Recent Results and Near-FuturePlanned Changes for the Noah LSM

at NCEP

Ken Mitchell, Mike Ek, Helin Wei, Vince Wong, Youlong Xia, Rongqian Yang, Jesse Meng,

Yihua Wu, Weizhong Zheng

Unified Noah LSM Development WorkshopNCAR

25-26July 2007

Strategic Issues of this Noah LSM Workshop

(for later workshop reference: not subject of this talk)

• Migrating all Noah developers to “Unified Noah” in public WRF Repository

• Centralized “version control” Noah software• Migrating to ESMF and LIS

– Maintaining ESMF compatibility in WRF repository

• Coordinating Noah development paths• Using new satellite-derived land surface fields• “Strategic” (major) Noah physics improvements

– See next frame for examples

• “Tactical” (modest) Noah physics improvements– This is emphasis of remainder of this talk and emphasis of

current Noah upgrades in NCEP Land Team

Strategic (major) improvements to Noah physics:often by collaborators outside of NCEP and NCAR

(not subject of this talk)

– Groundwater state and Topmodel approach (U. Texas)– Multi-layer snowpack (U. Texas)– Multi-layer vegetation canopy (U. Texas)– Ball-Berry canopy resistance (Purdue U.)– Snow albedo treatment (U. Arizona)– Irrigation treatment (NASA HSB)– Van Genuchten soil hydraulics (NASA HSB)– Spatially varying soil layer thickness– Unified surface layer treatment (over land)– Others (TBD at this workshop)

Recent Results and Plans for Noah LSM at NCEP:Outline of Remainder of Talk

• GLDAS-LIS/Noah: Jesse Meng– T126 and T382– Component in next NCEP Global Reanalysis

• Climate Forecast System (CFS): Rongqian Yang– Impacts of Noah and T126 GLDAS

• Global Forecast System (GFS):Helin Wei, Weizhong Zheng– Impact of T382 GLDAS– Impact of new weekly realtime global vegetation fraction (GVF)– Surface layer scheme: Zot changes

• N. American WRF model (NAM): Mike Ek, Vince Wong– Unified Noah: Impacts– Surface layer scheme: Zot changes

• NLDAS: Youlong Xia, Helin Wei– improve Noah physics (seasonal LAI, canopy resist., snow albedo, sfc layer, infiltration)– Add seasonal prediction capability (with U. Washington and Princeton U.)

• Hurricane WRF Model (HWRF): Yihua Wu– Replace surface slab model with Noah LSM

• PILPS:– San Pedro (Luis Bastidas) and LBA (U. Texas/Yang & Niu)

• Joint Center for Satellite Data Assim (JCSDA): Weizhong Zheng– New satellite land surface fields (mostly MODIS)– Land surface skin temp (LST) and land surface emissivity

CFS Land Experiments: 4 configurations Land Experiments of T126 CFS with CFS/Noah and CFS/OSU

Choice of Land Model

“GR2” denotes NCEP/DOE Global Reanalysis 2

Choice of

LandInitial

Conditions

GR2/OSU (CONTROL)GR2/OSU

GLDAS/Noah

GLDAS/Noah Climo

CFS/Noah CFS/OSU

• 25-year (1980-2004) 10-member 6-month T126 CFS runs ( GFS-OP3T3, MOM-3 )

– Four configurations of T126 CFS:• A) CFS/OSU/GR2: - OSU LSM, initial land states from GR2 (CONTROL)

• B) CFS/Noah/GR2: - Noah LSM, initial land states from GR2

• C) CFS/Noah/GLDAS: - Noah LSM, initial land states from T126 GLDAS/Noah

• D) CFS/Noah/GLDAS-Climo: - Noah LSM, initial land states from GLDAS/Noah climo

• Initial conditions: 00Z daily from Apr 19-23 29,30, and May 1-3

Key Theme From Results that Follow:Configuration B is clearly the worst configuration for CONUS JJA precip

CFS Summer-Season Land ExperimentsObjective: Demonstrate Impact on CFS of: A) new land model (Noah LSM vs OSU LSM)

B) new land initial conditions (GLDAS vs GR2)

Soil Moisture Comparison:

T126 GLDAS/Noah vs. T62 Global Reanalysis-2/OSU

Monthly Time Series (1985-2004) of Area-mean

Illinois 2-meter Soil Moisture [mm]:

Observations (black), GLDAS/Noah (purple), GR2/OSU (green)

Total

Anomaly

Climatology

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)

Top: observed 90-dayPrecipitation Anomaly(mm) valid 30 April 99Bottom: Climatology

GLDAS/Noah (top) versus GR2/OSU (bottom)

2-meter soil moisture (% volume)

May 1st Climatology 01 May 1999

Anomaly

Left column: GLDAS/Noah soil moisture climo is generally higher then GR2/OSUMiddle column: GLDAS/Noah soil moisture anomaly pattern agrees betterthan that of GR2/OSU with observed precipitation anomaly (right column: top)

