MultiScale Atmospheric Numerical Modeling and Data Assimilation for Planetary Applications With a Fo

1
Multi-Scale Atmospheric Numerical Modeling and
Data Assimilation for Planetary Applications -
With a Focus on MarsClaire E. Newman, Mark I.
Richardson (PI)(California Institute of
Technology)AIRSP - PI MeetingOctober 3, 2006
2
Overview

• Introduction - motivation for and development of
  the PlanetWRF model
• Results - progress report and selected
  applications
• Users of PlanetWRF
• Website - www.planetWRF.com

3
Our goal to develop a multi-scale, multi-purpose
model for planetary atmospheres
Introduction to PlanetWRF
We have developed a new model based on NCARs
Weather Research and Forecasting (WRF) model -
PlanetWRF
Main advantages of PlanetWRF

• Easily converted to different planets
• Efficient code that may be run in parallel on a
  cluster
• Takes advantage of continuing improvements and
  additions made to the WRF model, including work
  on data assimilation schemes
• May be run as a limited area, global, or
  multi-scale model (using nesting - see next page)

4
Nesting
Introduction to PlanetWRF
E.g., for Mars
Mother domain
Mother domain
or
siblings
Child
Child 2
Grand-child
Child 1
WRF permits 1-way or 2-way nesting
1-way nesting
2-way nesting
Information can travel both ways gt Feedbacks
between scales
Nest is forced at boundaries, but no information
goes in the other direction
5
WRF --gt planetWRF
Introduction to PlanetWRF
WRF is an Earth-based, limited area model, so we
first needed to

• Convert for application to a general planet
• Planet-specific constants moved to single file
• Constants and parameterizations of physical
  processes added for Mars, Titan and Venus (to
  date)
• Changed from Earth-specific to generalized
  planetary timing (Ls)
• Convert to a global mother domain (see also next
  page)
• Map scale factors
• Polar boundary conditions
• Polar filtering

6
More on the main global changes
Introduction to PlanetWRF
1. Map scale factors

dx, dy gap between grid points in map
coordinates, dX, dY actual distance (in
meters),
dX (1/mx) dx dY (1/my) dy
WRF uses a conformal grid mx my m at all
points This cant reach both poles
PlanetWRF uses a regular lat-lon grid BUT gt mx
? my (mx sec?, my 1) gt mx or my where
previously just m

• 2. Avoid instabilities due to E-W points becoming
  close near poles
• CFL (Courant Friedrichs Lewy) criterion gt need
  ? t lt ? x / U
• gt As ? x -gt 0, ? t must -gt 0 also,
  which is very expensive
• We increase effective scale ? x by filtering out
  smaller wavelengths

7
Progress to date
Results

• Basic global and planetary modifications finished
• Rotated pole (see later) modifications finished
• Parallel implementation heavily utilized
• Mars version of PlanetWRF (MarsWRF) working and
  validated
• Titan and Venus versions (TitanWRF VenusWRF)
  working
• Mars and Titan limited area and Mars large eddy
  simulations
• Mars global mesoscale and active dust
  simulations
• First PlanetWRF paper submitted to JGR planets!
• Still to finish nesting in global and limited
  area mode

CITerra _at_ Caltech
8
Global MarsWRF
Results
We have validated MarsWRF against observations,
and recently participated in an intercomparison
of Mars GCMs - e.g., zonal mean temperatures in
northern winter
Pressure (mbar)
From Mars General Circulation Model
Intercomparison, held at the Second Mars
Atmosphere Modeling and Observations Workshop,
Granada, Spain, Feb. 2006, coordinated by John
Wilson
Pressure (mbar)
9
Global TitanWRF
Results
Simulations are lengthy due to Titans thick,
sluggish atmosphere, with long radiative and
dynamical timescales, and 1 year 30 Earth
years gt PlanetWRFs efficient, parallel and
accurate set-up is vital for Titan
Mass streamfunction at
Northern summer solstice (Ls 90º)
Northern autumn equinox (Ls180º)
Pressure (Pa)
Latitude (degrees north)
Latitude (degrees north)
From Richardson et al., PlanetWRF A General
Purpose, Local to Global Numerical Model for
Planetary Atmospheric and Climate Dynamics,
submitted to JGR-Planets
10
Global VenusWRF
Results
We use simplified physics to represent the
Venusian atmosphere - both Venus and Titan
results show strong sensitivity to horizontal
diffusion
From Richardson et al., PlanetWRF A General
Purpose, Local to Global Numerical Model for
Planetary Atmospheric and Climate Dynamics,
submitted to JGR-Planets
11
Limited area MarsWRF and TitanWRF
Results
Methane clouds triggered by orography, using
TitanWRF in limited area mode
Winds in the Hellas basin on Mars, using MarsWRF
in limited area mode
12
Mars large eddy simulation
Results
We allowed convection to develop in MarsWRF, run
at resolutions of a few hundred meters - the
plots show potential temperature
in cross-section showing plumes
in map view showing convection cells, and
Height (km)
13
Rotated pole versus seaming
Results
We originally proposed seaming together two
polar stereographic grids to form an alternative
global model without pole problems This would
have shifted the problem area to the equator,
allowing polar dynamics to be captured more
accurately It was just as effective but easier
to use a rotated pole approach - this is so
simple that its a runtime option in the standard
PlanetWRF code
Model grid rotated through 90º
N. pole
N. pole
Regions requiring filtering
14
Rotated pole results
Results
15
Users of planetWRF
Users
Us (of course!), and other model users in close
collaboration with us (at Caltech, Kobe, JPL,
Cornell, ) Us in collaboration with NCAR, e.g.,
data assimilation for Earth and Mars (one of our
main intended applications of MarsWRF) Also,
GlobalWRF will soon be provided within the NCAR
WRF release Other MarsWRF users e.g., NCSU /
NIA / Nasa Langley - some may add significant
code to PlanetWRF, such as adaptive meshing Many
current and potential users of model output
e.g., engineers, designers of future missions,
planetary geologists, general public..
16
www.planetWRF.com
Website
We are currently adding content to our new
website, including latest results and
presentations, and a list of current users