Browsing by Author "SHETTY, AMBA"

Now showing 1 - 2 of 2

Advanced Spectral Spatial Approaches for Dimensionality Reduction of Hyperspectral Data
(National Institute of Technology Karnataka, Surathkal, 2024) C, DEEPA; SHETTY, AMBA; NARASIMHADHAN, A.V.
Recent advances in sensor technology have enabled the collection of large data in hyperspectral remote sensing. Although rich spectral information is captured in hundreds of narrow contiguous bands, the hyperspectral data possess several limitations such as mixed pixels, high intraclass variability, interclass similarity, and the curse of dimensionality which restricts the potential of conventional machine learning classifiers. Dimensionality reduction (DR) and incorporation of spatial information can be taken into account to increase the interpretability of hyperspectral data. The thesis mainly focuses on the implementation of different approaches for DR of hyperspectral data to address the curse of dimensionality, limited samples and labelled data issues inherent in hyperspectral data. First, a quality measure based on the co-ranking matrix has been proposed for the performance evaluation of 15 DR techniques for mineral exploration. The selection of appropriate techniques for a particular task is challenging due to the diversity and ever-increasing number of DR techniques. A few important aspects in this regard have been explored in detail. Clustering is performed using the K-means algorithm and the relationship between the quality index and clustering accuracy has been examined concurrently for the first time in hyperspectral remote sensing. Furthermore, the loss of quality in the process of DR has also been analyzed which provides sufficient input for the end-user to select an appropriate DR technique. Second, the ability of the Convolutional Neural Network (CNN) for supervised learning of hyperspectral data is explored. A fast and compact hybrid CNN which combines the strengths of 3D and 2D convolutions to extract joint spectral-spatial information has been proposed to analyze the impact of different feature extraction techniques on classification performance. The effect of input patch size on final results has been well demonstrated. A detailed investigation of classification accuracy, execution time, and comparison with nine state-of-the-art approaches has been demonstrated. ii Next, a novel deep feature selection strategy using autoencoders inspired by knowledge distillation has been implemented for the model compression and selection of informative bands. The potential of convolutional autoencoders has been well explored in selecting discriminative bands. Sensitivity analysis tests and different applications have been considered to verify the generalization capability of the proposed model. The potential of unsupervised learning schemes has been discussed in detail. Finally, a generator model based on Generative Adversarial Networks (GAN) has been proposed for virtual sample generation and compact representation of hyperspectral data. The training instability issue in Vanilla GAN has been addressed by the effective implementation of deep convolutional GANs. By comparing the spectra of the generated hyperspectral images to the corresponding real ones, the quality of the images is assessed. The potential of augmented data for improvement in classification accuracy has also been investigated.
Flood Modeling and Mapping in the Upper Awash River Basin, Ethiopia
(National Institute of Technology Karnataka, Surathkal, 2024) TOLA, SINTAYEHU YADETE; SHETTY, AMBA
Climate variability, land cover change, and catchment characteristics significantly impact hydrological extremes. However, their impact on flood response behavior varies spatiotemporally, and quantifying possible causes is essential for effectively mitigating floods. Identifying the potential flood areas, and flood hazard mapping, considering changing environmental factors and an alternative flood frequency model for developing flood hazard management and mitigation strategies are crucial in the flood-prone basin. This thesis comprehensively investigates the potential factors affecting and explaining floods, possible flood sites, and flood hazard maps over the Upper Awash River Basin (UARB), Ethiopia, during the study period from 1985 to 2015. First, the study investigated the variability of extreme hydroclimatic conditions and the relationship between anomalies in extreme local precipitation, El Niño Southern Oscillation indicators (ENSO) (Southern Oscillation Index (SOI), Niño 3.4, and Multivariate ENSO Index (MEI)), and extreme flow indices. The analysis used standardized anomaly index and coefficient of variation statistics to examine variability, the modified Mann-Kendall and Pettitt tests for trend and change point analysis, and Spearman's correlation test to explore relationships. Results showed that the basin-wise extreme precipitation indices had less variability but higher variability spatially, while the extreme flow indices showed high variability. The maximum temperature increased significantly, while the minimum temperature decreased significantly (except at a few northwest stations), with a considerable shift in the 1990s and 2000s. Anomalies and a decrease in extreme precipitation were consistent with the extreme flow at the basin outlet, Hombole station. However, the extreme flow indices at Melka Kunture increased significantly and shifted upward (2003/2005), and the anomalies in extremely wet and very wet precipitation in the northwest were possibly responsible for this change. The annual wet and very wet days of precipitation strongly affected the extreme flow in the basin. The effect of annual wet day precipitation, maximum yearly precipitation, and ENSO anomalies on extreme flow at the Hombole were significant. ii Secondly, the study quantitatively assessed the effects of individual and coupled changes in land cover and climate on peak and high flows at Melka Kunture and Hombole over the baseline (1988-2001) and evaluation (2002-2015) period. The impact of these changes was estimated using the Soil Water and Assessment Tool (SWAT). The model satisfactorily simulated daily and extreme flows. The SWAT model showed that the main factor which affected the changes in upstream flow was the land cover change, increasing peak and high flow by 38.69% and 11.95%, respectively, compared to the baseline period. However, combined changes resulted in downstream peak and high flow reductions of 19.55% and 50.33%, respectively. In addition, the spatial flood characteristics based on morphometric parameters were performed in four subbasins to understand the hydrological behaviour better. The topographic wetness index (TWI) and topographic position index (TPI) were also used to determine the potential flood areas and inundation extent. The aggregated parameters revealed that subbasin SB-1 comprises Melka Kunture, is highly susceptible to flooding, SB-3 and SB-4 are moderately susceptible, and SB- 2 is low. The degree of susceptibility was also determined by incorporating the TWI and TPI through overlay analysis. The UARB accounts for 22.8%, 41.7%, and 35.6% of the total basin classified as high, medium, and low flood-prone, respectively. Furtherly the study developed a flood hazard map based on the nonstationary flood frequency using a generalized extreme value distribution model for the highly susceptible subbasin (SB-1), the identified flood spot area, Becho floodplain. The distributional location parameter was modeled as a function of rainfall amount of different durations, annual total precipitation from wet days, yearly mean maximum temperature, and time as covariates. The one-dimensional Hydrological Engineering Center River Analysis System (HEC-RAS) hydraulic model with steady flow analysis was used to generate flood hazard map input, depth and velocity, and inundation extent for different return periods. The result indicated that the model as a function of rainfall, such as monthly rainfall (August) and annual wet day precipitation, best fit the observed hydrological data. The developed hazard map based on depth alone and the combination of depth and velocity thresholds iii resulted in more than 70% of the floodplain area being classified as a high hazard zone under 2, 25, 50, and 100 years return periods.