slide 1 pa surface iii of iv - training course 2006 slide 1 introduction general remarks model...

PA Surface III of IV - training course 2006

Slide 1

Slide 1

Introduction

General remarks

Model development and validation

The surface energy budget

Soil heat transfer

Soil water transfer

Surface water fluxes

Initial conditions

Snow

Conclusions and a look ahead

Layout


Slide 2

Slide 2

Evaporation: Idealized surfaces

Lake Snow

Wet vegetation Dry vegetation

Bare ground


Slide 3

Slide 3

Potential evaporation, bucket model

Potential evaporation- The evaporation of a large uniform surface, sufficiently moist or

wet (the air in contact to it is fully saturated)

)p,T(qquCE ssksatLLhpot Evaporation efficiency

- Ratio of evaporation to potential evaporation 10

EE pot

Bucket model Budyko 1963, 1974 Manabe 1969

properties soil and ground) bare vs.n type(vegetatiocover

soil offunction defined wella be should Ideally,

1

critcrit

crit

1

0

crit


Slide 4

Slide 4

A general, algebraic formulation

Two limit behaviours

- Bare soil: Evaporation dependent on soil water (and trapped water vapour) in a top shallow layer of soil (~ 20 mm).

- Vegetated surfaces: Evaporation controlled by a canopy resistance, dependent on shortwave radiation, water on the root zone (~ 1-5 m deep) and other physical/physiological effects.

)cover soilsoil, theof nature and state,,(

),(

, sLsL

ssksatsLLLh

Tqfa

pTqaqauCE


Slide 5

Slide 5

Transpiration: The big leaf approximation

Sensible heat (H), the resistance formulation

Lha

aa

a

skLpskLLhp

uCr

smrr

r

TTCTTuCCH

1

,resistance caerodynami

)(

1

Evaporation (E), the resistance formulation (the big leaf approximation, Deardorff 1978, Monteith 1965)

LLL uqT ,,

ar

crcqE H

skT

ar

resistancecanopy

conditions saturatedfor

ie, stomata, theofinterior

for thehumidity Specific )(

)(

c

sksatc

ca

sksatL

ca

cL

r

Tqq

rr

Tqq

rr

qqE


Slide 6

Slide 6

Some plant science

rs, the stomatal resistance of a single leaf. Physiological control of water loss by the vegetation. Stomata (valve-like openings) regulate the outflow of water vapour (assumed to be saturated in the stomata cells) and the intake of CO2 from photosynthesis. The energy required for the opening is provided by radiation (Photosynthetically Active Radiation, PAR). In many environments the system appears to be operate in such a way to maximize the CO2 intake for a minimum water vapour loss. When soil moisture is scarce the stomatal apertures close to prevent wilt and dessication of the plant.

)definition afor later (seeindex area Leaf

/

L

dL

LL

rr sc


Slide 7

Slide 7

Jarvis approach(1)

1

deficit saturation surface-near

re temperatusoil 4321min

i

asata

s

ascc

f

eTeD

T

fDfTfPARfrr

3/1 f

2/1 f

1/1 f

Dickinson et al 1991


Slide 8

Slide 8

Jarvis approach(2)

4/1 f

Shuttleworth 1993

capacity) holding(water water soil available

tyavailabili

(ECMWF) modelsmany in

zoneroot in the water soil

dava

pwpdava

capd


Slide 9

Slide 9

Bare ground evaporation

Soil (bare ground) evaporation is due to:

- Molecular diffusion from the water in the pores of the soil matrix up to the interface soil atmosphere (z0q)

- Laminar and turbulent diffusion in the air between z0q and screen level height

All methods are sensitive to the water in the first few cm of the soil (where the water vapour gradient is large). In very dry conditions, water vapour inside the soil becomes dominant

watercm) few (alayer soil Top

soil" theofhumidity Relative"

method

1

1

f

r

TqqE

a

sksatL


Slide 10

Slide 10

TESSEL transpiration

watersoil liquid theof average tedroot weigh a is

1

0

n vegetatiolowfor 0 exp

1,1min

)(

n tile vegetatiohigh/low theis

1

3

14

1

1

431min,

,,,

cap

cappwp

pwp

pwpcap

pwp

DaDa

S

SS

aSs

c

isksataiaic

ai

f

gDgDf

cbR

bRaRf

DffRfLAI

rr

Tqqrr

E

i


Slide 11

Slide 11

TESSEL root profile


Slide 12

Slide 12

TESSEL evaporation: shaded snow

)T(qqrr

E

EEE

i

snsataas,a

asnow

snowveg

tile snow shaded the is

Evaporation is the sum of the contribution of the snow underneath (with an additional resistance, ra,s ,to simulate the lower within canopy wind speed) and the exposed canopy. The former is dominant in early spring (frozen soils) and the latter is dominant in late spring.


