1 00/XXXX © Crown copyright Carol Roadnight, Peter Clark Met Office, JCMM Halliwell Representing convection in convective scale NWP models : An idealised.

1 00/XXXX © Crown copyright

Carol Roadnight, Peter Clark

Met Office, JCMM

Halliwell

Representing convection in convective scale NWP models : An

idealised study


Introduction Severe weather conditions

occur on small spatial scales. It would be beneficial to be

able to represent these phenomena in NWP models.

Nature, location and organisation of convection not predicted very well by the current mesoscale model

– the convection scheme only represents average effects of an ensemble of clouds.

Difficulty in representing convection in models that aim to resolve scales between 1km and 10km is well known.


Introduction cont...

In the absence of an appropriate scheme decrease the grid length so that convection is resolved?

Introduction of non-hydrostatic dynamics into the Met Office UM possibility of using high resolution grids.

Current computer power restricts us to grid lengths of 1km - 10km subgrid representation of physical processes still required.

JCMM: research into problems associated with modelling and Data Assimilation with partially resolved convection.


Do we need a convection scheme?

Target grid length 4km - 1km

– convection partially resolved.

Current convection scheme (Gregory Rowntree mass flux scheme) not designed for these resolutions.

We have option of using

– vertical mixing: convection scheme

– horizontal mixing: diffusion scheme.

– other schemes.


Does the addition of the convection scheme and diffusion have an effect on the scales in the system?

Do the choices of diffusion coefficient and CAPE closure timescale determine these scales?


Idealised UM Uses same dynamics as full UM but allows physics

schemes to be “switched off”

Allows us to study processes of interest in controlled conditions.

Radiative-convective equilibrium set up.

– Bi-periodic domain: 4km grid length, 64x64 points in

horizontal. 50 vertical levels.

– Constant cooling profile

– Run until equilibrium

– Compare with LEM

– Higher resolution reference run


Does the scheme behave in a physically

sensible way? Grid length CCT


With and without convection scheme














4km, no diffusion


4km, no diffusion


4km, no diffusion


4km, no diffusion


4km, no diffusion


4km, no diffusion


4km, no diffusion


2km, no scheme


2km, no scheme


2km, no scheme


2km, no scheme


Scheme and diffusion












Summary of results


Summary of results


Summary of results


Summary of results


Summary of results


Summary of results


Summary of results


Summary of results


Can the scheme be modified to be more scale selective?

Scales observed in system controlled by CCT?

Modify convection scheme

– If size of cells related to CAPE limit intensity of the scheme when CAPE is large enough to suggest cells are resolved

– Limit mass flux


Summary of results


Summary of results


Summary of results


Summary of results


Summary and conclusions Does the addition of the convection scheme

and diffusion have an effect on the scales in the system?

– Convection scheme: No significant changes with grid lengths 2km, 4km unless CCT was reduced, in which case larger scales more dominant.

– Diffusion: had more impact, damping small scales.

– Both schemes: More significant impact.


Do the choices of diffusion coefficient and CAPE closure timescale determine these scales?

– CCT? Larger scales observed as CCT decreases (monotonic).

– Diffusion coefficient? Smaller scales damped as diffusion increases (monotonic).

– Both schemes together: more complex.

– More quantitative analysis needed.


Future.. Some type of scheme required.

– Modified version of current scheme

– Shallow convection scheme

– other convection schemes

– turbulence scheme

Targetted diffusion



Current UM convection scheme Gregory-Rowntree mass flux scheme aims to represent all

types of moist and dry convection.

Effects of ensemble of convective plumes characterised by

– total mass flux associated with updraughts

– average thermodynamic properties

Characteristics obtained using cloud model based around parcel theory and modified by entrainment/detrainment.

CAPE closure scheme determines amount of convection

– mass flux calculated based on removal of CAPE over

– mass flux CAPE/ 30 minutes, 60 minutes


Power spectra of vertical velocity

60 30

min min

20 10

min min


Effects of diffusion


CAPE dependent CCT

LS/totalCCT=60 min 0.32CCT=30 min 0.08CCT=10 min 0.04Function 1 0.92Function 2 0.84

Function 1

Function 2


Summary of results

qcl cmf


CCT=60min and diffusion










“half” diffusion and scheme















1 00/XXXX © Crown copyright Carol Roadnight, Peter Clark Met Office, JCMM Halliwell Representing convection in convective scale NWP models : An idealised.

Documents

xxxx crown copyright

diffusion slide

diffusion scheme

idealised study slide

appropriate scheme

grid length cct slide

limit mass flux slide

organisation of convection