The Ocean-Land-Atmosphere Model (OLAM)

Robert L. Walko

Roni Avissar

Rosenstiel School of Marine and Atmospheric ScienceUniversity of Miami, Miami, FL

Martin Otte

U.S. Environmental Protection AgencyResearch Triangle Park, NC 27711

David Medvigy

Department of Geosciences and Program in Atmospheric and Oceanic Sciences, Princeton University, Princeton, NJ

Motivation for OLAM originated in our work with the Regional Atmospheric Modeling System (RAMS)

RAMS, begun in 1986, is a limited-area model similar to WRF and MM5

Features include 2-way interactive grid nesting, microphysics and other physics parameterizations designed for mesoscale & microscale simulations

But, there are significant disadvantages to limited-area models

External GCM domain

RAMS domain Informationflow

Numerical noise at lateral boundary

OLAM global lower resolution domain

OLAM local highresolution region

Well behaved transition region

Informationflow

So, OLAM was originally planned as a global version of RAMS.

OLAM began with all of RAMS’ physics parameterizations in place.

Global RAMS 1997: “Chimera Grid” approachLateral boundary values interpolated from interior of opposite grid

Not flux conservative

OLAM dynamic core is a complete replacement from RAMS

Based on icosahedral grid Seamless local mesh refinement

OLAM: Relationship between triangular and hexagonal cells(either choice uses Arakawa-C grid stagger)

OLAM: Hexagonal grid

cells

Downscaling Regional Climate Model Simulations to the

Spatial Scale of the Observations

Terrain-following coordinatesused in most models

OLAM uses cut cell method

One reason to avoid terrain-following grids:Error in horizontal gradient computation (especially for pressure)

P

V

P

P

PV

Wind

Terrain-following coordinate levels

Terrain

Another reason:Anomalous vertical dispersion

Thin cloud layer

iiiiii FgvpVvt

V

2

HVt

)(

QVst

s

)(

V

d

V

P C

R

C

C

vvdd pRRp

0

1

Continuous equations in conservation form

vV

Momentum conservation(component i)

Total mass conservation

conservation

Scalar conservation(e.g. )

Equation of State

Momentum density

)253,max(1

TC

q

p

lat = potential temperature = ice-liquid potential temperature

cvd Total density

/vvs

MVt

dd

dFdVdt m

dFdVdt

dFdVsdst s

.

Discretized equations:

Finite-volume formulation:Integrate over finite volumes andapply Gauss Divergence Theorem

dFdgdvdx

pdVvdV

t iiii

ii 2

d

iiiij

jjjijjii Fgv

x

pVvSGSVv

t

V

2},{1

jjjV

t 1

HVSGSVt j

jjjjj

},{1

QVsSGSVst

s

jjjjjj

},{1

Conservation equations in discretized finite-volume form

V

d

V

P C

R

C

C

vvdd pRsRsp

0

1

cell face area

cell volume

d

(SGS = “subgrid-scale eddy correlation”)

Discretized momentum density is consistent between all conservation equations

Grid cells A and B have reduced volume and surface areaFully-underground cells have zero surface area

A

B

Land cells are defined such that each one interacts with only a single atmospheric level

Land grid cells

Cut cells vs. terrain-following coordinatesHigh vertical resolution near groundDirection of atmospheric isolines

C-staggered momentum advection method of Perot (JCP 2002)

3D wind vectordiagnosed

Normal wind prognosed

iw

iw1

iw6

iw5

iw4

iw3

iw2

iw7

iv1

iv7

iv6

iv5

iv4

iv3

iv2im7im1

im2

im3

im4im5

im6

Neighbors of W point on hexagonal mesh itab_w(iw)%im(1:7)itab_w(iw)%iv(1:7)itab_w(iw)%iw(1:7)

iv

iu

im1

im2

im3

im5 im6

im4

iv1

iv3iv9

iv15

iv13

iv5

iv12

iv16

iv14

iv8

iv4

iv2

iv11iv10

iv7iv6

iw1iw2

iw3

iw4

Neighbors of V point on hexagonal meshitab_v(iv)%im(1:6) itab_v(iv)%iv(1:16)itab_v(iv)%iw(1:4)

im

iw1iw2

iw3

iv3

iv1iv2

Neighbors of M point on hexagonal mesh itab_m(im)%iv(1:3)itab_m(im)%iw(1:3)

RAMS/OLAM Bulk Microphysics Parameterization

• Physics based scheme – emphasizes individual microphysical processes rather than the statistical end result of atmospheric systems

• Intended to apply universally to any atmospheric system (e.g., convective or stratiform clouds, tropical or arctic clouds, etc.)

• Represents microphysical processes that are considered most important for most modeling applications

• Designed to be computationally efficient

Physical Processes Represented

• Cloud droplet nucleation• Ice nucleation• Vapor diffusional growth• Evaporation/sublimation• Heat diffusion• Freezing/melting• Shedding• Sedimentation• Collisions between hydrometeors• Secondary ice production

Hydrometeor Types

1. Cloud droplets

2. Drizzle

3. Rain

4. Pristine ice (crystals)

5. Snow

6. Aggregates

7. Graupel

8. HailH

G

S

A

C

P

R

D

Stochastic Collection Equation

yxymgyxmgx

ytyxtxa

tytx

t

x

dDdDyxEDfDf

DVDVDyDxDxmFNN

d

dr ff

,

42

00

Table Lookup Form of Collection Equation

nynxa

tytxx DDyxJ

tyxEFNNr ,,,

4

,

LANDCELL 1 LANDCELL 2

LEAF–4

fluxes

wgg

wca hca

rvchvc wvc wvc hvc

wgvc1

A wav

C

VV

G2

G1

hav

ravC

G2

S2

S1

rsahscwsc

hca

haswas

wca

wss

wgs

wgg hgg

hgs

hss

rsv

hvswvs

wgvc2hgcwgcrga

hgg

rgv wgvc2

wgvc1G1

longwave radiation

sensible heat

water

Ks = saturation hydraulic conductivity

ys = saturation water potential

rw = density of water

[h / hs ] = soil moisture fraction

b = 4.05, 5.39, 11.4 for sand, loam, clay

z

zF wwgg

32

b

ss

0;

s

b

ss

Water flux between soil layers

Hydraulic conductivity(m/s) Soil water potential (m)

How should models represent convection at different grid resolutions?

Conventional thinking is to resolve convection wherepossible and to parameterize it otherwise.

0.1 1 10 100 Horizontal grid spacing (km)

Deep Convection

ShallowConvection

resolve

resolve

parameterize

parameterize

?

?