The Emergence of Land-Surface Modelingin Modern-Era NWP:

The NCEP Experience and Collaborations

NWP 50-Year Anniversary Symposium15-17 June 2004

Ken MitchellNCEP Environmental Modeling Center

NCEP: Where America's Weather and Climate Services Begin

Improving Weather and Climate Prediction:Becoming a Complete Earth System Endeavor

1 - ATMOSPHERE: troposphere, stratosphere

(GARP) - initial conditions require atmosphere data assimilation

2 - OCEAN: deep ocean, seas, coastal ocean, sea ice

(TOGA/CLIVAR) - initial conditions require ocean data assimilation

3 - LAND: soil moisture, snowpack, vegetation, runoff

(GEWEX/GAPP) - initial conditions require land data assimilation

Historical Timeline of NCEP LSMs:With respect to NCEP atmospheric models

• 1955-1965: Barotropic Model– no land surface, no radiation, no diurnal cycle

• 1965-1985: Multi-layer PE and LFM models– simple surface friction effect on wind velocity

– surface sensible/latent heat fluxes over ocean only

– assume zero sensible/latent heat flux over land

– no diurnal cycle, no radiation

• 1986-1995: global MRF, regional NGM & Eta models– first viable land surface models included

– bucket model hydrology and slab model thermodynamics

– first diurnal cycle of land surface energy balance & radiative forcing

Historical Timeline of NCEP LSMs:With respect to NCEP atmospheric models

• 1995-2004: global GFS, regional Eta & WRF models– The Oregon State University (OSU) LSM

– The NCEP Noah LSM descendant of the OSU LSM• Four soil layers

– Includes liquid and frozen soil moisture (OHD)

– Vertical profiles of soil moisture and soil temperature

• Explicit vegetation canopy with root zone– Satellite NDVI-based seasonal cycle of green vegetation fraction (NESDIS)

• Snowpack physics, including water content and density– Daily snow cover and snowpack analyses from NESDIS and AFWA

– Dynamic snowmelt and snow sublimation

• Stream network and streamflow simulation

Multi-institution Land-Surface Partners:1990 - present

• Air Force (AFWA and AFRL)• NESDIS Land Team (ORA)• NWS Office of Hydrological Development (OHD)

• NOAA Office of Global Programs (OGP): – GEWEX Programs: GAPP, GCIP, PILPS, ISLSCP

– NLDAS: N.American Land Data Assimilation System

• Six university partners plus above partners

• Many are GAPP/OGP sponsored

• NASA Hydrological Sciences Branch and GMAO• NCAR WRF Land Surface Working Group

– USWRP sponsored

NESDIS Interactive Multi-sensor Snow (IMS) Product: Daily 4-km Snow/Ice Analysis

Used along with AFWA Snowdepth Analysis for the dailyInitialization of snowpack in NCEP global and regional models

28 Feb 2004 13 May 2004

Partitioning of Incoming Solar Radiation

34% reflected to space-- 25% reflected by clouds-- 7% back scatter by air-- 2% reflected by earth sfc

19% absorbed by atmos-- 17% absorbed by air-- 2% absorbed by clouds

47% absorbed by earth sfc

Land Surface Energy Balance (Exp: Monthly mean, mid-day summer, central U.S.)

Sd - αSd + Ld - Lu = H + LE + G

800 - 150 + 400 - 550 = 125 + 300 + 75

Complexity of LSM driven by representation of LE and G

Sd = Downward solar insolation: 800 W/m**2 -αSd = Reflected solar insolation: -150 Ld = Downward longwave radiation: 400 -Lu = Upward longwave radiation: -550 (based on land skin temp)

G = Ground heat flux 75 H = Sensible heat flux: 125 LE = L*E = Latent heat flux (evaporation) 300

Land Surface Water Balance(Exp: monthly, summer, central U.S.)

dS = P – R – E

dS = change in soil moisture content: - 75 mm

P = precipitation: 75R = runoff 25E = evaporation 125

(P-R) = infiltration

Evaporation is a function of soil moisture and vegetationtype, rooting depth/density, fractional cover, greenness.

All terms in units of mm.

