Please use this identifier to cite or link to this item:
|Title:||Remote sensing-based mapping and modelling of salt marsh habitats based on optical, lidar and sar data|
|Authors:||van Beijma, Sybrand Jucke|
|Presented at:||University of Leicester|
|Abstract:||There is much interest in the ability of Remote Sensing (RS) technologies for mapping natural environments. Meanwhile, coastal zones need monitoring in order to find a balance between human use and sustainable functioning of coastal zone ecosystems. This research explores methods for characterising coastal salt marsh zone habitats using multi-source RS data, focussing on under-exploited Synthetic Aperture Radar (SAR) remote sensing data, thereby providing additional information in support of the mapping of natural habitats in coastal zones. This research examined the use of quad-polarimetric airborne S-band and X-band SAR data, in conjunction with optical and LiDAR RS data variables, for assessment of environmental parameters, mapping and modelling of salt marsh habitats in a research area set in the Llanrhidian salt marshes in Wales. In the first analysis it was researched how SAR descriptors (backscatter intensity and polarimetric decomposition variables) were affected by salt marsh environmental and botanical factors. It was found that SAR backscatter from the most seaward pioneer zone of the salt marsh was most affected by soil moisture variations. Differences in botanical structure caused variations in SAR backscatter mechanisms active in different habitats. In the second analysis habitat mapping was carried out with optical, LiDAR and SAR variables, with the supervised classifiers Support Vector Machine (SVM) and Random Forest (RF). With these classifiers accurate salt marsh habitat maps were produced, the most accurate classification achieved was 78.20% with RF based on all available RS variables. The last research experiment involved multivariate regression analysis of correlations between RS variables and biophysical parameters vegetation cover, height and volume and showed that multivariate SVM regression was the most accurate technique for all three biophysical parameters. This research indicated that SAR is complementary to optical and LiDAR data for ecological mapping and therefore recommended to be included in similar ecological studies.|
|Rights:||Copyright © the author. All rights reserved.|
|Appears in Collections:||Theses, Dept. of Geography|
Items in LRA are protected by copyright, with all rights reserved, unless otherwise indicated.