NEW CLOUD-TOP HEIGHT PRODUCTS

Stereo Imaging Principles:

• Not dependent on any assumptions of atmospheric physics, purely geometric
• Sensitive to accuracy of knowledge of sensor position, attitude and knowledge of along track drift component and the advection of clouds by the local wind-field
• Need accurate automated digital photogrammetry to deter- mine correspondences between pixels (disparity values)

Automated Digital Photogrammetry:

• Operational potential sensitive to stereo matching accuracy and computational performance
• M2 algorithm developed at UCL for ESA-ERS ATSR2 and NASA- EOS MISR will be rigorously tested A sample of Cloud Top Heights retrieved using ATSR2 images.
• New algorithm to be developed for MOMS at ETH to be compared against M2 performance
• Use stereo visualisation to assess cloud-top height completeness, contrail drift reliability and accuracy within a Digital Photogrammetric Workstation (DPW)

Oxygen A-Band Cloud-top Pressure/Oxygen A-band Principles:

• Depends on the absorption of solar radiation by a well-mixed gas (O2) which is related to the air mass in the line of sight and hence to the cloud-top pressure
• Artificial neural networks, trained with radiative transfer simulations, are used to determine the relationship between input parameters (radiances, surface albedo, observation geometry) and cloud-top pressure
• Cloud optical thickness, cloud geometrical thickness and cloud droplet radius can also be retrieved with an algorithm to be developed for this purpose

Fuzzy Integral Detection Of Cirrus And Contrails

• Cirrus clouds, particularly thin cirrus, are difficult to detect at either visible/NIR or thermal IR wavelengths because their boundaries are diffuse and their grey-level values are almost indistinguisable from their background, especially when overlaid over lower cloud layers
• Contrails are also difficult to detect, particularly at visible/NIR wavelengths as they are frequently broken and may be very difficult to distinguish from natural cirrus clouds if they are more than a few hours old
• Through hierarchical application of fuzzy integrals, cirrus clouds can be distinguished from their background using properties such as grey-level differences at different inter -point distances and small grey-level differences across a boundary
• Contrails will be detected using similar fuzzy integrals but in addition employing the fact that they are likely to be quasi-linear in nature

  ---

  Go to the next topic - Validation Methods
  Back to the Cloudmap Home Page
  Back to the GE Home Page

  ---

  Please note that any views expressed here do not necessarily reflect UCL College policy. No claim is made, explicit or implied, about the suitability of any data presented here for any purpose. All rights are reserved by individual members of the Dept., the Dept. itself, or UCL as appropriate, unless otherwise indicated.

  Any links to pages provided by individuals in the department are done so on the understanding that they agree to abide by the College and draft UCL Guidelines.
  www-admin@ge.ucl.ac.uk

  Last modified: 13th May 1998