WP 3000 Macroscopic Parameter Retrieval

1
WP 3000 Macroscopic Parameter Retrieval 

• Jane Hurley, Anu Dudhia, Don Grainger
• University of Oxford

2

• Aim to retrieve most obvious macrophysical cloud
  properties
• Cloud Top Height CTH (relative to instrument
  pointing)
• Cloud Top Temperature CTT
• Cloud Extinction Coefficient kext

3
Cloud Forward Model (CFM) Radiance in MIPAS FOV

• Assume that
• a cloud in the MIPAS FOV is horizontally
  homogeneous that is, has a constant cloud top
  height across the FOV and can be characterized by
  a single extinction coefficient.
• the temperature structure within the cloud can be
  determined by the wet adiabatic lapse rate
  estimated downwards from the cloud top
  temperature.

4
The radiance is considered in the clearest
microwindow of the MIPAS A band 960 cm-1 961
cm-1
O3 CO2
O3 CO2
O3
O3
O3
O3 NH3
O3
5
Total radiance measured within two MIPAS FOVs
(the FOV containing the cloud top and the FOV
immediately below) calculated by the CFM for
varying cloud top heights and extinction
coefficients.
6
Gas Correction and Validation with Simulations
Real MIPAS measurements Rm will include
significant gaseous radiation contributions Rg,
while the CFM calculates only the radiation
contribution by the cloud itself Rc. It is thus
necessary to deduce what portion of the measured
signal is due to the cloud. Assume that the
cloud has a continuum signal and that the gaseous
contribution has emission/absorption lines.
7
Want blue to overlay red
Comparing the CFM output with RFM simulations of
cloud corrected for gaseous emission in the
manner described above, it is clear that the CFM
does a good job at estimating the radiance
measured by the MIPAS FOV.
8
Retrieval and A Priori Dependence

• Cloud modelling is a highly non-linear process,
  even when considered on the vastly simplified
  scale.
• Given a pair of radiances from two adjacent
  sweeps in a scan pattern, there can be two
  possible clouds present a high thin cloud, or a
  low thick cloud.
• Depending upon which a priori is supplied to an
  OER, equally valid different solutions will
  result.
• NEED TO ADD MORE INFORMATION, to better
  characterize the a priori extinction.
• Easiest way to get information is to use quantity
  we already have
• THE COLOUR INDEX CI
• which should already be highly correlated with
  Cloud Effective Fraction EF, which is the
  effective blocking power of the cloud in the
  FOV.

9
CI and EF are clearly correlated, but how?? Needs
optimization down the road
10
Using three different extinction coefficients
(0.001, 0.01,and 0.1 km-1), four standard RFM
atmospheres (day, polar summer, polar winter and
tropical) and nine different cloud top heights at
a tangent height of 9 km, for a total of 108
different cloudy atmospheric conditions
11
Optimal Estimations retrieval of form
with state
vector and using
Real Measurements 2 radiance measurements from
MIPAS spectrum the first sweep flagged as
cloudy and the one immediately below DIRECT
Pseudo-Measurements Tret temperature
corresponding to first flagged cloudy sweep EF
Cloud effective fraction, as estimated from
CI RELATE
12
The retrieval has been applied to and validated
with RFM simulations for cloudy FOVs with known
CTH, CTT and kext.
13
Application to MIPAS Spectra

• Hot spot of high cloud over Indonesian toga
  core, mountainous regions such as the Southern
  Andes and Rockies, Amazon Basin and the Congo
• Increasing cloud top height towards the tropics
• Retrieved CTT is nearly fully correlated with
  CTH
• Retrieved log(kext) is more or less constant over
  the globe.

14
cirrus?
thick
Double peaked distribution in kext thick cloud
and cirrus??
15
If there IS a cloud in the FOV at a certain
latitude, this shows the probability that it will
occur at a given altitude
16

• If there IS a cloud in the FOV at a certain
  latitude, this shows the probability that it will
  occur at a given temperature
• Basically anti-correlated with cloud top
  height/altitude

17
If there IS a cloud in the FOV at a certain
latitude, this shows the probability that it will
have a given extinction
18
Rapid rise in CTH in S.Pole in August/September
PSCs? Higher (lower) clouds in summer (winter)
hemisphere
19
Future Work

• Check retrievals sensitivity to reference
  temperature.
• Check retrieval against other retrievals of
  macroscopic properties McClouds etc
• Run over larger MIPAS dataset to get a high
  cloud climatology
• Compare high cloud climatology with others
  ISCCP etc

20
(No Transcript)
21
The forward model of the radiance encountered in
the MIPAS FOV can be expressed as
, where

is the pencil beam radiance for temperature



, pathlength


, whereby Re Earths
radius, thc tangent height where cloud is
first detected, thFOV tangent height of FOV
modelled, T temperature, z relative position
within the FOV, F FOV convolution and Gwet
wet adiabatic lapse rate.
22
Basically, the sequential retrieval consists of
two separate retrievals (preliminary and
main) which independently use difference pieces
of information in an attempt to retrieve the
state vector using
the one-step formulation and then feed their
results back and forth until a converged set of
values is reached.