Probabilistic machine learning-based phytoplankton abundance using hyperspectral remote sensing
Abstract
ABSTRACT Remote sensing is a crucial tool for understanding the spatial dynamics of algal blooms by quantifying and detecting algal proliferation in water bodies. Hyperspectral remote monitoring enables precise pigment concentration measurements of Cyanobacteria, facilitating the accurate quantification of algal blooms with high spatial and spectral resolutions. However, current water quality management policies in the Republic of Korea predominantly rely on phytoplankton concentration to assess algal bloom status, which presents challenges for effective pigment estimation from remote sensing. To address this gap, this study employed airborne remote sensing using hyperspectral imagery and a deep-learning approach to directly estimate phytoplankton cell concentrations across extensive water bodies. Airborne monitoring was conducted to comprehensively capture the spatiotemporal features of algal dynamics from 2016 to 2022, complemented by concurrent in situ assessments of phytoplankton concentration, including Cyanobacteria, diatom, and Green algae. Utilizing a Bayesian neural network and natural gradient-boosting algorithm, we simulated phytoplankton abundance using airborne remote sensing data. The probabilistic models achieved test accuracy with coefficients of determination (R2) of approximately 0.6 and 0.4 for the cell concentration of different algal phyla, respectively. Furthermore, the algorithms provided spatial distributions of algal cell concentrations, enabling the identification of critical management zones for water quality. This study demonstrates that probabilistic deep learning algorithms can deliver timely and accurate phytoplankton concentrations, improving decision-making processes in water quality management.
BibTeX
@inproceedings{Kumar2023SelfNoiseFrameworks,
author = {Kumar, Pawan and Ali, Murtiza and Nathwani, Karan},
booktitle = {OCEANS 2023 - Limerick},
title = {Self-Noise Cancellation in Underwater Acoustics using Deep Neural Network Frameworks},
year = {2023},
volume = {},
number = {},
pages = {1-6},
abstract = {The direction-of-arrival (DOA) estimation is a challenging task for towed array sonars in the presence of self-noise. Hence, self-noise cancellation (SNC) is necessary for correct detection and DOA estimation of the targets. Deep learning techniques, with their high-feature extraction capability and self-learning ability have been used for de-noising in image and audio processing, but have not been attempted in the underwater acoustics for SNC. We have therefore proposed SNC using autoencoders and VGG-16 based encoder-decoder (VGG-ED), trained in both supervised and semi-supervised manner. With the knowledge of the clean signal (without self-noise) and actual target DOAs, the autoencoder learns in a supervised manner to estimate the clean signal from the noisy signal. Since the clean signal is unavailable in reality, we also propose to use a semi-supervised learning approach. Herein, the autoencoder is trained with the estimated clean signal produced by the null space projection technique using self-noise and corresponding signal-to-interference-noise-ratio (SINR). The proposed autoencoder can reduce self-noise by 51 dB when SINR is −32 dB with a fewer sensors and snapshots.},
keywords = {Deep learning;Direction-of-arrival estimation;Sonar;Estimation;Interference;Sensors;Underwater acoustics;Self-noise cancellation;deep learning;convolutional neural networks;sensor array},
doi = {10.1109/OCEANSLimerick52467.2023.10244641},
issn = {},
month = {June},
}