land (GLDAS)
• 2. Atmospheric Model Operational GFS
• 3. Land Model New Noah Land Model
• 4. Ocean Model New MOM4 Ocean Model
• New SEA ICE Model

5
EMC Land Surface Partnerships GCIP/GAPP/CPPA
(NOAA/CPO) and JCSDA
GLDAS
COLA/GMU
Paul Houser Paul Dirmeyer
6
(No Transcript)
7
History of the Noah LSM

• Oregon State University 1980s
• OSU/CAPS LSM was forerunner of Noah LSM
• Initial development was funded by Air Force
• Transitioned to Air Force in late 1980s
• Implemented in Air Force GLDAS (known as AGRMET)
• Transitioned to NCEP Ops mesoscale Eta model in
  1996
• Coined NOAH LSM after NCEP, OSU, Air Force and
  OHD upgrades
• Transitioned to NCAR in late 1990s
• Implemented in NCAR Community MM5 mesoscale model
  (F. Chen)
• Applied in NCEP N. American Regional Reanalysis
  (NARR)
• 1979 to present
• Implemented with NCEP Ops WRF meso model in Jun
  06
• Implemented in NCEP Ops medium-range GFS in May
  2005
• GFS Global Forecast System

8
GFS and CFS Land Model UpgradeNoah LSM (new)
versus OSU LSM (old)

• OSU LSM
• 2 soil layers (10, 190 cm)
• No frozen soil physics
• Only one snowpack state (SWE)
• Surface fluxes not weighted by snow fraction
• Vegetation fraction never less than 50 percent
• Spatially constant root depth
• Runoff infiltration do not account for subgrid
  variability of precipitation soil moisture
• Poor soil and snow thermal conductivity,
  especially for thin snowpack

• Noah LSM (vegetation, snow, ice)
• 4 soil layers (10, 30, 60, 100 cm)
• Frozen soil physics included
• Add glacial ice treatment
• Two snowpack states (SWE, density)
• Surface fluxes weighted by
• snow cover fraction
• Improved seasonal cycle of vegetation
• Spatially varying root depth
• Runoff and infiltration account for sub-grid
  variability in precipitation soil moisture
• Improved thermal conduction in soil/snow
• Higher canopy resistance
• Improved evaporation treatment over bare soil and
  snowpack

Noah LSM replaced OSU LSM in operational NCEP
medium-range Global Forecast System (GFS) in late
May 2005
9
Noah LSM Testing Sequence

• Uncoupled testing
• 1-d column model
• 3-d NLDAS and GLDAS
• National and Global Land Data Assimilation
  Systems
• Coupled testing (then Ops Implementations)
• ETA and WRF mesoscale model (NAM)
• N. American Regional Reanalysis (NARR)
• Global Forecast System (GFS)
• Coupled Forecast System (CFS)

10
NLDAS surface energy Fluxes across ARM- CART
sites of Oklahoma. Multi-station average ff
model and obs Jan 98 Sep 99
Three land models shown Blue -- Noah Green --
VIC Red -- Mosaic Noah performs well, arguably
the best.
11
Noah LSM in N. American Regional Reanalysis NARR
Soil moisture availability (percent of
saturation) Top 1-meter of soil column
The hallmark assimilation of high-resolution
hourly precipitation analyses in the NARR is not
feasible in the Global Reanalysis owing to lack
of timely Global precip analysis of sufficient
quality and retrospective availability.
12
GFS Implementation of Noah LSM31 May 2005

• NCEP TPB
• http//www.emc.ncep.noaa.gov/gc_wmb/Documentation/
  TPBoct05/T382.TPB.FINAL.htm
• Increase in horizontal resolution
• Noah LSM replaces OSU LSM
• New sea-ice treatment
• Enhanced mountain blocking
• Modified vertical diffusion
• Analysis upgrades
• Additional satellite radiance data
• Enhanced quality control
• Improved surface emissivity calculations over snow

13
Annual mean biases in surface energy fluxes In
five operational GCMs during 2003-2004 w.r.t.
nine flux-station sites distributed
world-wide from K. Yang et al. CEOP Study (2007,
J. Meteor. Soc. Japan)
lE Latent Heat Flux H Sensible Heat Flux Rn
Net Radiation µ Global mean
Mean Bias Error (MBE)
Pre-May 2005 NCEP GFS had large positive bias in
surface latent heat flux and corresponding large
negative bias in surface sensible heat flux. Also
large positive bias in precipitation in humid
regions (not shown).
14
Mean GFS surface latent heat flux 09-25 May
2005 Upgrade to Noah LSM significantly reduced
the GFS surface latent heat flux (especially in
non-arid regions)
Pre-May 05 GFS with OSU LSM
Post-May 05 GFS with new Noah LSM
15
Global Land Data Assimilation System
(GLDAS)with Noah LSM(Next 7 Frames)

