CloudSat Snowfall Retrievals: Potential, Status, and Challenges

1
CloudSat Snowfall Retrievals Potential, Status,
and Challenges

• Tristan S. LEcuyer, Richard T. Austin, and
  Min-Jeong Kim
• Colorado State University

2
?
3
ERA40 SnowfallDistribution (DJF 98-00)
4
Role of CloudSat
Accurate accounting of the distribution of
snowfall in high latitude land regions is
critical for water resource management
applications. CloudSat may provide

• Improved snowfall discrimination for refining
  statistics on the global frequency of occurrence
  of snow events
• Zonal distributions of the vertical profiles of
  falling snow over both land and oceanic surfaces
  on monthly timescales for (eg. for the evaluation
  of water cycle in NWP)
• Potential for limited (narrow swath/snapshots)
  assimilation experiments using the vertical
  structure of falling snow

Drawbacks include

• Sampling
• Complexity of the relationship between
  reflectivity and snowfall rate/IWC

5
The CloudSat Mission
Objective To provide, from space, the first
global survey of cloud profiles and cloud
physical properties, with seasonal and
geographical variations needed to evaluate the
way clouds are parameterized in global models,
thereby contributing to weather predictions,
climate and the cloud-climate feedback problem.
The Cloud Profiling Radar

• Nadir pointing, 94 GHz radar
• 3.3?s pulse ? 500m vertical res.
• 1.4x2.5 km horizontal res.
• Sensitivity -28 dBZ
• Dynamic Range 80 dB
• Antenna Diameter 1.85 m
• Mass 250 kg
• Power 322 W

6
The A-Train Constellation
Formation Flying

• Orbit is sun-synchronous at an altitude of 705 km
  and inclination of 98.2º
• Allows nearly simultaneous views of the Earth
  from distinct instrument payloads
• Of particular relevance to CloudSat snowfall
  retrievals is the 89 GHz channel of the AMSR-E
  instrument aboard Aqua

7
CloudSat Data Products
8
Probabilistic Retrieval Philosophy

• Many of CloudSats algorithms are framed around
  Bayesian statistics in a form that is an
  extension of Rodgers (1976, 1990, 2000) optimal
  estimation formulation of the inverse problem.
• The technique offers the following benefits
• Allows formal inclusion of multiple forms of
  information including a priori knowledge
• Requires explicit specification of input
  uncertainties
• Provides quantitative measures of uncertainty in
  retrieved values including relative contributions
  of all forms of assumed knowledge and measurement
  error
• Readily expanded to include additional
  measurements and/or constraints
• CloudSats standard liquid and ice cloud
  algorithms as well as its experimental
  precipitation algorithms are all formulated using
  variations of this basic framework.
• Output consists of a multi-dimensional PDF in
  retrieval parameter space that describes the
  probability that the scene exhibits any
  particular set of microphysical properties given
  the measurements and our best physical model of
  the relationship between them

9
Snowfall Algorithm

• Measurements
• Vertical profile of radar reflectivity observed
  by the CPR
• Surface return for estimating PIA (not yet
  implemented)
• Retrieved parameters
• Vertical profiles of exponent parameter ? and
  column-mean number concentration N0 for an
  exponential distribution of snow crystals
• Profile of snowfall rate
• Properties
• Optimal-estimation (probabilistic) framework
• Spherical ice particles currently assumed but can
  easily be modified to model more complicated
  crystal shapes
• Strengths
• Retrieval framework provides a suite of Q.C. and
  error diagnostics
• CPR offers higher spatial resolution and improved
  sensitivity to many other satellite-based
  precipitation sensors
• Weaknesses
• Single-frequency requires simplifying assumptions
  concerning PSD and snow crystal shape
• CPR is nadir-pointing providing only a 2D slice
  of the real world

10
Example from Wakasa Bay
11
Error Analysis (Challenges)

• Measurement errors, forward model errors, and
  those associated with the mathematics of
  inversion are much smaller than those due to
  variability in the influence parameters
• To demonstrate this, the impacts of varying
  crystal shape and PSD parameterization on Z-S
  relationships are investigated
• Sekhon and Srivastava (1970) PSD is adopted as a
  baseline
• Scattering properties for non-spherical
  snowflakes generated by Min-Jeong Kim using
  Discrete Dipole Approximation (DDA)