GLDAS/Noah GLDAS/Noah

GR2/OSU GR2/OSU

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/NoahGLDAS/Noah

GR2/OSU GR2/OSU

JJA Precip Correlation Skill w Different LSMs and ICs

10 Members each case (same initial dates)

Worst

Noah/GLDAS Noah/

GR2

Noah/GLDAS Climo

OSU/ GR2

JJA Precip Correlation Skills South America

Best

OSU/ GR2

Noah/GR2

Noah/GLDAS

Noah/GLDAS Climo

JJA Mean SST Correlation Skill w Different LSM/ICs

Globally

Noah/GLDAS

Noah/GLDAS Climo

OSU/ GR2

Noah/GR2

10 Members each case (same initial dates)

System: dew. Daily execution.

Software: LIS (Land Information System)

Land surface model: Noah (as in operational GFS/GDAS)

Forcing: GDAS hourly sflux CMAP precip (1-7 days behind realtime)

Output: Global land states validate at 00z. GLDAS-based gdas1.t00z.sfcanl

Realtime T382 GLDAS/Noah

CONUS CPC Precip Anomaly

http://www.cpc.ncep.noaa.gov/cgi-bin/anom_realtime.sh

CONUS GDAS vs CMAP precip

Up to 50 mm differences in30-day precip accumulationof 200 mm.

CONUS GDAS/SMC

CONUS LIS/SMC – GDAS/SMC

The Impact of using GLDAS/Noah initial land states on GFS Forecasts(T382 GFS Static Runs)

Experiments:

GFS/GLDAS: 31 static 6-day T382 forecasts with GLDAS/Noah land states as initial conditions

Period: May 4-June 3, 2007

Hypothesis:

GDAS tends to overestimate precip over

eastern half of CONUS in warm season-->wet

soil moisture-->high evaporation-->cool

temperature

With CMAP precip in T382 GLDAS:

less precip over east half of CONUS-->lower

soil moisture-->lower evaporation-->warmer

temperature

As in previous slide, but for CONUS

Daytime (18-21h) 40-100 cm Soil Moisture averagedfrom 20070511 to 20070610

tested GFS – ops GFS

Mean Daytime (18-21h) Surface Latent Heat Flux (W/m**2)

from 20070511 to 20070610

tested GFS – ops GFS

CONUS:Precip scores

12-36hThreatScore

Bias

07 May – 03 Jun

Valid period: 20070507-20070603

East CONUS

West CONUS

New NESDIS fields of weekly realtime global green vegetation fraction (GVF)

and their impact on

T382 GFS forecasts

GVF Data Set: (i) Old climatology GVF monthly data (Gutman)

(ii) Multi-year mean GVF weekly data (24 years: 1982-2005)

(iii) Real-time weekly GVF data (from 1982 to present)

GFS Sensivity Case Selection:Summer 2006: August 1, 2006

Global GVF Data: August 01, 2006

• Old climatology GVF data (Gutman);

• Mult-year mean GVF weekly data (24 years);

• Difference of two climatology datasets;

• Anomaly of real-time weekly GVF data;

Climatology: for Aug 01

Old GVF: Upper leftNew GVF: Upper right

Difference (New-minus-Old):Lower right

New GVF is generally higher

Climatology: for Aug 01

Old GVF: Upper leftNew GVF: Upper right

Difference (New-minus-Old):Lower right

New GVF is generally higher

CONUS version of previous frame

Top Panel:Climatology of New GVF for Aug 01

Bottom Right Panel:Realtime Anomaly for 01 August 2006:(New Weekly Realtime GVF minus Climatology of New Realtime GVF)

Bottom Left Panel:CONUS version of bottom right panel

Two Sensitivity Tests with GFS: August 01, 2006

• (1) Impact of climatology difference:

– Control: Execute GFS with climatology of old GVF product

– Test: Execute GFS with climatology of new GVF product

• (2) Impact of weekly anomaly:

– Control: Execute GFS using climatology of new GVF product

– Test: Execute GFS using the realtime weekly field of new GVF product

• GFS: 7 days forecast, starts from Aug.01,2006.

• Only Experiment 1 is completed to date and its results shown in next frame

(Plots of Test-minus-Control Differences: Tsfc, Ta, Sensible heat flux and Latent heat flux at 1800 UTC for GFS 36-

hour forecast valid at 1800 UTC from 0000 UTC initial conditions on 01 August 2006)

•Test-minus-Control Difference Fields: Tskin (upper left), 2-m air T (upper right),

•Sensible heat flux (lower left), Latent heat flux (lower right)

GFS 36-hour forecast valid at 1800 UTC from 0000 UTC initial time on 01 Aug 06)

Where differencesare significant,higher GVF valuesin new vs old climoResult incooler Tskinand higher surfacelatent heat flux.

