Back to results list
Please use this identifier to cite or link to this item:
|Title:||Development of geographic image cognition approach for land degradation assessment with hyperion images||Authors:||Wang, Jing||Keywords:||Hong Kong Polytechnic University -- Dissertations
|Issue Date:||2010||Publisher:||The Hong Kong Polytechnic University||Abstract:||Land degradation is a major problem world-wide. The degradation process is much related to the soil characteristics, topography, vegetation, land types, land use, climate, and human activities. Quantification of land degradation is difficult mainly due to the ambiguously expressed knowledge and the lack of appropriate information. However, there is a pressing need for an objective intelligent methodology of monitoring and assessment of land degradation at a regional scale. This research is to develop an approach of geographic image cognition (GEOIC) to study land degradation and explore its applications by combining hyperspectral images, geographic information and multi-source data/information. The approach is developed based on the methodology of object-based image analysis (OBIA) and realized through the segmentation of land degradation spectral response units (DSRUs) using the diagnostic indicators related to land degradation. The approach was tested and validated in a study area, located in an agriculture-pasture mixed region in the edge of Loess Plateau area with complex physical and geographical situations and widely distributed land degradation.
In this research, the definition, conceptual issues, theoretical underpinning, and the framework of the GEOIC approach were first proposed. Its applications in mapping soil organic matter (SOM) and assessment of land degradation were investigated with the data collected in the study area. The GEOIC for the study of land degradation is to simulate the function and process of the visual interpretations of geoscience experts, and to extract spatial feature, spatial object and spatial pattern of land degradation from remote sensing images and multi-source information. Its realization was done through the DSRU segmentation by land type classification with integrating Hyperion images, geographic information, vegetation, soil, DEM and local information. The developed approach can improve the accuracy of the extraction of land degradation information. The research is the first attempt to apply the approach of GEOIC in the extraction of land degradation information. Moreover, the research provides a methodology of determining the diagnostic indicators related to land degradation and their combinations. The method of determining the diagnostic indicators from local farmers’ perception and from the comparison among different combinations of the diagnostic indicators was also proposed. The results showed that the overall classification accuracy was improved by 11.5% when the optimal combination of diagnostic indicators was used. The higher classification accuracy was achieved at an appropriate level for GEOIC approach than at similar pixel level and for the SAM method and DSLI method. The differences among the results with the GEOIC method and the methods of DSLI and SAM are significant. Investigation on SOM mapping with the approach of GEOIC was also performed. The GEOIC approach based on DSRU estimation models for soil parameter mapping is of advantage. The results using Hyperion images are comparable well with the field survey results and close to the results with the Kridge interpolation of soil samples. The developed method can be used for mapping soil features at a regional scale by integrating field data, remote sensing images and various regional variables.
|Description:||177 leaves : ill. (some col.) ; 30 cm.
PolyU Library Call No.: [THS] LG51 .H577P LSGI 2010 Wang
|URI:||http://hdl.handle.net/10397/4285||Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|b24250120_link.htm||For PolyU Users||162 B||HTML||View/Open|
|b24250120_ir.pdf||For All Users (Non-printable)||4.84 MB||Adobe PDF||View/Open|
Citations as of Mar 12, 2018
Citations as of Mar 12, 2018
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.