Atmosphere-landsurface modelling

Mathl. Comput. Modelling Vol. 21, No. 9, pp. 5-10, 1995 Copyright@1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0895-7177/95 $9.50 + 0.00

Pergamon

08957177(95)00045-3

Atmosphere-Landsurface Modelling A. HENDERSON-SELLERS* Climatic Impacts Centre, Macquarie University New South Wales 2109, Australia annQcic.mq.edu.au R. E. DICKINSON Institute of Atmospheric Physics, University of Arizona Tucson, AZ 85721, U.S.A. robtedQhail.atmo.arizona.edu A. J. PITMAN Climatic Impacts Centre, Macquarie University New South Wales 2109, Australia andyQcic.mq.edu.au Abstract-The atmosphere-landsurface interface is the locus of the vast majority of all human activities: we live at this interface. Improved modelling of the exchanges of energy, water and momentum at this critical boundary depends upon (1) better understandmg of how to scale the physical processes, (ii) better agreement amongst existing numerical schemes and (iii) provision of adequate data with which to validate improved atmospherelandsurface models. This paper reviews progress in these tasks. Keywords-Climate,

Landsurface,

1.

Biosphere, GCMs

INTRODUCTION

Understanding the processes that control the near-surface climate and chemistry is of critical importance to the future well-being of humans. Without fresh water, adequate crops and hospitable weather conditions human activities are severely hampered. Current climate models do incorporate some features of the land surface-atmosphere interface, but development of this aspect of environmental modeling has been slow and today still suffers from a severe lack of appropriate data for initialization and evaluation. Land has a dual role in the climate system. First, it acts as a lower boundary for approximately 30% of the atmosphere, exchanging moisture, momentum, heat, and trace gases. Second, from a practical viewpoint of human requirements, land is the most important component of the climate system. The practical importance of studying land surface-atmosphere fluxes derives from human dependence upon the processes that occur near the continental surface. We cannot farm in a desert, in the tundra or in a rain forest. We must have water, food, shelter and warmth, simply to survive. We must understand how the climate will affect the availability of water resources, agriculture and forestry in order to plan for sustainable development. *Author to whom all correspondence should be addressed. This research was funded in part by grants from the Model Evaluation Consortium for Climate Assessment, the NASA/EOS program, the Australian Department of Environment, Sport and Territories and by the Australian Research Council. This is Contribution No. 93/15 of the Chmatic Impacts Centre Typeset 5 MCM 21:9-B

by &S-w

A. HENDERSON-SELLERSetal.

6

The coupling of land to the atmosphere differs in many ways from that of the ocean. The continents are very much more heterogen~us than the oceans. When wet, it can exchange water with the atmosphere more rapidly than the oceans because of greater surface roughness; but when

dry, it provides

no water

the land surface

means

from or through

the atmosphere:

temperatures

at all to the atmosphere.

that local thermal

conditions

the presence

The relatively

low heat

are much more responsive

or absence

can vary by 10°C or more over the diurnal

capacity

of clouds has a substantial cycle.

The fraction

of

to net radiation effect, and

of solar radiation

reflected (the albedo) varies with type of surface cover, and vegetation and soils have large spectral variations of albedo from generally low values at visible wavelengths to much higher values in the near-infrared. In high latitudes and in temperate latitudes during winter, additional processes become important for land’s role in the climate system. Snow accumulates during the winter season and acts 8s a good insulator, allowing little heat to be exchanged between soil and atmosphere. It enhances

the surface albedo and, by this reflection

effect can be considerably soil impedes

weakened

by overlying

of solar radiation, vegetation.

delays spring warming.

Seasonally

or permanently

This frozen

the flow of water into the soil.