Mike Ek will present recent work with Unified Noah LSM in

WRF/NMM (NAM) mesoscale model

plus:

1 - surface layer test results in NAM2- iterative surface energy balance in Noah

NLDAS Phase 2:

1 – Add seasonal prediction component

2- Employ improved versions of Noah (and VIC, Mosaic, SAC)

3- Add CLM

Following frames focus on improvements to Noah LSM:-- seasonal LAI-- canopy resistance (vapor pressure deficit function, soil moisture stress)-- infiltration parameters (Ksat, Kdt) as tested in DMIP-2

Impact in NLDAS simulations of a bundle of following seven changes to Noah LSM

• -- replace constant LAI treatment with a seasonally and spatially varying LAI • -- allow a seasonally varying vertical profile of root density

• -- change the function for the vapor-pressure deficit term in the canopy resistance

• -- change the upper threshold of soil moisture at which the vegetation feels a soil moisture deficit

• -- change the minimum stomatal resistance parameter for a few vegetation classes

• -- change the treatment for the diurnal variation of surface albedo

• -- change the single parameter in the formulation for the roughness length for heat (to increase the daytime aerodynamic conductance)– Decrease CZIL from 0.10 to 0.05

Adding Seasonal LAI treatment to Noah:Use current green veg fraction (GVF), scale between

annual min and max values of LAI

Seasonal LAI Approach

LAImin and LAImax are from Brian Cosgrove’s table.

)/()(,

*)(

minmaxmin

minmaxmin

ffffawhere

LAILAILAILAI

LAI Greenness

Exp: scaling LAI (left hand Y-axis) by means of green veg fraction (right hand Y-axis)This example for a winter-wheat location in central Oklahoma

From Chen et al. 1996

Using a narrow range of this tends to overestimate the evaporation during wet periods (spring) and underestimate the evaporation during dry periods (summer).

SMHIGH and SMLOW

BXEXP

To reduce low LH biases during the Summer

Changing humidity stress and soil moisture stress functions in Jarvis treatment in Noah

Bundle of 7 changes to Noah LSM physics reduces Noah’s generally negative runoff bias

BUNDLECONTROL

Normalized bias of mean-annual basin average runoff from 2-year uncoupled Noah LSMsimulations (Oct 97 to Sep 99) for selected unregulated basins in the NLDAS test bed for a control version (left panel) and test version (right panel) of the Noah LSM. The test version (labeled “BUNDLE”) includes the seven physical changes listed in Section 5 of the progress report. The changes in the test version reduce the negative runoff bias in the east half of the CONUS. Those basins in central Great Plains with high positive bias (dark green) are likely basins were streamflow is diverted for irrigation and hence should be discarded from the set of validating basins.

Normalized bias is defined as [(model-minus-observed)/observed]

BUNDLE +GRNDWTR

BUNDLE of 7 Noah changes

CONTROL

Monthly mean diurnal cycle (horizontal axis) by month (vertical axis) of the difference of model-minus observed surface latent heat flux as computed from the average of such differences at 24 ARM/CART flux stations for a 21-month simulation in the NLDAS test bedof three configurations of the Noah LSM described in the text.

Diurnal cycle (x-axis) by month of year (y-axis) of differences between Noah simulated and observed surface latent heat flux

Noah simulations for DMIP-2:improving Noah streamflow simulations

by tuning Ksat and Kdt parameters

New Ksat = 5.5 * (Old Ksat)

New Kdt = 2.59 - 0.044*(new Ksat above)

Mean annual cycle from six years (Oct 96 – Sep 02) of observed and Noah LSM simulated monthlystreamflow for the Blue River and Eldon River basins in Oklahoma. Changes to two Noah runoffrelated parameters recommended by the MOPEX project reduces the low bias in the simulation.

As in previous figure, but for monthly time series spanning six years (Oct 96 – Sep 02).

PILPS San Pedro (semi-arid: S. Arizona)Lead: Luis Bastidas

• See PILPS San Pedro arcticle in most recent issue of GEWEX Newsletter

– Noah simulations fared reasonably well• Runoff timing one of the best of participating models• Runoff magnitude much too low (as in almost all models)• Latent heat flux was low

New Land-Surface Fieldsfrom JCSDA thrusts

To be shown on Day 2 of workshop

New Satellite-Based Land-Surface Fieldsfrom JCSDA thrusts

To be shown on Day 2 of workshop

• Landuse (Mark Friedl, Boston U.)

• Albedo: snow-free (Mark Friedl, Boston U.)

• LAI (Mark Friedl, Boston U.)

• Deep snow albedo (Xubin Zeng, U. Arizona)

• Greenness fraction– Realtime weekly (NESDIS/STAR)– Monthly climo (Xubin Zeng, U. Arizona)

• Land Surface emissivity (Ben Ruston, NRL)

NCEP recommended modestchanges to Noah

• Seasonal LAI

• Modify Ksat and Kdt parameters

• Change humidity stress function in Jarvis

• Modify RSmin for some veg classes

• Modifyy SMCREF and SMCWLT

• Modify Zot for stable conditions