• Motivation for GLDAS
• high precip bias over tropical land mass in
  coupled GDAS
• GLDAS Configuration for T382 CFSRR
• T126 uncoupled (about 1-deg resolution)
• Precip forcing CPC global 5-day CMAP precip anal
• Only over Tropical Latitudes (otherwise model
  precipitation)
• Non-Precip forcing T62 NCEP/DOE Global Reanal 2
  (GR2)
• GLDAS Results from low-res multi-decade test
• Period 1979-2006
• Cold Start 5-year spin up with 1979 forcing
• Compared with Global Reanalysis 2

16
Precipitation JJA 2007
OPI CDAS1 CDAS2 GDAS
Global 2.62 2.97 3.42 3.23
Land 2.11 2.73 2.83 2.72
Ocean 2.84 3.07 3.67 3.45
Greatest GDAS high precip bias over land appears
over tropical land mass Next Frame (e.g. central
Africa, northern S. America, India and Southeast
Asia)
17
GDAS-minus-OBS Jun-Jul-Aug 2007 Precipitation
Total
From land perspective Largest positive bias over
tropical latitudes.
GR1-minus-Obs
GR2-minus-Obs
18
Motivation for Using CMAP Precipitation in
Tropical Latitudes in GLDAS GDAS shows high bias
in tropical precipitation compared to CMAP
analysis 10 July 09 Aug 2007 Example in
tropical Africa
CMAP Precip Analysis 10Jul07 09Aug07
GDAS Precip Field 10Jul07 09Aug07
19
Two-Year (Oct 05 Sep 07) Soil Moisture Time
Series at Four global locations for 10-40 cm
layer in Noah LSM of Ops GDAS(does not utilize
CMAP precipitation forcing)
North Central USA
Equatorial Africa
Three of four locations look reasonable, except
tropical Africa is spinning up to very moist state
Central Amazon
Southeast Asia
20
GLDAS uses computational infrastructure
ofNASA/GSFC/HSB Land Information System (LIS)
21
CFSRR GLDAS Configuration

• Uncoupled execution of NASA LIS computational
  infrastructure
• Same Noah LSM source code as in coupled GDAS
• same four soil layers (10, 30, 60, 100 cm)
• same parameter values
• Same computational grid (T382 Gaussian) as in
  coupled GDAS
• same terrain height, same land mask
• same land surface characteristics (soils,
  vegetation, etc)
• Applies GDAS atmospheric forcing
• hourly from previous 24-hours of coupled GDAS
• except precipitation forcing (see next line)
• Precipitation forcing is from CMAP precip
  analysis over tropical lands
• temporally disaggregate the 5-day CMAP
  precipitation with GDAS 6-hrly precip
• linearly blend GDAS and CMAP precip forcing
  between 30-40 deg latitude
• Reach-back every 5 days to apply latest 5-day
  CMAP anal
• then reprocess last 6-7 days to maintain
  continuous cycling from CMAP-driven land states
• Same realtime and retrospective configuration
• Once daily update of coupled GDAS soil moisture
  states from GLDAS

22
ONE DAY OF REANALYSIS Note daily GLDAS (spans
prior 24-hrs)
12Z GSI
18Z GSI
0Z GSI
6Z GSI
0Z GLDAS
12Z GODAS
18Z GODAS
0Z GODAS
6Z GODAS
9-hr coupled T382L64 forecast guess (GFS MOM4
Noah)
1 Jan 0Z
2 Jan 0 Z
3 Jan 0Z
4 Jan 0Z
5 Jan 0Z
2-day T382L64 coupled forecast ( GFS MOM4
Noah )
23
GLDAS versus Global Reanalysis 2 (GR2)Land
Treatment

• GLDAS an uncoupled land simulation system
  driven by CMAP observed precipitation over
  tropics
• Executed using same grid, land mask, terrain
  field and Noah LSM as GFS in experimental CFS
• Non-precipitation land forcing is from Global
  Reanal 2 (GR2)
• Executed retrospectively from 1979-2006 (after
  spin-up)
• GR2 a coupled atmosphere/land assimilation
  system wherein land component is driven by model
  predicted precipitation
• applies the OSU LSM
• nudges soil moisture based on differences between
  model and CPC CMAP precipitation