2. ISSUES OF SCALING VALUATION OF CURRENT

AND MODELS

For current, or future, landsurface parameterization schemes to be of use, their performance must be evaluated. This task demands coordinated intercomparison and improvement in and synthesis of current observational programs of exchanges of momentum, energy, water and trace gases. Observations range in scale from smaller than 1 m-10 m (a plant) through AVHRR Local Area Coverage (LAC) pixels of size 1 km to the geostationary satellites and radar pixels of 10 km. Mesoscale models require data aggregated to 30 km while global models need aggregation to 300 km. This wide range in scales of observation and model requirements [l] may seem, at first sight, to be insurmountable. There are, however, ways of focusing the problem. Horizontal mixing in the surface turbulent boundary layer homogenizes transfer on scales of 10 x the boundary layer depth (i.e., 10 x loom). Thus inhomogeneities of sizes of 1 km must be considered, but smaller heterogeneities can be neglected for atmospheric modelling purposes. Similarly, mesoscale circulation (sea breezes, slope circulations) are triggered by terrestrial inhomogeneities of size 10 km or larger. Thus, inhomogeneities in features smaller than 10 km can be neglected when co~idering mesoscale circulations. Thus the problems to be addressed become: 1. How to scale energy 2. How to scale energy

and water fluxes from 1 km to 10 km for mesoscale models? and water fluxes from 10 km to 100 km for global models?

These problems of spatial aggregation and integration of data are being answered through detailed field campaigns. The World Climate Research Programme (WCRP) has instigated GEWEX (the Global Energy and Water Cycle experiment). Paradoxically, the atmospheric climate was not believed to be critically dependent upon the distribution or the type of vegetation or soil until ocean models were incorporated into GCMs. For this reason, the development of land-surface-atmosphere exchange schemes lagged behind some other components of climate models in the 1960s and 1970s. In this era, the continental surface was ch~acterized as a flat, bare soil surface with a globally constant “soil moisture capacity.” The only real differences amongst land surface representations at this time were in the specification of surface albedos. Even with these simple schemes, it was shown that the global climate was sensitive to extreme changes in specification, and early models were used to investigate, for example, the possible mechanisms of the Sahel drought of the 1970s. During the 1980s a small number of “complex” land surface schemes were developed and tested in global climate models. These schemes had many similarities and depended upon global data

Atmosphere-Landsurface Modelling

7

sets of soil and vegetation information, the latter often derived from rather inadequate They were primarily concerned with the exchanges of moisture, energy and momentum

sources.

between

the atmosphere and the continental surface. They were developed on the assumption of spatial homogeneity in the surface specification across areas typical of grid ‘Lpoints” in global climate models (rectangles a few hundred examine rather gross disturbances for improved

data for input

field experiments Recently, reasons. coupled

there

kilometres on a side). As such, they could be used only to such as the wholesale removal of tropical forests. The need

and for evaluation

and to the coordination has been

an upsurge

prompted

by these studies

of surface and satellite of interest

in land

derived

surface

gave rise to large-scale information.

simulation

for a variety

of

Schemes designed to compute the exchanges of carbon with the atmosphere have been to the soil-vegetation-atmosphere-transfer schemes to give rise to biophysical models

of land surface processes. At the same time, models which calculate ecology of the continental surface have been linked to climate models to produce “interactive” vegetation prediction schemes which can, in principle, global change

models

make the land surface

a dynamic

component

of the next generation

of

[2].

Recognizing the urgent need to evaluate current landsurface models, the WMO-CAS Working Group on Numerical Experimentation (WGNE) and the Science Panel of the GEWEX Continental-scale International Project (GCIP) (e.g., [3]) h ave agreed to launch a joint WGNE/GCIP Project for Intercomparison of Land-surface Parameterization Schemes (PILPS) [4]. The principal goal of this project is to achieve greater understanding of the capabilities and potential applications of existing, and new, land-surface schemes in atmospheric models. It is not anticipated that a single “best” scheme will emerge. Rather, the aim is to explore alternative models in ways compatible with their authors’ or exploiters’ goals and to increase understanding of the characteristics of these models in the scientific community. The phases of PILPS follow a systematic and staged intercomparison and evaluation procedure: (i) improved and community-wide documentation including intercomparisons and evaluations derived from the existing literature; (ii) intercomparison

