Partner und Internationale Organisationen
(Englisch)
|
Dept. of Geomatic Engineering - University College London (UK), Institut für Weltraumwissenschaften - Freie Universität Berlin (D), Koninklijik Nederlands Meteorologisch Institut (NL), Deutsches Zentrum für Luft und Raumfahrt (D): Institut für Physik der Atmosphäre, Institut für Planetenerkundung, Institut für Optoelektronik, Institut für Weltraumsensorik
|
Abstract
(Englisch)
|
The main scientific objective of the project was the development of new cloud-top products (heights, type, optical thickness, effective droplet size), especially for cirrus clouds, from existing and new sensors for weather forecasting and climate change analysis. There are three different techniques which were studied and compared by the participating institutes (UCL, KNMI, DLR, FUB, ETH): brightness temperature (CO2 slicing method), stereoscopy and Oxygen A-band. The new cloud products were additionally validated with contemporaneous observations from ground-based active and passive remote-sensing instruments. The role of ETH Zuerich, Institute of Geodesy and Photogrammetry, was to provide algorithms for stereo photogrammetric data analysis. It included the modelling of linear CCD array sensors and the matching of satellite- and ground-based cloud images for the purpose of 3D cloud modeling.
Existing sensor models for CCD linear array sensors have been investigated and used for cloud-top height (CTH) estimation. A general and flexible sensor model for a wide class of CCD linear cameras was studied and partially implemented. The model had to take into account not only the sensor geometry (number of lenses and CCD lines), but also any available data about the sensor internal orientation, provided by pre- or in-flight calibration, and about the sensor's external orientation, provided by orientation instruments carried on board. If the sensor position and attitude is known, a forward intersection provides the object coordinates of two or more corresponding points measured in the images. Otherwise, the sensor's external orientation is modelled with second order polynomial functions depending on time, which parameters are estimated from ground control and tie points.
Stereo pairs from different satellite sensors (ATSR-2 on ERS-2, MISR on EOS-Terra and MOMS from Priroda mission) and at various resolutions (18m - 1km) have been processed into CTH. All images were enhanced and radiometrically equalized before the matching to improve texture, especially within clouds, and to correct the illumination differences of the two looking angles. The matching was done with the Multi-Photo Geometrically Constrained Matching Software developed at our institute, which is based on Least-Squares Matching. For ATSR2, a wind correction algorithm with Meteosat-6 Rapid Scans was implemented, to correct the ATSR2 CTH for the height error introduced by the cloud motion within the time delay of around 120 seconds between the forward and nadir view. Such a wind correction is necessary for any stereo CTH retrievals from a single satellite, unless the height and motion are estimated together which is possible with more than two views (e.g. MISR with 9 views).
A ground-based imager system has been developed to automatically derive cloud-base heights (CBH). It consists of two Kodak DCS460 colour digital CCD cameras with wide-angle lenses, positioned with a horizontal distance of about 1 km, both connected to a laptop for camera control and image storage and to a radio clock for time synchronization. A fieldwork campaign was carried out within the Mesoscale Alpine Programme (MAP) from October, 7 - October, 22, 1999. An important step for reliable CBH results is an accurate determination of both the interior and exterior orientation of the cameras, which had been carried out by a testfield calibration at our Institute, an airplane flight equipped with differential GPS and star path images during clear nights. The CBH results of selected cases were compared with radiosondes, lidar and visual observations from the MAP Composite Observing Network and validated with manually measured cloud points at a digital photogrammetric station.
|
