FIM Cloud Visualization for SOS and TerraViz
Steve Albers
Team Members
Jebb Stewart
Eric Hackathorn
Julien Lynge
Bob Lipschutz
Judy Henderson
Visualization Overview – Simulated VIS
• Assume both sun and viewpoint are overhead
at all points on the sphere
• Cloud Albedo derived from model data is
combined with multi-spectral land albedo
inferred from NASA’s Blue Marble image
• Simulated VIS provides best realism to produce
an “animated Blue Marble” (more so than IR)
• Physically and Empirically based for best
efficiency and reasonable accuracy
Visualization Technique• Vertically integrated hydrometeor fields are
computedo Cloud liquid, cloud ice, rain, snow, graupel (as
available)
o A.k.a. liquid/ice water path (LWP/IWP)
o Units can be either kg/m**2 or m (based on water
density)
o FIM presently groups everything into cloud liquid
• Convert LWP/IWP into optical depth
o Use typical values of droplet/crystal size (re) and
density for each type of hydrometeor (ρ)
o Account for lower density of snow or graupel
o τ ≈ (1.5 LWP)/(reρ) (Stephens 1978, with ρ term
added)
Scattering of Sunlight by Clouds & Precip
•Calculate fraction of incident sunlight scattered
upward
○ Based on optical depth and backscattering efficiency
•Backscattering efficiency
○ Ratio of backscatter coefficient to extinction coefficient
○ .063 for liquid, .14 for cloud ice or snow, 0.3 for
graupel
○ Low values explain why clouds can look opaque yet still
darker gray as seen from above
Cloud / Precip Scattering - II• Apply equation to yield cloud albedo (a) using
optical depth (τ) and backscatter efficiency (b)
○ a = τ / (τ + 1 / b)
○ Reproduces figure in Mishchenko et. al. (1996) within a
few % (for non-absorbing clouds)
○ Works with cloud liquid and cloud ice (random fractal
crystals)
○ Reduces to expected relationship: a = τ × b for small
values of τ
Cloud / Precip Scattering - III
•Cloud albedo definition and assumption
○ Top Of Atmosphere (TOA) albedo (fraction of sunlight
scattered upward by clouds) assuming dark surface
•TOA albedo (at) used for visualization
○ Combine cloud albedo (ac) derived from FIM with
ground albedo (ag) from NASA’s Blue Marble image
○ Consider cloud semi-transparency and multiple
reflections between ground and cloud
○ at = ac + (1 – ac)2 × (ag ⁄(1-acag))
○ Equivalent to equation (12) in Stephens (1978)
○ Using just cloud albedo (ac) in a linear fashion with ag
(e.g. in TerraViz) would introduce a further
approximation
Use of Blue Marble Image•Pros
○ The high resolution Blue Marble image allows for finer
detail to be shown (10km resolution), compared with
the FIM at ~13km
○ Allows for visualization in color
○ Blue Marble RGB values accurately convert to albedo
(still to do)
○ Allows for more accurate blending of cloud albedo and
ground albedo, compared with a simple overlay
○ Relatively simple and efficient, already demonstrated
•Questions / Cons
○ Wouldn’t use TOA (total) albedo or outgoing short-
wave (if available) from model
○ Wouldn’t allow “progressive disclosure” compared with
an overlay
Potential use of cloud overlay
•Pros
○ Can allow very high resolution (much less than 10km)
for land
○ Ease of use with layering in TerraViz
•Questions / Cons
○ Can it “blur” the land underneath translucent clouds
when scales go finer than ~10km?
○ Can it consider more accurate calculation of TOA (top
of atmosphere albedo), based on cloud & ground
albedo?
Implementation
•Case study with 3-D cloud water from FIM
○ Shown recently at AMS conference
○ 700MB of input for each of 168 time steps
○ Takes ~2.5 hours to process in IDL on SOS server
○ Will rerun to include recent refinements
•Speedup of processing for real-time runs
○ Precalculate vertically integrated hydrometers (i.e.
LWP)
○ About 40 times less data for ITS to write to /public
Implementation - II•Should we switch to a more rigorous RTM?
○ Would be more somewhat more accurate, particularly
when NIM comes online with improved microphysics
○ Would TOA or cloud albedo already be available from
RTM output within the FIM/NIM?
○ Can it be configured for a “sun/viewer always
overhead” setup
○ Consider just cloud albedo output to merge with higher
resolution multi-spectral land surface (e.g. Blue
Marble) data, or alternatively with “progressive
disclosure”
○ Would it allow for visualization in color (if multi-
spectral radiances are available)?
○ Separate radiation package (e.g. CRTM)?
○ What are computational resource needs?
References
Stephens, G., 1978: Radiation Profiles in Extended Water Clouds. II: Parameterization Schemes. J. Atmos. Sci. , 35, 2123-2132
Mishchenko, M., Rossow, W.B., Macke, A., 1996: Sensitivity of cirrus albedo, bidirectional reflectance, and optical thickness retrieval accuracy to ice particle shape. JGR, 101, 16973-16985
Backup Slides
Other Wavelengths?•11μ IR also shown at AMS
○ Less physically consistent with Blue Marble (visible)
○ Simplified approach works well for brightness
temperatues
○ Model OLR converted to brightness temperature with
Stefan-Boltzmann relationship, then a linear correction
applied
○ Agrees within 5-10K with satellite observations
•3.9μ, 6.7μ, 13μ, etc.
○ More impacted by various absorption lines, etc.
○ RTM more needed and appropriate