by means of the execution

of agreed simulations

by all PILPS

(iii) evaluation against agreed observed data and, ultimately, (iv) coupled intercomparisons using selected host climate and numerical

weather

participants; forecast

mod-

els. PILPS has moved from Phase 1 in which schemes were described and existing sensitivity studies reported (e.g., [5]) to the first stage of Phase 2 in which the same set of stand-alone simulations are being conducted by all the participating models. Using atmospheric forcing data generated from a general circulation climate model, twenty-two participating land-surface schemes (Table 1) were run to equilibrium. Forcing data for a tropical forest and a grassland point were used. The land surface parameters (roughness length, albedo, soil depth, etc.) for both locations were provided for each scheme so that any differences in the results should be due to physical differences between the models rather than differences in the surface characterization. Results for surface temperature, evaporation, sensible heat flux, snow depth and runoff were reported and compared. It is found that there is some agreement between the models in the prediction of annually averaged temperatures with a range, between all the models, of 2.5 K in the case of the tropical forest and 3.8 K for the grassland. Evaporation and sensible heat, averaged over a year, shows less intermodel agreement. Predictions for the tropical forest range from -40 to +50 W rnp2 for the sensible heat flux and f90 to +170 W mV2 for evaporation (Figure 1). For the grassland, predictions range from -22 to +22 W me2 for the sensible heat flux and f25 to +62 W rnp2 for the latent heat flux. In the analysis of monthly and diurnal results, it is shown that there is no consensus between the model predictions of energy fluxes, snow depth or runoff. There is a marginally better agreement in the prediction of temperature. The major differences in the monthly total runoff have implications for the soil moisture distribution and thereby the partitioning of available energy.

J. Lean

CLASS

CSIRO

GISS

ISBA

TOPLATS

LEAF

LSX

MIT

Mosaic-SiB

NMGNFS

CAPS

PLACE

BSTOM

SECHIBA

SSIB

UKMO

VIC

D

E

F

G

H

I

J

Iv1

N

0

P

Q

R

S

T

u

V

_ .._,._,

%emperature,

Y. Xue

BUCKET

A

Kowalczyk

Rasenzweig

Pielke

Bona

Suarez

M

Ducoudre

Milly

Wood

xu

‘also

L1ang

-

treats

D. Lettenmaier

E

K. Lava1

N

C

P Wetzel

M. Ek

L. Mahrt

H. Pan

K. Mitchell

Koster

Entekhabi

R

D

G

S Thompson

R

R. Avissar

E. Wood

*soil moisture,

-_

Mahfouf

J Famiglietti

J-F

J. Noilhan

C

F. Abramopoulos

E

D. Verseghy

Robock

Schlosser

A

Pitman

C

A.J.

R E. Dickmson

BATSlE

BEST

contact

B

Model

A

Key

-..

honzontal

1

1

1

1

1

1

1

wata

__.~_

1

4

2

2

0

15

I

the so11

2

1

3

2

1

2

Z-10

1

3

3

6

7

2

2

6

2

3

1

2

3

Q*

.

for

1

1

1

1

1

2

1

1

2

3

6

3

1

1

6

1

3

1

2

2

Roots

-.

with

soil

law analogy

full energy

,,,

balance

Penman/Mont&h

Penman/Monteitb

Penman/Monteith

PenmanfMonteith

Ohm’s

PenmsnfMonteith

Lumped

Penman/Montelth

Penman/Monteith

Penman/Monteith

Penman/Monteith

aerodynamic

aerodynamic

aerodynamic

Penman/Monteith

-

Penman/Monteith

Penman/Montelth

canopy

or

.

