Effects of Cloud/rain on Ka band altimeter

1
Effects of Cloud/rain on Ka band altimeter

• J. Tournadre et al
• Ifremer Brest

2
OSTS Hobart, March 2007

• Altika altimeter in Ka band (see Noubel and
  Verron presentations)
• Better performances but one major problem at Ka
  band (35.75 GHz) rain and cloud can strongly
  attenuate the signal and distort the echo
  waveform.
• Important to quantify
• Impact on echo waveforms and geophysical
  parameters estimate (use of MLE4 algo)
• Percentage of data possibly lost due to rain
  and/or clouds

3

• Previous studies (Tournadre 1999). In presence of
  rain larger than 1-2 mm/hr the distortion of the
  waveform will inhibit the geophysical parameter
  retrieval.
• At Ka band the attenuation by cloud droplet is
  about 1.1 dB/km par g/m3 (10 time larger than Ku
  band), not negligible. Cloud at Ka rain at Ku .
• Cloud more frequent than rain. Necessity to
  analyze in detail the effect of cloud on the
  signal
• Waveform modeling in presence of cloud
• Estimation of attenuation, off-nadir angle,
  leading edge slope at 1Hz and 20 Hz as a function
  of cloud parameters (IWC, height, diameter,..)
• Estimation of the impact of cloud on the
  geophysical parameters (ssh,s0, SWH) retrieval
  waveform modeling and MLE4 (Stenou et al).
  Determination of data availibility
• Rain/cloud flag definition based on the signal
  analysis (Altika is single-frequency)

4
Waveform modeling

• Analytical model of waveform based on Brown model
  an attenuation term
• A attenuation by cloud or rain
• k attenuation coefficient in dB/km
• 
  rain
• 
  cloud
• R rain rate, mv liquid water content (g/m3)

5

• For clouds
• Attenuation depends on Integratedd cloud liquid
  water (ICLW)
• Simpler than the rain case

6
Cloud liquid water
Typical values for the properties of clouds. The
values are merely modal-means. The range of
observed values is quite large. The radius of
cloud droplets is r (microns), the effective
optical radius optically is r', N is the number
of droplets per cubic centimetre, L is the liquid
water content of the cloud (g/m3). For all
clouds, the level of observation is just below
the freezing level, except for fog and cirrus.
(Hess et al 1998)
7
Cloud water droplet and ice cristalsIce has a
much lower attenuation coefficient and can be
neglected in the computation.
8
Cloud at a given time!!
9
Cloud liquid water content data

• Several satellite sensors gives estimates of the
  integrated cloud liquid water content such as
  SSM/I, AIRS or MODIS on Aqua/Terra.
• However to have a good modeling we need high
  resolution data.

10
Waveform distortion by rain and cloud

• Ka band strong attenuation by rain and cloud
• But more important
• strong distortions of the waveform shape
  modification of leading edge and plateau slope

0.1
1
10
Return power
Attenuation
Altika WF over a 10 km 2.5 mm/hr rain cell
11
Altika unavailability by rain Use of JASON
ENVISAT,TOPEX rain climatology same sampling as
Altika Probability of rain greater than 1-1.5
mm/hr
12
(No Transcript)
13
Waveform modeling cloud with gaussian
distribution of ICLW

• ICLW.
• Cloud radius
• Distance nadir/cloud
• Attenuation
• Off-nadir angle

Stronger distortion for small high ICLW cloud
14
Comparison 1Hz 20Hz
Small difference on attenuation for cloud radius
larger than the footprint (8km) For ICLW
lt0.5kg.m² weak impact of cloud (attlt1 dB zlt0.01²)
15
Quantification of the impact of cloud on
geophysical parameter see Desjonquéres
Presentation

• Other approach definition of a rain and cloud
  flag
• Altika single frequency altimeter implies a
  definition of a flag based on the analysis of the
  signal itself (sigma0 and or waveform).
• Critical for light to very light rain and cloud
  with high water content.
• Radiometer liquid water estimate are not
  sufficient for cloud/rain flagging (see past
  experience with Topex and Jason)
• Flag based on sigma0 and off-nadir angle estimate
  variations.
• Test on modeled waveform using satellite cloud
  liquid water measurements.

16
Analysis of s0 and z² variations within a cloud.
Nuage à IWC constant
Nuage IWC exponentiel
High small scale highly correlated variations of
s0 et z
17
Modeled waveforms using (10 km resolution 10 km)
AIRS liquid water data from AIRS on AQUA
18

• Detection of s0 et z² variations
• Method Wavelet decomposition (here symmlet 8)
  of s0 et z
• Use  matching pursuit  algorithm (Mallat et al,
  1997)
• Determination of the most significant wavelet
  from an energy point of view (named atoms)
• Each event is characterized by its amplitude and
  scale (possibility to discriminate by scales)

19

• Exemple AIRS modeled waveforms
• White noise added to attenuation and z

Reconstructedd signal from atoms
20
Synthesis

• IWC
• Attenation
• Off_nadir
• attenuation atoms
• O z atoms

21
MODIS05 (1km)
22

• IWC
• Attenation
• Off_nadir
• attenuation atoms
• O z atoms

23

• Deformation can be important even for low IWC
  (0.2kg/m²)
• For IWC gt0.6 kg/m2 rain most probable
• Below 0.2 kg/m² few data will be contaminated
• Distortion depends more strongly n the
  variability of IWC within the footprint than mean
  value

24
To be done Availability maps Test of the
feasibility of an operational rain/cloud flag
based on s0 and z variations

• Modis06 level3 daily products clouds.
• Cloud fraction, cloud liquid water histograms

25
Conclusion

• Model shows that the mean impact of cloud is weak
  for CLWC lower than 0.2 kg/m²
• For higher CLWC it will be necessary to flag the
  data
• Wavelet analysis of the sigma0 and off-nadir
  angle variations can be used to flag the data
  affected by rain and clouds..

26
Problem Definition of a rain flag

• As Altika certainly mono-frequency the rain flag
  can not be similar to the JASON or ENVISAT one
  (dual frequency).
• Critical for light to very light rain (lt1 mm/hr)
  and some non-raining clouds where microwave
  radiometer data can not be used for rain
  detection
• New rain flag to be defined modelisation show
  that it can be based on variation of sigma0 and
  off-nadir (plateau) angle

Example variation of off-nadir angle induced by
rain
Off_nadir as function of distance from rain cell
and rain cell diameter for 0.5 mm/hr
Off-nadir as a function of rain celle diameter