University of Twente Student Theses


Incorporation of time in data analysis for surface mineralogical mapping with multispectral remote sensing

Sampatirao, Akhil (2020) Incorporation of time in data analysis for surface mineralogical mapping with multispectral remote sensing.

Link to full-text:
(only accessible for UT students and staff)
Abstract:Geological time is very different from the general time; changes in geology do not happen in a month or a year, unlike its surroundings. As a result, the applications of multi-temporal remote sensing in mineralogical mapping are minimal. Although geology does not change in short periods, its surroundings – vegetation cover, atmospheric and illumination conditions change rapidly with time. With these changes in surroundings happening, the question ‘Do remotely obtained spectra of geological objects – rocks, outcrops, soil surfaces, etc. change with time?’ raises. Vegetation cover is considered one of the toughest challenges in Geological Remote Sensing (GRS), which is also the reason for most GRS studies to take place in desert-like areas. In most GRS studies, pixels rich in vegetation are masked out. Techniques such as spectral unmixing, principal component analysis, regression models, etc. were developed over the years to tackle the influence of vegetation cover on spectral signatures of geological objects. This research attempts to take on the challenge of vegetation cover by incorporating the paradigm of time. The abilities of time in tackling vegetation cover in geological remote sensing using multi-temporal Sentinel-2 imagery are evaluated in this research. Changes in remotely sensed spectra of geological objects due to natural vegetation over a period of thirteen months were observed to understand the level of information that can be retrieved from pixels with mixtures of vegetation and geological objects. Forty Sentinel 2 level 2A images over Rodalquilar, southeast Spain, were analyzed using Google Earth Engine; pixels were selected for tracing changes over time, and classified into ‘pure geology,’ ‘mixed,’ and ‘vegetation’ pixels. Time-series analysis was then performed on spectra, band ratios, and reflectance values of individual bands of all the selected pixels. Results from this research show that spectra of geological objects change with time; both pure geology and mixed pixels showed variations over time. Seasonal differences were observed in spectra and band ratios of mixed pixels; the trends were similar to that of the variation of vegetation in the region. Time helped in differentiating between signal due to different parts (vegetation, geology) in mixed pixels. Several possibilities of misinterpretations in the traditional GRS approach were observed. This research demonstrated the advantages of using multi-temporal remote sensing applications in surface mineralogical mapping and contributes to understanding the variations in remotely sensed spectra of geological objects over time.
Item Type:Essay
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page