24
GLDAS/Noah (top row) versus GR2/OSU (bottom row)
2-meter soil moisture ( volume) GLDAS/Noah
values are higher Climatology (left column) is
from 25-year period of 1981-2005) May
1st Climatology 01
May 1999 Instantaneous Anomaly
GLDAS/Noah
GLDAS/Noah
GR2/OSU
GR2/OSU
25
GLDAS/Noah (top ) versus GR2/OSU (bottom)
2-meter soil moisture ( volume) May 1st
Climatology 01 May 1999 Anomaly
Observed 90-day Precipitation Anomaly (mm) valid
30 April 99
GLDAS/Noah
GLDAS/Noah
GR2/OSU
GR2/OSU
Left column GLDAS/Noah soil moisture climo is
generally higher then GR2/OSU Middle column
GLDAS/Noah soil moisture anomaly pattern agrees
better than that of GR2/OSU with observed
precipitation anomaly (right column top)
26
Monthly Time Series (1985-2004) of Area-mean
Illinois 2-meter Soil Moisture
mmObservations (black), GLDAS/Noah (purple),
GR2/OSU (green)
The climatology of GLDAS/Noah soil moisture is
higher and closer to the observed climatology
than that of GR2/OSU, while the anomlies of all
three show generally better agreement with each
other (though some exceptions)
27
New T126 CFS Reforecast TestsLand Component
Impact

• -- New Noah LSM versus Old OSU LSM
• -- GLDAS/Noah versus Global Reanal-2/OSU
• -- SST high correlation skill in tropical
  Pacific (not shown)
• -- CONUS precipitation (low correlation skill in
  summer, later frame)

28
CFS Land Experiments (4 configurations)
Experiments of new T126 CFS with Noah LSM and OSU
LSM25-year CFS 6-month summer reforecasts (10
member ensembles) from late-April and early-May
initial conditions (00Z) of 1980-2004Initial
Dates of Ten Members Apr 19-23, Apr 29-30, May
1-3(GFDL MOM-3 Model is ocean component)
29
JJA Precipitation Correlation Skill
CFS/Noah/GR2 case is clearly worst case (least
spatial extent of positive correlation). Remaining
three cases appear to have similar spatial
extent of positive correlation, but distributed
differently among sub-regions. Still
disappointingly small spatial extent
of correlations above 0.5 in all four
configurations.
From hindcasts for years 1981-2004. Ten-member
ensemble mean shown for each panel.
30
JJA Precipitation Correlation Skill
CFS/Noah/GR2 case is clearly worst case (least
spatial extent of positive correlation). Remaining
three cases appear to have similar spatial
extent of positive correlation, but distributed
differently among sub-regions. Still
disappointingly small spatial extent
of correlations above 0.5 in all four
configurations.
From hindcasts for years 1981-2004. Ten-member
ensemble mean shown for each panel.
31
All the configurations of New CFS are superior to
Ops CFS over CONUS (Most likely owing to
inclusion of CO2 trend in New CFS)
10 Members each case (same initial dates)
32
Conclusions from CFS Land-component Experiments

• The relatively low CFS seasonal prediction skill
  for summer precipitation over CONUS is not
  materially improved by the tested upgrade in land
  surface physics and land data assimilation
• Lack of positive impact likely due to more
  dominant influence from SST anomalies and
  internal chaotic noise in the coupled global
  model
• Corollary The use of initial soil moisture
  states with instantaneous soil moisture anomalies
  did not provide an advantage over the
  climatological soil moisture states, provided the
  climatology was a product of the very same land
  model
• Separate study by CPC (Soo-Hyun Yoo, S. Yang, J.
  Schemm) evaluated these same summer CFS
  experiments over the Asian-Australian Monsoon,
  showing modestly positive impact from Noah LSM
  and GLDAS upgrades presented here.
• An upgrade to the land surface model of a GCM can
  possibly degrade GCM performance if the upgraded
  land model is not also incorporated into the data
  assimilation suite that supplies the initial land
  states
• The addition of a CO2 trend to the experimental
  CFS is likely the major source of the improvement
  in experimental CFS summer season surface
  temperature forecasts relative to the currently
  operational CFS
• Future work will carry out this same suite of CFS
  reforecasts for winter season
• One focus will be snow cover prediction (Ops CFS
  has notable low bias in snow cover)

33
CFSRR Land ComponentSummary and Pending Issue

• Motivation for GLDAS
• High tropical precipitation bias in GFS/GDAS
• GLDAS uses CPC CMAP precip anal to force land
  over tropical latitudes
• T382 GLDAS for CFSRR
• Codes and scripts delivered to and executing in
  CFSRR suite
• Low-res 28-yr GLDAS retrospective run done
  assessed
• CMAP precipitation applied globally to force land
  surface
• Non-precipitation land surface forcing from
  Global Reanalysis 2
• CFS land-component summer reforecasts run for
  25-yrs
• Land upgrade not yielding better summer precip
  fcst skill
• Winter reforecast tests are underway
• Pending Issue
• Length of overlap in four CFSRR production
  streams
• I urge at least12-months overlap (6-months is
  insufficient)