Cloud Resolving Model Studies of Tropical Deep Convection Observed

During HIBISCUS 2004.

By Daniel Grosvenor, Thomas W. Choularton, & Hugh Coe

- The University of Manchester, United Kingdom

With thanks to:

Gerhard Held - IPMET, Brazil;Jorge Gomes – CPTEC, Brazil;Andrew Robinson – UCAM, United Kingdom.

Aims of Work

• To simulate transport of material from lower to upper troposphere by deep convective clouds– Gases, aerosols, water vapour, ice hydrometeors

• Testing and improvement of model– Comparisons to observations

• GCM parameterisations– Data set for development and testing

The LEM Model• Cloud Resolving Model• UK Met Office• Bulk microphysics parameterisation

– 38 conversion processes between:• Vapour, liquid, rain, ice, snow, graupel.

• Double moment for ice hydrometeors• Highly variable resolution:-

– Boundary layer processes– Deep convection

• Periodic boundary conditions

24th February, 2004 Case Study• Large squall line moving from north passes

over Bauru.

Model Initialisation• One sounding for whole domain

– Time forcing possible

• Available soundings:– 09:00 LT Campo Grande– 17:15 LT Bauru– 21:00 LT Sao Paulo

• Bauru sounding fairly stable - no deep convection produced by model– Campo Grande sounding used and model forced

towards Bauru sounding

Model Initialisation

• Squall line initialised using a warm perturbation

• 2-D, 500km domain, 1km resolution

• Sensitivity to aerosol concentration tested

• Comparisons to radar statistics of echo tops and 3.5km CAPPI data

Timeseries of max 3.5km radar reflectivity

Timeseries of 3.5km radar reflectivity modes

Log-Normal Distribution of 3.5km dBZ from 14:00-23:00

Timeseries of Max EchotopsM

axim

um o

f 10d

BZ

rad

ar e

cho

tops

(km

)

Local Time

Ma

xim

um

of

10

dB

Z r

ad

ar

ech

o t

op

s (k

m)

CCN = 720cm-3

CCN = 240cm-3

Radar data

Local Time

Mo

de

of

10

dB

Z r

ad

ar

ech

o t

op

s (k

m)

CCN = 720cm-3

CCN = 240cm-3

Radar data

Timeseries of Echotop Modes

Timeseries of Echotop Variances

Local Time

Va

ria

nce

of

10

dB

Z r

ad

ar

ech

o t

op

s (k

m2)

Log-Normal Distribution of Echotops from 14:00-23:00

-5

-4.5

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

Log

10 o

f N

orm

alis

ed D

istr

ibut

ion

of E

cho

Top

s

Max Tracer at each Height as Percentage of Max Input

CFC-11 tracer measurements from SF-4 (DIRAC, UCAM)

Profile of Mean Liquid Water

Conclusions• Echotop agreement reasonable but lack of high

echotops• Simulated 3.5km dBZ generally too high – related to

above?• 2-D simulations produce highly time variable

statistics• Mean values hard to compare with 2-D simulations –

slices through radar data should be better• Tracer outflow height close to apparent outflow

from observations

Future Work

• Vertical radar slice comparisons (RHIs)

• ECMWF/Meso model for soundings and forcing – comparisons to clouds obtained in these models

• Full double moment scheme

• Vertical aerosol transport

• EMM (Explicit Microphysics Model)