Fisher, DN; (2014) Application of Stereo-Photogrammetric Methods to the Advanced Along Track Scanning Radiometer for the Atmospheric Sciences. Doctoral thesis , UCL (University College London).

Preview

PDF Fisher_Daniel_thesis_excl_third_party.pdf Available under License : See the attached licence file. Download (8MB)

Abstract

This thesis studies photogrammetric techniques applied to the ATSR instruments for the extraction of atmospheric parameters with the objective of generating new scientific datasets. The atmospheric parameters under observation are cloud top height, smoke plume injection height, and tropospheric wind components. All have important applications in various tasks, including the initialisation and validation of climate models. To generate accurate stereo measurements from the ATSR imagery the forward and nadir views need to be accurately co-registered. Currently this is not the case, with differences of up to 2 pixels in both axes recorded. In this thesis an automated image tie-pointing and image warping algorithm that improves ATSR co-registration to ≤1 pixel is presented. This thesis also identifies the census stereo matching algorithm for application to the ATSR instruments. When compared against a collocated DEM, census outperforms the previous stereo matching algorithm applied to the ATSR instrument, known as M4, significantly: RMSE ~700m vs. ~1200m; bias ~60m vs ~600m; R2 ~0.9 vs ~0.7. Furthermore, this thesis reviews the M6 algorithm developed for application within the ESA ALANIS Smoke Plume project. Using census a climatological cloud fraction by altitude dataset over Greenland is generated and demonstrated to agree well with current observational datasets from MISR, MODIS and AATSR. The 11μm channel stereo output provides insights into high cloud characteristics over Greenland and appears to be, in comparison with CALIOP, practically unbiased. The ALANIS Smoke plume project is introduced and the inter-comparison of the M6 algorithm against MISR and CALIOP is presented. M6 demonstrates some ability for determining smoke plumes injection heights above 1km in elevation. However, the smoke plume masking approach currently employed is demonstrated to be lacking in quality. Finally, this thesis presents the determination of cloud tracked tropospheric winds from the ATSR2-AATSR tandem operation using the Farneback optical flow algorithm. This algorithm offers accuracy on the order of 0.5 ms-1 at full image resolution, which is unprecedented in comparison to similarly derived datasets.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item