Validating the AVHRR Cloud Top Temperature and Height product using weather radar data

1
Validating the AVHRR Cloud Top Temperature and
Height product using weather radar data
COST 722 Expert Meeting 9.-11. 6. 2004 Sauli Joro
2
Outline

• Introduction
• Weather radar TOPS product
• Cloud Top Temperature and Height product
• Data
• Product comparison
• Results
• Summary

3
Introduction

• Visiting Scientist Activity within EUMETSAT
  SAFNWC framework
• EUMETSAT Satellite Application Facilities (SAFs)
  are programs specialized in developing and
  processing satellite data
• SAFNWC is consentrated on nowcasting and very
  short range forecasting

4

• The Cloud Top Temperature and Height (CTTH)
  product is developed within the SAFNWC framework
• Objective is to validate CTTH product with
  weather radar data known to be very ambitious
• Study mainly concentrated on opaque and
  semi-transparent high clouds (clouds containing
  ice)

5
Weather radar TOPS product

• Direct cloud measurement with good temporal and
  spatial resolution

6
(No Transcript)
7

• What is the threshold value for an echo in dBZ
  to be considered as the cloud top?
• FMI Finnish Air Force in case of raining
  cloud treshold value of 10 dBZ should give
  reliable cloud top heights
• Poutiainen (1999) Various cirrus types can
  create Z values between 15 - 5 dBZ
• -5 dBZ and -10 dBZ selected to represent the true
  tops of ice clouds

8
Thick cirrus on RHI screen taken from Radar
Vantaa on 25 June 1998 at 1059 UTC (Poutiainen,
1999)
9
Thin cirrus on RHI screen taken from Radar Vantaa
on 22 July 1998 at 0729 UTC (Poutiainen, 1999)
10
TOPS product

• Uses 3-D weather radar data
• Vertical resolution 100 m

• For each pixel downward search in cylindrical
  coordinates at constant range for the chosen dBZ
  treshold value and determines if it is crossed -gt
  interpolation

11
TOPS underestimation
12
CTTH product

• Aims providing reliable estimates of the cloud
  top temperature and height for all the cloudy
  pixels within an AVHRR scene
• Vertical resolution 200 m
• Consists of two algorithms one for opaque and
  one for semi-transparent cloudiness
• Results of Cloud Mask and Cloud Type products
  used as input data

13
CTTH opaque retrieval

• Applied to all pixels classified as opaque by
  the Cloud Type product
• Based on Radiative Transfer Model simulations and
  results of NWP model
• For each pressure level the RTM simulates the
  AVHRR channel 4 cloudy brightness temperature -gt
  the best fit to the measurement is selected -gt
  CTH from NWP

14
CTTH semi-transparent retrieval

• Measurement itself is contaminated as the
  radiation from the surface is partly passed
  through the cloud -gt brightness temperatures too
  warm
• Method uses AVHRR channels 4 and 5 brightness
  temperatures
• Result is applied to all pixels within an image
  segment classified as semi-transparent

15

• Single cloud layer
• Constant absorption coefficient throughout the
  cloud layer
• Brightness temperature depends linearly on
  radiance
• No atmospheric absorption
• Local thermodynamic equilibrium

thermodynamic cloud top temperature
16
Data

• Selected time period April and May 2003
• 301 satellite overpasses
• FMI radar data, 15min temporal resolution
• NWP used HIRLAM
• RTM used RTTOV

17
Radar data features
18
Product comparison

• Cases selected subjectively from AVHRR imagery
  together with Cloud Type and CTTH product outputs

19

• Simple comparison method used data may not be
  normally distributed
• Pixel-by-pixel comparison not reasonable -gt data
  needs to be averaged
• Data is classified to histogram with 200 m class
  intervals -gt resolution difference vanishes
• Modes are taken as the results of the products
• Mode difference TOPS CTTH
• Comparison is considered to be successful if the
  absolute value of mode difference is less than
  500 m

20
Results
both -5 dBZ and -10 dBZ cases 41 success
TOPS-CTTH
21
radar echoes below 2000m discarded -10 dBZ
applied to all cases 57 success
TOPS-CTTH
22
no discarded echoes cases within r-10 coverage
area 75 success
TOPS-CTTH
23
timing -2 minutes
Case A Opaque high cloud from Radar Ikaalinen -5
dBZ coverage. Satellite over pass is received on
1 May 2003 at 1043 UTC.
24
Case A frequency distributions and HPs.
25
timing -3 minutes
Case B Opaque high cloud from Radar Anjalankoski
-5 dBZ coverage. Satellite over pass is received
on 29 April 2003 at 1915 UTC.
26
Case B frequency distributions and HPs.
27
TOPS-CTTH
28
timing -6 minutes
Case C Semi-transparent high cloud from Radar
Utajärvi -10 dBZ coverage. Satellite over pass is
received on 13 May 2003 at 1047 UTC.
29
Case C frequency distributions and HPs.
30
Summary

• A new approach for the CTTH product validation
  presented
• The approach presented here uses 3D radar data
• Result interpretation very challenging as both
  methods are based on numerous assumptions lots
  of different error sources

31

• Results on the opaque high cloud categoty
  promising more than 55 of the comparisons
  successful
• Still lots of unsuccessful comparisons with no
  explicit explanation, also Cloud Type
  misclassifications occurred
• -10 dBZ proved to be a good first guess for the
  top heights of thick ice clouds

32

• Results on semi-transparent high cloud category
  disappointing with only three successfull
  comparisons
• -10 dBZ threshold turned out to be too high for
  the semi-transparent high clouds and more
  sensitive values should be applied in the future

33
Conclusion
A weather radar is not the best possible tool for
the CTH validation, however, it can offer
valuable information about the cloud tops in
various situations being at its best when the
cloud consists lots of ice particles. The lack of
45-degree elevation angles limited proper -10 dBZ
cases to Radar Utajärvi
34
Report available at http//www.smhi.se/saf