EUROCS WP2: Idealized humidity case for CRM and SCMs
Background
The aim of this part of the EUROCS project is to quantify
the dependence of convection on mid-tropospheric humidity
and compare with various SCMs. For this purpose parallel CRM and SCM
runs of an idealized forced humidity case will be run.
This focus is motivated partly by the observed
associations between convection and mid-tropospheric moisture.
E.g. Johnson (1996, ECMWF Workshop
Proceedings) quotes TOGA-COARE results categorized according to
the type and strength of organized convection. The composite humidity
profiles are around 50% RH for weak convection but 80% for strongly
organized convection. However the causal relations are not fully
understood or quantified. Presumably they lead to some differences
between maritime and continental convection.
Mechanisms probably contributing to the
observed associations include the following:
- Inhibition of convection
through entrainment of subsaturated
environmental air and consequent evaporative cooling
- Direct radiative effect of humidity, which affects convection
through the temperature profile
- Dynamical response to convection,
leading to local ascent and moisture convergence
We need to separate these influences, because each
has quite different implications for parametrization. For instance
the entrainment mechanism depends on the assumed rates of entrainment
(in which there are wide differences between schemes).
Another motivation to study the convective response to humidity
comes from the problem of gridpoint storms in GCMs, when resolved-scale
saturation occurs in a convective situation, because the subgrid
convection scheme is insufficiently active. Typically such schemes are
tuned to prevent
resolved-scale saturation (e.g. Betts and Miller 1996, ECMWF Workshop
Proceedings). Perhaps if we understand better how convection should
strengthen
in a moister atmosphere, we can address this GCM requirement in a more
physical way.
Of course convection is also influenced by other factors including
the temperature profile and boundary-layer q, surface triggers etc.,
but the aim here is isolate
one particular issue.
Methodology
- we want a quasi-steady scenario, to avoid complications of
convective cloud-development (cf. the diurnal cycle case)
- we want to control the T and q profiles, to isolate the
impact of changes in q
These aims can be met by nudging the mean T and q profiles on a
timescale comparable with a typical convective adjustment time.
The nudging of T approximately represents the feedback from large-scale
dynamics which tends to oppose convective heating ("Q1").
Randall and Cripe (JGR 1999) discuss different ways of setting
up a CRM/SCM case.
Pilot CRM runs
Pilot simulations have been conducted with the Met Office CRM.
These were based on
- Strong (essentially instantaneous) nudging of T and q profiles.
- The same T-profile between cases
- Varying q-profiles above 2km
- Simple warm-rain cloud-physics (Kessler)
- No mean wind
A statistically steady state developed within 24 hours. With
50% RH above 2km only shallow convection was observed, whereas with higher
values progressively stronger deep convection was found.
See ARM-GCSS talk (presented at NOAA, November 2000).
Case specification
- Domain depth = 15km
- Surface pressure = 1000 hPa
- The CRM will be run at 0.5km horizontal resolution
and non-uniform vertical resolution. The horizontal domain
will be 48kmx48km.
- The SCMs should use their own preferred resolution
- The CRM will be run with 3-phase cloud physics including
ice, snow and graupel. SCMs should use full cloud physics
where available.
- There will be no radiation in the CRM and the SCM radiation schemes
should be switched off.
- Wind, temperature and humidity profiles in the CRM will be nudged
towards target profiles using a relaxation time of 1 hour.
However this relaxation will be imposed only at heights
above 1km .
- The sea-surface temperature (T=theta) will be 294K. The surface will
be defined to have roughness lengths 0.1m for wind, temperature
and humidity.
- The target temperature profiles
for nudging will be defined through potential temperature
theta.
Specifically: thetatarg=293K at z=1km, increasing linearly
with height up to 335K at z=15km.
Below 1km, thetatarg may be
interpolated to the surface but in this layer it will not be used.
- A specific humidity profile qtarg is prescribed by combining
a relative humidity target with thetatarg.
The target relative
humidity is specified as 80% for z between 1km and 2km.
Above 2km, rhtarg is specified as (a) 25% (b) 50% (c) 70% (d) 90%.
Here RH=max(RHw,Rhi).
- The wind target profile utarg=(0.5ms-1)*ln(1+z/z0)
- Initial mean profiles should match the target profiles.
- The simulations will be run for 12 hours, by which time they are
expected to be statistically steady
Diagnostics
- The diagnostics and format should be similar to the
GCSS-WG4 case 3.
- The diagnostics should be averaged over 3 hours.
Timetable
- For the meeting on 18 January 2001, CRM results will be available
- I hope to have results also to present from the Met Office/
Hadley Centre SCM
- Other SCM participants are asked to try to run the above cases
by 18 January and summarize results and issues arising, even
if we cannot do a formal intercomparison by 18 Jan.
Last updated on 17.12.2000 by Steve Derbyshire |
shderbyshire@meto.gov.uk