Simple “one-layer” slab LSMs of 1985-1995 era at NCEP

Bucket Model for hydrology

Surface Evaporation: LE = B * EP

B = Surface Wetness coefficient (fraction)

EP = potential evaporation:

function of atmospheric conditions

(humidity, wind speed, temperature)

Slab Model (“force-restore”) for ground heat flux

The Surface Wetness Field in the NGM Model(Range: ~ 0.04 – 0.20 )

(values plotted are actual * 100)

Land Surface Evaporation Treatment

in modern-era land models

E Edir Et Ec

WHEREIN:

E = total evapotranspiration from combined soil/vegetation

Edir = direct evaporation from top soil layer

Ec = evaporation from canopy-intercepted precipitation or dew

Et = transpiration through plant canopy via root uptake, and

constrained by the canopy resistance to evaporation

Noah Land Model Prognostic Equations

Soil Moisture:

t

z

Dz

Kz

F

– “Richard’s Equation” for soil water movement

– D, K functions (soil texture)

– F represents sources (infiltration) and sinks (evaporation)

Soil Temperature

C Tt

z

K t Tz

– C, Kt functions (soil texture, soil moisture)

– Soil temperature information used to compute ground heat flux

Vegetation Greenness

April Climatology

Vegetation Greenness

July Climatology

Developed and providedby NESDIS/ORA

-- New NESDIS realtime weekly update now being tested by NCEP

Ground Heat Flux Evaluation in Eta Model using FIFE Field Exp: Slab/Bucket LSM versus Noah LSM

(Betts et al., 1997, MWR)

Validation of surface fluxes of four LSMs vs 15 ARM flux stations.Monthly mean Rnet, LE, H and G for Jan 98 to Sep 99

Improving the Mesoscale NWP Forecastsvia Land-Surface Influences

• NWP prediction improvement goals

- 2 meter air temperature and humidity

- 10 meter wind vector

- PBL T and Td profiles

- convective stability indices

- integrated moisture flux convergence

- precipitation and cloud cover

July 2003 Monthly Mean Diurnal Cycle of 2-m Air Temperature:Obs vs NCEP Models (3) for Midwest U.S.: Eta, GFS/AVN, NGM

NGM

Obs

ETA

AVN

NCEP Eta model forecast during July 1998:

Texas/Oklahoma drought, 24-hour forecast valid 00Z 27 July 1998

In late July1998, after nearly twomonths of self-cyclingthe land states in theEDAS, the Eta modelsuccessfully capturedthe extremely dry soilmoisture (upper left)and warm soil temps(upper right) over theTexas/Oklahomaregion, yieldingforecasts of high 2-mair temps (lowerleft) and deep, dry,hot boundary layersthat verified wellagainst raobs (e.g.,at Norman, OK –lower right). air temperature (2-meter) Norman, OK sonde

(obs=solid, model=dashed)

soil moisture availability (1-m) soil temperature (5-cm)

In the forecast period between the analysis steps of the 12h pre-forecast data assimilation period, at each time step and at each point where observed precipitation is available, we compare Pmod to Pobs, then modify the model’s temperature, moisture, cloud and rain field to be more consistent with observed precipitation.

The Eta Data Assimilation System: EDASA Coupled Land Data Assimilation System with

hourly assimilation of observed precipitationPre-forecast data assimilation period

Free forecast period

Figure 8. (a) 1-15 July 1998 gage-observed total precip (mm), (b) 'snapshot' of hourly Stage IV radar/Gage precip (06Z, 15 July 1998); EDAS total precip of 1-15 July 1998 for (c) control run without precipassim, and (d) test run with hourly Stage IV precip assim; EDAS soil moisture availability (% saturation)of top 1-m soil column valid at 12Z 15 July 1998 (e) without precip assim, and (f) with precip assim.

IMPACT OF HOURLY PRECIPITATION ASSIMILATION IN ETA MODEL

(a) (c) (e)

(f)(d)(b)

OPS EDAS: OPS EDAS:

TEST EDAS: TEST EDAS:

15-DAY OBS PRECIP (1-15 JUL 98)

SOIL MOISTURE

SOIL MOISTURE

15 JUL 98

15 JUL 98

15-DAY PRECIP

15-DAY PRECIP

1-15 JUL

1-15 JUL1-HR STAGE IV PRECIP

25-Year EDAS-based Regional Reanalysis:Example of July 1988 vs. 1993

Difference of observed monthly total precipitation from gauge-only analysis (Higgins and Shi, Schaake personal comm.)

Difference of monthly total precipitation produced by Regional Reanalysis with its precipitation assimilation

Drought of 1988 vs Flood of 1993

SAMPLE LAND-SURFACE OUTPUT FROM RR

DROUGHT

1988July

15 July, 21Z

FLOOD

1993July

15 July, 21Z

soil moisture (percent of saturation) in top 1-meter

DROUGHT

YEAR (1988):

15 July, 21Z

FLOOD

YEAR (1993):

15 July, 21Z

Boundary layer depth [m]

SAMPLE LAND-SURFACE OUTPUT FROM RR

Conclusions

• Land surface modeling has advanced intensely at NCEP from mid 1980’s to present

• Above advancements have benefited greatly from multi-institution and multi-disciplinary partnerships

• These land surface advancements have improved the skill/accuracy of NWP predictions