Slide 13

Slide 13

TESSEL bare ground evaporation

capliq,1

capliq,1pwp

pwpliq,1

pwpcap

pwpliq,1113

13,minsoilsoil

i,sksatai,asoil

ai

1

0

f

frr

)T(qqrr

E

i

tile ground bare the is


Slide 14

Slide 14

Tiles

Interception layer

Bare ground

Snow on low vegetation

High vegetation with snow beneath

Sea ice / frozen lakesLow vegetation

Open sea / unfrozen lakesHigh vegetation

Sea and iceLand


Slide 15

Slide 15

TESSEL geographic characteristics

Dependent on vegetation type

1 m

Global constant

Root depth

Root profile

Dependent on vegetation type

Global constantsLAI

rsmin

MonthlyAnnualAlbedo

Dominant low type

Dominant high type

Global constant (grass)

Vegetation type

Fraction of low

Fraction of high

FractionVegetation

TESSELERA15Fields


Slide 16

Slide 16

Interception (1)

Interception layer represents the water collected by interception of precipitation and dew deposition on the canopy leaves (and stems)

Interception (I) is the amount of precipitation (P) collected by the interception layer and available for “direct” (potential) evaporation. I/P ranges over 0.15-0.30 in the tropics and 0.25-0.50 in mid-latitudes.

Leaf Area Index (LAI) is (projected area of leaf surface)/(surface area) 0.1 < LAI < 6

Two issues- Size of the reservoir

- Cl, fraction of a gridbox covered by the interception layer

T=P-I; Throughfall (T) is precipitation minus interception


Slide 17

Slide 17

Interception: Canopy water budget

canopy theof bottom at the drainage of rate

waterdintercepte ofn evaporatio

ionprecipitat modified

onintercepti of efficiency

waterdIntercepte

*

*

D

Ec

P

e

w

EcIDEcPet

w

ll

i

l

llllil

D

*Pei llEc


Slide 18

Slide 18

TESSEL: interception

Interception layer for rainfall and dew deposition

soil) top to(input lThroughfal

ionprecipitat by coveredbox -grid of fraction

ionprecipitat modified

IPT

k

k/PP

t

ww,PcemaxI

EcIt

w

*

nllmx*

li

lll


Slide 19

Slide 19

Example: Deep tropics interception

Viterbo and Beljaars 1995

Interception

Evaporation

Interception

ARME, 1983-1985, Amazon forestAccumulated water fluxes


Slide 20

Slide 20

Case study: Aerodynamic resistance (1)

Cabauw (Netherlands), is a grass covered area, where multi-year detailed boundary layer measurements have been taken

Observations were used to force a stand-alone version of the surface model for 1987

The first model configuration tried had z0h=z0m


Slide 21

Slide 21


Beljaars and Viterbo 1994


Slide 22

Slide 22



Slide 23

Slide 23



Slide 24

Slide 24



Slide 25

Slide 25

Runoff and infiltration

Infiltration is that part of the precipitation flux that contributes to wet the soil

Runoff occurs in- Parts of the watershed where hydraulic conductivities are lowest (Horton

mechanism, in upslope areas)

- Parts of the watershed where the water table is shallowest (Dunne mechanism, in near channel wetlands)

Runoff depends on orography, nature and moisture state of the soil, precipitation intensity, and sub-grid scale effects

Runoff in NWP/climate models should be called runoff generation (upon application of a routing algorithm becomes “runoff”)

runoff

roughfall th

oninfiltrati

Y

T

I

YTI

f

f


Slide 26

Slide 26

TESSEL: runoff generation

Surface runoff is based on a maximum infiltration limit concept, but with no sub-grid scale variability of either the precipitation flux or the top soil water content. Physically, it is the Hortonian concept, but applied at the wrong spatial scales (the model grid-box)

Deep runoff is free drainage at the bottom

All computations are performed with soil liquid water only. The frozen fraction of the surface is impervious to vertical water fluxes


Slide 27

Slide 27

Introduction

General remarks

Model development and validation

The surface energy budget

Soil heat transfer

Soil water transfer

Surface water fluxes

Initial conditions

Snow

Conclusions and a look ahead

Layout


Slide 28

Slide 28

Initial conditions: general remarks

There are no routine direct observations of soil temperature and water at a global scale

Long time scales in deeper soil temperature and water: Forecast drifts are possible. A way of controlling model drift (initialisation of soil variables) is needed