12
Impacts of Crystal Shape

• DDA calculations averaged over the PSD
  parameterization of Sekhon and Srivastava (1970)
  suggest that snow crystal habit impacts on
  simulated reflectivities range from 2 dBZ in
  light snow to as much as 7 dBZ in heavy snowfall

13
Impacts of PSD

• Uncertainties in the shape of the PSD can
  introduce 3-6 dBZ errors in simulated
  reflectivities
• Errors are largest in light snowfall where ZD6
• Note changing the function relating N0 and ? to
  S indirectly tests different fall velocity
  relations

Sensitivity to ?
? 0 ? 1 ? 2
Reflectivity (dBZe)
Snowfall Rate (mm h-1)
14
Implications for Retrieval

• Given a perfect forward model, 1 dB measurement
  errors lead to errors in retrieved snowfall rate
  of less than 10

• Algorithm assumptions concerning PSD and snow
  crystal shape, however, spread the range of
  allowable solutions in the absence of additional
  constraint

15
PMW Constraint

• Passive microwave brightness temperatures
  simulated from DDA-derived scattering properties
  also demonstrate sensitivity to snow crystal
  shape and PSD
• Since TB decreases with increasing scattering
  (opposite to Z), the combination of active and
  passive microwave measurements may provide a
  means for discriminating certain sets of
  PSD/shape assumptions

16
Combined Retrieval

• With a 140 GHz brightness temperature accurate to
  5 K as a constraint, the range of PSD and
  crystal shape assumptions can, in principle, be
  narrowed improving quantitative snowfall
  retrievals.

17
Wakasa Bay Revisited
18
Validation Plans

• Validation involves a partnership of activities
  with other national agencies as well as
  international partners.
• The validation funded by the CloudSat project is
  limited to a few activities and this is highly
  leveraged through other support mentioned above.
  There is also a significant amount of validation
  activity being conducted through support from
  other programs.

19
Planned Activities

• Field Programs
• Japan
• Scheduled cruise of Mirai
• Scheduled SPIDER flights
• Europe
• African Monsoon Multidisciplinary Analysis (AMMA)
  (France/Germany)
• Cirrus and Anvils European Satellite and
  Airborne Radiation measurements project (CAESAR)
  (UK)
• University of Bremen low cloud optical property
  experiment (Germany)
• United States
• TWP-ICE (ARM Darwin site) also support from
  CloudSat/CALIPSO and UK
• Two sets of UWyo King Air flights in winter-time
  clouds and marine stratocumulus (supported by
  EPSCor and DoD)
• Canada
• extensive Canadian high-latitude winter cloud
  precipitation CloudSat validation experiment, Dec
  1-April 30 2005/6 and 2007, funded by CSA
  specifically for CloudSat (Canadian
  CloudSat/CALIPSO Validation Project, C3VP)
• Systematic Measurements
• ARM sites
• Cabauw (Netherlands)
• Chilbolton (UK)
• Mirai (Japan - floating platform)
• Lindenberg Observatory (Germany)

20
The Canadian CloudSat/ CALIPSO Validation Project
(C3VP)

• David Hudak, Howard Barker, Walter Strapp, and
  Kevin Strawbridge
• Réjean Michaud

21
C3VP Scientific Goals

• Funded by CSA, this project will undertake a
  thorough and careful evaluation of the CloudSat
  products
• Their applicability to the Canadian climate will
  be emphasized.
• The program will focus on stratiform cold-season
  cloud systems. These are frequently mixed phase
  in nature and occur throughout Canada much of the
  year. The widespread and slowly changing nature
  of these systems are particularly well suited to
  validation studies.

