Impact of Model Choice

1
Improving Seasonal Hydrologic Prediction at NCEP
via Land Surface Modeling and Assimilation
Advancements and Collaborations in Coupled and
Uncoupled Prediction Systems in the Climate
Prediction Program for the Americas (CPPA)
Kenneth Mitchell and Youlong Xia Environmental
Modeling Center (EMC) National Centers for
Environmental Prediction (NCEP) AND
Collaborators Rongqian Yang, Jesse Meng, Helin
Wei, Eric Wood, Lifeng Luo, Justin Sheffield,
Dennis Lettenmaier, Andrew Wood, Brian Cosgrove,
Charles Alonge, Pedro Restrepo, John Schaake,
Kingtse Mo
ICCES, Institute of Atmospheric Physics Beijing
Normal University
2
CPPA Climate Prediction Program for the
Americas(Predecessor programs GCIP and GAPP)

• CPPA Science Objectives
• Improve the understanding and model simulation of
  ocean, atmosphere and land-surface processes
• Determine the predictability of climate
  variations on intra-seasonal to interannual time
  scale
• Advance NOAAs operational climate forecasts,
  monitoring, and analysis systems
• Develop climate-based hydrologic forecasting
  capabilities and decision support tools for water
  resource applications.

PACS
3
Drought/Hydro Prediction Outline

• CPPA Climate Prediction Program for the Americas
• Seasonal forecast scale is current emphasis
  (sub-seasonal coming later)
• Two strategic approaches (objective,
  reproducible, retrospective)
• Coupled prediction models (plus data
  assimilation/analysis for them)
• Uncoupled prediction models (plus data
  assimilation/analysis for them)
• Coupled Monitoring Prediction (coupled
  atmosphere-land-ocean)
• Global Models
• GFS (NCEP Global Forecast System) medium-range
• CFS ( NCEP Climate Forecast System)
  seasonal-range
• Regional Models
• RCMs (Regional Climate Models)
• Uncoupled Monitoring Prediction (land component
  only)
• Motivation Downscaling, bias correction,
  multiple models
• National focus (but with global potential)
• NLDAS N. American Land Data Assimilation System
• Climate Test Bed NCEP-NCPO Partnership
• Achieve future upgrades and NOAA operations for
  all above

4
Short range to seasonal range
NARR N. American Regional Reanalysis
Launching at 2006 CPPA PI Meeting, August, Tucson
5
Two Dynamical Approaches to Drought / Hydro
PredictionA) Coupled B)
Uncoupled
precipitation
Atmospheric Model (GCM or RCM)
Bias-corrected Precipitation Forecasts (ensemble)
Post Processor Downscaling Bias Correction)
Precipitation
Fluxes

Multi Land Surface Models Noah, VIC, Mosaic, SAC
One Land Surface Model
Runoff
Runoff
River Routing Model
River Routing Model
Stream Flow
Stream Flow
Post Processor
Post processor
Final Products
Final Products
Both approaches should be executed in ensemble
mode.
6
CPPA PI Partners in Drought/Hydro Prediction

• Ken Mitchell NCEP/EMC
• Youlong Xia, Jesse Meng, Rongqian Yang
• Eric Wood Princeton U.
• Lifeng Luo, Justin Sheffield
• Dennis Lettenmaier U. Washington
• Andy Wood, Ted Bohn
• Brian Cosgrove NASA/GSFC/HSB
• Christa Peters-Lidard, Chuck Alonge, Matt Rodell,
  S. Kumar
• Kingtse Mo NCEP/CPC
• Wanru Wu, Muthuvel Chelliah
• Huug Van den Dool NCEP/CPC
• Yun Fan
• Pedro Restrepo NWS/OHD
• John Schaake, DJ Seo