,

balanaz

diffusion

diffusion

heat

__.

diffusion

heat diffusion

force-restore

force-restore

force-restore

(tmwdependent

-

heat ddfusion

heat

heat diffusion

force-restore

aerodynamic

heat

heat diffusion

heat

instantaneous

force-restore

.-

Temperature

Philosophy

-

equations

surface

for

Law

+ variatnn

Law

Law

Vries

Law

Phil&de

diffusion

diffusion

Choisnel

bucket

._

Vries’

,_

+ variatmn

force-restore

for heat & mcnsture)

Darcy’s

Day’s

Phdlpde

Day’s

Philip-de

Vries’

Law

force-restore

Darcy’s

Vnes

Law

force-restore

Day’s

bucket

Philipde

Darcy’s

Soil moistwe

;

et al. (1984,

._

Warrilow

Xue

_

.,I

-

et al. (1986)

et al. (199x) et al. (1991)

(1992)

(1988)

(1984) and Chang

and Pan

Ducoudre

Milly

Wetzel

.

(1990)

(1989)

(1988) and Eagleson

and Eagleson

(1990)

Mahrt

Pan

Koster

Entekhabi

(1989)

(submitted)

(1989)

et al

and Pielke

and Wood

and Jacquemin

Abramopoulos

Avissar

Famiglietti

Mahfouf

1993)

et a(. (1988)

et al. (1991)

(1991)

et al. (1993)

et al. (1991)

Abramopoulos

Kowalczyk

Verseghy

Robock

Pitman

Dickinson

Reference

Some basic information describing each scheme is included.

force-restore

who have provided data to PILPS.

of layers

Included

Number

Z-10

1

2

3

6

7

1

2-3

6

3

3

0

3

2

To

flow through

Ye=

Yes

Yes

Yes

no

yes

Yes

Yes

Yes

no

1

yeles

2

Yes

Ye

Yes

Yes

Yes

Yes

no

Yes

Yes

-

1

1

1

1

1

1

0

1

1

treated

layers

Inter-

ception

canopy

Number

Table 1. A list of those models and primary contacts

Atmosphere-Landsurface

Modelling

60

h4

0

N

W

0 .E

?I

o-

60

I

I

80

100

120

140

R

z

,

I

160

180

E

P

-40

L

H

Mean

Annual

Evapotranspiration

(W/m**2)

Figure 1. Annually-averaged sensible and latent heat fluxes at equilibrium calculated ‘off-line’ by participating PILPS schemes for a tropical forest.

3. CONCLUSIONS The dearth of data at appropriate scales remains to be fully addressed, although the upsurge of interest in models of climate and global change has prompted an international effort to intercompare current land surface schemes. This project will serve both science and the wider community by identifying model strengths and weaknesses, and hence, additional data needs. Global databases not now available, but needed to address the land modelling questions, include the properties of vegetation entering into the landsurface schemes, soil properties, especially how much water can be stored and made accessible to the atmosphere, and high resolution topographic data. As such data become available on a subkilometre scale, procedures to scale their effects up to the hundred kilometre scale of the climate models will be crucial. hrthermore, descriptions of precipitation on the time and space scales upon which it actually occurs must be represented in climate models to make proper use of the above information. Without a full understanding of land surface-atmosphere interactions, it will not be possible to manage the Earth’s resources in sustainable development. Future land surface schemes will be a critical component of global and regional environmental models which must serve policy makers and their advisers.

REFERENCES 1. P.G. Jarvis and K.G. McNaughton, Stomata1 control of transpiration: Scaling up from leaf to region, Adv. Ed. Res. 15, l-49 (1986). 2. A. Henderson-Sellers, Continental vegetation as a dynamic component of a Global Climate Model: A preliminary assessment, Climatic Change 23, 337-377 (1993). 3. M.T. Chahine, GEWEX: The global energy and water cycle experiment, EOS 73 (2), 9-14 (1992).

10

A. HENDERSON-SELLERSet al. 4. A. Henderson-Sellers and R.E. Dickinson, Intercomparison of land-surface parameterizations launched, EOS 73, 195-196 (1992). 5. A. Henderson-Sellers, Z.-L. Yang and R.E. Dickinson, Project for the intercomparison of land surface parameterization schemes, Bull. Amer. Meteor. Sot. 74, 1018-1034 (1993)