22
C3VP Instrumentation

• Ceilometer
• Visibility meter
• Meteorological measurements
• Radiation flux measurements
• 2 PMS FSSP probes (2-47 µm)
• PMS 2D2-C probe (25-800 µm)
• PMS 2DC-grey probe (15-960 µm)
• PMS 2D-P probe (200-6400 µm)
• SPEC 2D-S (10-1280 µm)
• PMS PCASP aerosol spectrum probe
• Nevzorov extinction probe

• Ground-based
• Alaska W-band polarimetric radar
• X-band radar
• Wind profiler
• Precipitation Occurrence Sensor System (POSS)
• Hot plates
• ALIAS lidar
• Microwave radiometer
• McGill University video disdrometer
• Penn. State Spectrometer
• Aircraft (NRC Convair-580)
• AERIAL cloud lidar
• PMS LWC probe
• Nevzorov LWC/TWC hot-wire
• Counterflow Virtual Impactor for TWC/IWC
• CSU Ice nucleus counter
• Small ice detector
• W/X-band radar

Possible Additions
23
Summary

• CloudSat observations are well-suited to
  assessing the frequency of occurrence of snowfall
  events and categorizing them into coarse
  intensity bins.
• Quantifying snowfall intensity represents a more
  challenging matter, since forward model
  calculations are very sensitive to the choice of
  PSD and crystal shape.
• Multi-sensor approaches that merge high frequency
  PMW and radar observations offer the potential to
  reduce these uncertainties by constraining range
  of possible solutions. The combination of
  CloudSat and AMSR-E represents a new paradigm for
  satellite-based snowfall measurement.
• Anticipate early snowfall measurements from
  CloudSat by years end!

24
Backup Slides
25
Aircraft Observations I
Observations from the Wyoming cloud radar (WCR)
in snowing clouds during the WYICE97 field
experiment at Laramie, WY in March 1997
In many cases reflectivities in snowing scenes
will fall below the MDS of TRMM PR and proposed
DPR.
26
Aircraft Observations II
Airborne Cloud Radar (ACR) and co-located
Millimeterwave Imaging Radiometer (MIR)
observations of snowfall during the AMSR-E
validation experiment at Wakasa Bay, Japan
Jan-Feb 2003
27
Surface Observations
Missed snowfall occurrence and amounts at a
reflectivity threshold of 18 dBZ (0.4 mm/h
snowfall water equivalent).
Courtesy of Paul Joe, MSC
28
ECMWF Re-analyses

• A significant fraction of the annual
  precipitation poleward of 42.5o falls in the form
  of snow.
• In ERA40, snowfall accounts for as much as 65
  of the winter-time precipitation in the Northern
  Hemisphere poleward of 42.5o

29
SH Winter
30
CloudSats Validation Plan

• Four-fold validation plan
• Verify radar performance
• Absolute calibration (2 dB required)
• End-of-lifetime sensitivity (-26 dBZ or better
  required)
• Stability
• Once per month CPR will measure oceanic surface
  return at 10º in cross-track direction
• Validate pointing knowledge error of the radar
  footprint (1 km or 0.583º required)
• On-orbit comparison of actual and predicted
  coastal crossings
• Verify radar sensitivity in cloud domain (eg.
  determine the minimum optical depth that can be
  discerned)
• Statistical evaluation of geometric profile
  product using volume matched surface and aircraft
  measurements
• Quantify both random and systematic errors in the
  standard data products
• Correlative analysis direct comparison of
  retrieved products to ground-based observations
  of the similar quantities
• Physical validation evaluation of algorithm
  assumptions and their variability

31
CAESAR (Phase 1 14/11/05-21/12/05) Cirrus and
Anvils European Satellite and Airborne Radiation
measurements project

• To improve the representation of cirrus clouds in
  Numerical Weather Prediction and Climate Studies
• To characterise the microphysical and
  macrophysical properties of mid-latitude cirrus
  clouds over different seasons
• To perform closure studies relating the cloud
  microphysics to remote sensing data across the
  visible, thermal infrared, far infrared and
  microwave regions of the electromagnetic spectrum
• To validate remotely sensed products from the
  A-Train, AATSR, Seviri and Chilbolton Radar

PIs Dr Anthony J. Baran (Met Office), Dr Clare
Lee (Met Office), and Dr Caroline Poulsen
(Rutherford Appleton Laboratory)
http//metresearch.net/CAESAR