7
Soil Moisture Anomaly Monthly Total
Column (Model by Model and 4-model Ensemble Mean)
Rerun of NCEP Realtime NLDAS for 10 Years
NOAH
MOSAIC
SAC
VIC
Multi-model Ensemble Mean
MARCH 2006
8
NASA NLDAS-based 10-Year Prototype Realtime
Drought MonitorLeft column shows prototype
example from 9 April 2006Web Address
http//ldas.gsfc.nasa.gov/monitor/
NLDAS Mosaic LSM Output
NDMC Weekly Drought Monitor
NLDAS Noah LSM Output
CPC - Leaky Bucket Model
9
Monthly Total Column Soil Moisture Anomaly
(Model by Model and 4-model Ensemble Mean)
NCEP/EMC Realtime NLDAS runs (1996 to present
realtime)
Noah
Mosaic
SAC
VIC
July 2006 large change since 2005 (not shown)
NLDAS Multi-Model Ensemble Mean Anomaly
10
Monthly Snowpack Water Content (SWE)(Model by
Model and 4-model Ensemble Mean) Rerun of NCEP
Realtime NLDAS for 10 years
MOSAIC
NOAH
SAC
VIC
Multi-model ensemble Mean anomaly
SWE Anomaly March 2006
11
Impact of Model Choice
LSM Total Column Soil Moisture (mm), Northern New
York, 1997-2007
750
650
550
450
Total Column Soil Moisture (Noah LSM)
Total Column Soil Moisture (Mosaic LSM)
350
1997 1999
2001 2003
2005 2007
LSM Total Column Moisture Climatology (mm),
Northern New York
750
Average Soil Moisture Climatology (Noah LSM)
690
Average Soil Moisture Climatology (Mosaic LSM)
630
570
510
450
January 1st

December 31st
12
Impact of Model Choice
Mosaic LSM Total Column Soil Moisture
Percentile July 1st, 2007, Based on 28 Year
Climatology
Noah LSM Total Column Soil Moisture
Percentile July 1st, 2007, Based on 28 Year
Climatology

• Choice of land surface model can greatly
  influence depiction of drought severity due to
  differences in model physics and parameterizations

Noah LSM Total Column Soil Moisture
Percentile July 1st, 2007, Based on 28 Year
Climatology
Mosaic LSM Total Column Soil Moisture
Percentile July 1st, 2007, Based on 28 Year
Climatology
D4
D3
D2
D1
D0
13
Soil MoistureRetrospective Forecast for Summer
1988 drought(Princeton East-Wide System An
Example)

Ohio Basin
Lead time
14
From the Princeton University Seasonal Forecast
System
The evaluation of streamflow predictions over
selected gauges. The ranked probability score
(RPS) for monthly streamflow for the first three
months are examined against the offline
simulation. The bars are for CFS, CFSDEMETER
and ESP from the left to the right,
respectively. RPS 01 with 0 being the perfect
forecast 3 tercels, below normal, normal and
above normal with probability of 1/3 each.
15
Example seasonal predictions and verification of
Spring 2007 drought conditions from the Princeton
U. VIC/CFS-based uncoupled seasonal forecast
system.
16
Princeton UniversityCFS-driven 5-month
VIC-based ensemble uncoupled forecast of Dry Soil
Areafor winter-spring-summer 2007

S
West U.S.
Southeast U.S.
17
U. Washington Seasonal Hydrologic Forecast System
18
Drought Prediction at U. Washington
sponsors NOAA TRACS (A. Wood, PI) CPPA program
(D. Lettenmaier, PI)
19
Drought Prediction at U. Washington
The UW Surface Water Monitor platform offers
weekly predictions of soil moisture, runoff and
other variables for the Continental US.
http//www.hydro.washington.edu/forecast/monitor
/
sponsors NOAA TRACS (A. Wood, PI) CPPA program
(D. Lettenmaier, PI)
20
Drought / HydrologicalCoupled Medium-Range
Prediction

• GEFS Global Ensemble Forecast System
• about 80 GFS two-week forecasts run daily

21
GEFS Forecast Precipitation
Week2 Forecast Made 19Aug2007
Week1 Forecast Made 26Aug2007
Verification 27Aug2007-02Sep2007
Large uncertainties over SE for week2 forecast,
overlapping with large errors
22
GEFS Forecast Soil Moisture
Verification 27Aug2007-02Sep2007
Week2 Forecast Made 19Aug2007
Week1 Forecast Made 26Aug2007
Errors corresponding to large uncertainties
23
Drought / HydrologicalCoupled Seasonal-Range
Prediction

• CFS Climate Forecast System

24
CFS Land Experiments 4 configurations Land
Experiments of T126 CFS with CFS/Noah and CFS/OSU
25
(No Transcript)
26
(No Transcript)
27
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 Anomaly
GLDAS/Noah
GLDAS/Noah
GR2/OSU
GR2/OSU
28
GLDAS/Noah (top) versus GR2/OSU (bottom) 2-meter
soil moisture ( volume) May 1st Climatology
01 May 1999 Anomaly
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)
29
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)
30
Noah/ GR2
Noah/ GLDAS
Noah/ GLDAS Climo
OSU/ GR2
10 Members each case (same initial dates)
31
Correlation Skill of Ensemble Mean CFS Seasonal
Forecasts of CONUS Precipitation for 25-year
retrospective CFS summer season predictions of
1980-2004
32
Same as previous frame, except showing areas of
both positive and negative correlation and
replacing one of four panels with histogram of
results from remaining three panels (from Koster
paper)
33
Future Efforts

