Please use this identifier to cite or link to this item: https://idr.l1.nitk.ac.in/jspui/handle/123456789/14039
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorAmba Shetty
dc.contributor.authorMinu S.
dc.date.accessioned2020-04-04T07:33:13Z-
dc.date.available2020-04-04T07:33:13Z-
dc.date.issued2018
dc.identifier.urihttp://idr.nitk.ac.in/jspui/handle/123456789/14039-
dc.description.abstractSoil organic carbon (SOC) is one of the most important constituents of soil due to its capacity to affect plant growth both as a source of energy and a trigger for nutrient availability through mineralization. Conventional laboratory methods of determination of SOC content are very laborious, time consuming and costly. For practical application, estimation of SOC from spaceborne VNIR/SWIR (Visible Near Infrared and Short wave infrared, 400-2500 nm) image using statistical regression is considered as an alternative technique. Spatial and spectral information from spaceborne hyperspectral VNIR/SWIR data can be used for quantification and better characterization of soil properties. The potential of this data has not been fully extracted till now because of noisy atmospheric components in spectral signature retrieved from spaceborne hyperspectral image. Though there are different atmospheric correction algorithms, retrieving biophysical characteristics from spaceborne hyperspectral data is a challenge. In this research, influence of atmospheric correction algorithms on the estimation of SOC is investigated. Research was initiated with ground VNIR/SWIR (400-2500 nm) spectroradiometer reflectance spectra obtained from bare agricultural sites in Narrabri, Australia, to find the effect of various pre-processing methods and estimation models on the estimation of SOC. Partial least square regression (PLSR) model performs better with Savitzky Golay as the best pre-processing method. The output from PLSR model was used to identify wavelengths that are significant in estimating SOC using a relative score defined as the product of the Variable Importance for Projection (VIP) values and the absolute value of PLS regression coefficient values. The most significant wavelengths in this PLSR model were located in the 600–680, 1860–1900, and 2180–2250 nm spectral regions. And, secondary significant wavelengths were located around 1000 and 2070 nm. Study was conducted to find the influence of atmospheric correction algorithms in the estimation of SOC from Hyperion data in sites located in two different geographical settings viz. Karnataka in India and Narrabri in Australia. Commonly used atmospheric correction algorithms, (1) ATmospheric CORection (ATCOR), (2) Fast Line-of-sight Atmospheric Analysis of Spectral Hypercubes (FLAASH), (3) 6S, and (4) QUick Atmospheric Correction (QUAC), were employed for retrieving spectral reflectance from radiance image. The results showed that ATCOR corrected spectra coupled with PLSR model, produced the best SOC estimation, in terms of coefficient of determination (R 2 ), Residual Prediction Deviation (RPD) and Ratio of Performance to Inter-Quartile (RPIQ), irrespective of the study area. Comparing the results across study areas, sites in Karnataka gave lower estimation accuracy than sites in Narrabri. This may be explained due to difference in spatial arrangement of field conditions. A spectral similarity comparison of atmospherically corrected Hyperion spectra of soil samples with field-measured VNIR/SWIR spectra was performed. Among the atmospheric correction algorithms, ATCOR corrected spectra is found to capture the pattern in soil reflectance curve near 2200 nm. ATCOR's finer spectral sampling distance in shortwave-infrared wavelength region compared to other models was identified as the main reason for its better performance. Two hybrid atmospheric correction (HAC) algorithms incorporating a modified empirical line (EL m ) method were proposed. The first HAC model (named HAC_1) combines i) a radiative transfer (RT) model based on the concepts of radiative transfer equations, which uses real-time in situ atmospheric and climatic data and ii) an EL m technique. The second one (named HAC_2) combines i) the well-known ATCOR model and ii) an EL m technique. Both HAC algorithms and their component single atmospheric correction algorithms (ATCOR, RT and EL m ) were applied to radiance data acquired by Hyperion satellite sensor over study sites in Australia. The performances of both HAC algorithms were analysed in two ways. Firstly, the Hyperion reflectances obtained by five atmospheric correction algorithms were analysed and compared using spectral metrics. Secondly, performance of each atmospheric correction algorithm was analysed for estimation of SOC using Hyperion reflectances obtained from atmospheric correction algorithms. The estimation model of SOC was built using PLSR model. The results show that i) both the hybrid models produce a good spectrum with lower Spectral Angle Mapper (SAM) and Spectral Information Divergence (SID) values and ii) both hybrid algorithms provided better SOC estimations accuracy, in terms of R 2 , RPD and RPIQ statistics. Thus, HAC algorithms, developed using EL m technique, may be recommended for atmospheric correction of Hyperion radiance data, when archived Hyperion reflectance data have to be used for SOC estimation mapping.
dc.publisherNational Institute of Technology Karnataka, Surathkal
dc.subjectDepartment of Applied Mechanics and Hydraulics
dc.titleEvaluation of Atmospheric Correction Algorithms in the Estimation of Soil Organic Carbon from Hyperion Image
dc.typeThesis
Appears in Collections:1. Ph.D Theses

Files in This Item:
File Description SizeFormat 
135010AM13P01.pdf8.22 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.