State variables (soil temperature and water) are non-linearly related (via the equations of the land-surface scheme) to observations (near-surface atmospheric variables)


Slide 29

Slide 29

Soil water initial conditions

Simple method 1: (Initial values)=(short term forecast values)

- Some form of relaxation to climatology in the land surface scheme (ECMWF prior to 1993) or outside (NCEP global model, NCEP ETA model prior to 1998)

- Problem: Model results difficult to validate

Simple method 2: (Initial values)=(short term forecast values)

- No relaxation to climatology

- Problems: It can lead to drift due to: (a) Errors in the atmospheric forcing; (b) Deficiencies in the land surface scheme. Used at ECMWF August 1993-December 1994

Use of proxy information- Screen level RH (ECMWF, 1994, NUDGING)

- Screen level T and RH [(ECMWF, 1999, Optimal interpolation (OI)]

- Rainfall rates

- Radiometric screen temperature (infrared, microwave)


Slide 30

Slide 30

Soil moisture increments

The nudging coefficient, D, is such that a specific humidity analysis increment of 1.5 g/kg fills 150 mm of water in the soil in 9 days (24 days in ERA15)

Exception conditions: (a) Snow; (b) ; (c)

Operational at ECMWF during 1994-1999. Used for ERA15

ECMWF 1994-1999: Nudging

tcoefficien nudging

fraction vegetation

value d)(backgrounforecast

value analysis

hours 6

D

C

x

x

t

qqtDC

v

b

a

bavba

cap pwp


Slide 31

Slide 31

Optimal Interpolation vs nudging (1)

Use is made of temperature (Ta) and relative humidity (RHa) information (instead of specific humidity only)

Increments of soil moisture linked in an optimal way to Ta and RHa increments, with coefficients depending on location (instead of an empirical global coefficient used in nudging)

Increments decrease in no information situations: low zenith angle, high wind speed, precipitation, cloudiness (instead of the constant value used in nudging)


Slide 32

Slide 32

Optimal Interpolation vs nudging (2)

OI Nudging


Slide 33

Slide 33

Soil moisture increments

Douville et al 2000

Nudging

OI


Slide 34

Slide 34

Temperature/humidity errors

Douville et al 2000

T

RH


Slide 35

Slide 35

LE and soil moisture

Douville et al 2000

LE

Soil water, root zone

Soil water, layer 4


Slide 36

Slide 36

July 1988 increments: NOTILES, Nudging


Slide 37

Slide 37

July 1988 mean increments: NOTILES, OI


Slide 38

Slide 38

July 1988 mean increments: TESSEL, OI


Slide 39

Slide 39

Soil water increments: USA June


Slide 40

Slide 40

Case study: Europe, May-June1994 (1)

ECMWF

German (DWD)

Day 2 forecasts


Slide 41

Slide 41



Slide 42

Slide 42



Slide 43

Slide 43


Observations suggest the model has not enough low clouds, and a warm and dry bias at the surface.

In data assimilation the observations only control the state of the atmosphere. Initial conditions for the surface state variables (e.g., soil moisture) were taken from short-term forecasts. They reflect the errors in the surface forcing (precipitation, radiation and clouds) in those forecasts.

A “free-running” soil moisture will have a inaccurate time evolution, even with a perfect surface model.

Cycling the short-term forecasts for soil moisture is tantamount to an extended model integration.

x

t

xb

xa

Analysis trajectoryBackground trajectory

H(x)

y

time

Atmospheric data assimilation


Slide 44

Slide 44


Control + 2 (one month) data assimilation experiments

- FREWAT Control No initialisation of soil water

- WAT Simple initialisation of soil water

- WATCLD WAT + prognostic cloud scheme (Tiedtke, 1993)

WAT uses information from screen-level humidity and temperature observations. If the short-term forecast parameters are warm and dry, the system produces a soil water increment (ie, adds soil water in the root layer); the converse also happens (Viterbo, 1996).


Slide 45

Slide 45


21 Jun1 Jun

2q Bias over Europe

1 Jun 21 Jun

2T Bias over Europe

FREWAT

WAT

WATCLD


Slide 46

Slide 46


Errors in the surface radiative forcing can affect the soil water in the root zone, which in turn affect the partitioning of surface energy into sensible and latent heat. This has a large impact on the quality of summer forecasts.

The impact exists not only on near-surface parameters but throughout the atmosphere.

It is essential to include in data assimilation a procedure to initialise soil moisture based on indirect usage of observations.

500 hPa Bias Europe

500 hPa RMS Europe

FREWATWAT

WATCLD

slide 1 pa surface iii of iv - training course 2006 slide 1 introduction general remarks model...

Documents