• Third NCEP Global Reanalysis (1979-2008)
• 3D ocean assimilation included
• 3D land data assimilation included
• New CFS 30-year retrospective forecast
• New global climate model
• Double previous resolution
• Upgraded physics of atmosphere, ocean, land
• CPPA Regional Climate Model seasonal forecast
  experiment (see next frame)

34
New CPPA Seasonal Prediction Experimentwith
Multiple Regional Climate Models (RCMs)

• About half dozen different RCMs
• WRF, RAMS, ETA, RSM, MM5
• RCM domain must span entire CONUS
• Resolution about 25-50 km
• Dynamical downscaling is a key objective
• Driven by CFS seasonal predictions of SST and
  lateral boundary conditions
• Winter prediction experiments first
• Summer experiments later
• 15 forecast members from November initial
  conditions for each of 22 years (1982-2003)
• each member will be forecast through end of April

35
Design of the 30-year NCEP CFSRR
T382L64 Global Reanalysis and T126L64 Seasonal
Reforecast Project(1979-2008)
Suru Saha and Hua-Lu Pan, EMC/NCEP With Input
from Stephen Lord, Mark Iredell, Shrinivas
Moorthi, David Behringer, Ken Mitchell, Bob
Kistler, Jack Woollen, Huug van den Dool,
Catherine Thiaw and others
36
For a new Climate Forecast System (CFS)
implementation Two essential components A new
Reanalysis of the atmosphere, ocean, seaice and
land over the 31-year period (1979-2009) is
required to provide consistent initial conditions
for A complete Reforecast of the new CFS over
the 28-year period (1982-2009), in order to
provide stable calibration and skill estimates of
the new system, for operational seasonal
prediction at NCEP
37
Following Frames Example ofRoutine Monthly U.S.
Drought Monitoring and Predictionvia CPPA
Collaborating PIs

• NCEP EMC
• NCEP CPC
• Princeton U.
• U. Washington
• NASA/GSFC/HSB
• NWS/OHD

38
P anomaly for SON 2007
NCEP-CPPA Drought Monitoring
Nov 2007 was dry for the entire US. except the
Northeast. The lack of rain in Nov off sets
positive anom in Oct over the N-Plains and the
East coast
39
November 2007 anomaly of monthly mean moisture
flux (vector arrows) of total atmospheric column.
Color shading shows the zonal (east-west)
component of the moisture flux anomaly. (Both
fields based on N. American Regional Reanalysis)
Despite LaNina pattern of cool tropical east
Pacific SST during Oct-Nov 07 (not shown), which
usually brings high moisture flux and above
normal precipitation from northeast Pacific into
northwest U.S, the moisture flux in northwest
U.S. in Nov 07 is northerly and weak, which has
precluded above-normal precipitation thus far in
northwest CONUS. (Aside this pattern later
changed in Dec 07, bringing the expected heavy
precipitation to northwest U.S.
40
Ensemble soil moisture ensemble anomaly
NOVEMBER 2007
SON 2007 SM total
SM top 10cm
SM top 40cm
SM total
SM top 1m
SM anomaly was dry for the top 40cm over the
entire US in Nov. At the deep soil level (top 1m
and total), wetness was from Oct rain.
41
Total soil moisture change Nov-Oct
All NLDAS systems give a consistent picture of
depreciation of soil moisture over the Great
Plains and the Southeast SM increased over the
Northeast coast
42
SPI for Nov 2007
Standardized Precipitation Index
SPIs at 6- month or longer all indicate A three
cell pattern with dryness over the Southeast and
Southwest and wetness over the Great Plains.
43
Runoff ensemble mean anomaly
NOV 2007
Negative Runoff anomalies over the Southeast
SON 2007
For the Southeast, SM, runoff and P all indicate
drought
44
Streamflow (USGS)
Last 28 days
A typical La Nina signal
45
U Washington SM runoff3- month lead ESP
forecastsIC Dec 2007
Soil Moisture
Runoff
46
Princeton University
47
Summary

• Uncoupled hydrological monitoring system is
  running at NCEP EMC, and support to CPC drought
  monitor
• Coupled and uncoupled prediction system is being
  installed at NCEP EMC via CPPA collaborators
• Methods and algorithms to generate forcing
  ensembles are still being investigated
• New CFS will use GLDAS reanalysis dataset and new
  Noah model to make seasonal predictions
• Multi-models and multi-ensembles will be applied
  to both 1) generate forcing datasets and 2)
  produce land surface model outputs