Water-Quality Classification of Inland Lakes Using Landsat8 Images by Convolutional Neural Networks

Abstract

Water-quality monitoring of inland lakes is essential for freshwater-resource protection. In situ water-quality measurements and ratings are accurate but high costs limit their usage. Water-quality monitoring using remote sensing has shown to be cost-effective. However, the nonoptically active parameters that mainly determine water-quality levels in China are difficult to estimate because of their weak optical characteristics and lack of explicit correlation between remote-sensing images and parameters. To address the problems, a convolutional neural network (CNN) with hierarchical structure was designed to represent the relationship between Landsat8 images and in situ water-quality levels. A transfer-learning strategy in the CNN model was introduced to deal with the lack of in situ measurement data. After the CNN model was trained by spatially and temporally matched Landsat8 images and in situ water-quality data that were collected from official websites, the surface quality of the whole water body could be classified. We tested the CNN model at the Erhai and Chaohu lakes in China, respectively. The experiment results demonstrate that the CNN model outperformed widely used machine-learning methods. The trained model at Erhai Lake can be used for the water-quality classification of Chaohu Lake. The introduced CNN model and the water-quality classification method could cover the whole lake with low costs. The proposed method has potential in inland-lake monitoring.water-quality measurement, the cost is high, limiting the usage of in situ sampling and measurement. Optimization of a monitoring network is considered a trade-off between water-monitoring precision and cost [2]. A comprehensive and low-cost water-quality monitoring method was our goal.Remote-sensing monitoring of water quality and its parameters is considered as a promising and cost-effective monitoring method. There are many free-of-charge remote-sensing data, such as Landsat and Sentinel open-access data. Remote sensing provides a synoptic, repetitive, and consistent view of a water body. Water-quality-related variables, such as surface temperature, chlorophyll-a (Chl-a), turbidity and suspended solids (TSS) can be estimated from remote-sensing images according to their reflectance from the water surface [3]. The accuracy and stability of the remote-sensing inversion of surface-reflectance variables (Chl-a and SS) have improved after several years [4].Nonoptically active variables related to water pollution, such as total nitrogen (TN), total phosphorus (TP), chemical oxygen demand (COD), and dissolved oxygen (DO), cannot be directly sensed from water-surface reflectance because of their weak optical characteristics and low signal-to-noise ratio [3]. These nonoptically active variables are closely related to optically active parameters such as Chl-a, TSS, and colored dissolved organic matter (CDOM) [3,5]. Different kinds of regression models have been introduced to establish the relation between nonoptically active variables and optical active variabl...

U-Net Convolutional Networks for Mining Land Cover Classification Based on High-Resolution UAV Imagery

“…Recently, the artificial neural network (ANN) has become a useful tool to deal with complex non-linear regression between input data that has not yet been optimized for ancient remote sensing processing techniques such as unsupervised learning, Random Forest, pixel-based and Support Vector Machine [23]- [25]. For example, F. Pu et al [26] and K. B. Dang et al [27] used Convolutional Neural Network (CNN) to classify (1) water quality of inland lakes based on Landsat-8 images; and (2) coastal types in Vietnam based on Sentinel-2 images. W. Ge et al [28] used ANN models for lithological classification based on various moderate-resolution satellite images.…”

A Review of the Artificial Neural Network Models for Water Quality Prediction

“…In contrast, feed-forward neural network can predict water quality with relatively little data. In addition to being able to perform prediction tasks, GRNN is also suitable for small data sets (28,32,61,151,159, 265 samples) compared with other types of ANNs [24,[39][40][41][42][43], so researchers should pay some attention to it. The artificial neural network has been widely used in water quality prediction.…”

Comprehensive Review on Application of Machine Learning Algorithms for Water Quality Parameter Estimation Using Remote Sensing Data

“…(65) Pu et al used a CNN with a hierarchical structure to determine water quality levels using Landsat-8 imagery. (66) They used CNN to mitigate the problem of estimating water quality parameters, which occurs because of the weak optical characteristics of water and the lack of explicit correlation between RS imagery bands and parameters.…”