Mapping Water Quality Parameters in Urban Rivers from Hyperspectral Images Using a New Self-Adapting Selection of Multiple Artificial Neural Networks

Abstract

Protection of water environments is an important part of overall environmental protection; hence, many people devote their efforts to monitoring and improving water quality. In this study, a self-adapting selection method of multiple artificial neural networks (ANNs) using hyperspectral remote sensing and ground-measured water quality data is proposed to quantitatively predict water quality parameters, including phosphorus, nitrogen, biochemical oxygen demand (BOD), chemical oxygen demand (COD), and chlorophyll a. Seventy-nine ground measured data samples are used as training data in the establishment of the proposed model, and 30 samples are used as testing data. The proposed method based on traditional ANNs of numerical prediction involves feature selection of bands, self-adapting selection based on multiple selection criteria, stepwise backtracking, and combined weighted correlation. Water quality parameters are estimated with coefficient of determination R 2 ranging from 0.93 (phosphorus) to 0.98 (nitrogen), which is higher than the value (0.7 to 0.8) obtained by traditional ANNs. MPAE (mean percent of absolute error) values ranging from 5% to 11% are used rather than root mean square error to evaluate the predicting precision of the proposed model because the magnitude of each water quality parameter considerably differs, thereby providing reasonable and interpretable results. Compared with other ANNs with backpropagation, this study proposes an auto-adapting method assisted by the above-mentioned methods to select the best model with all settings, such as the number of hidden layers, number of neurons in each hidden layer, choice of optimizer, and activation function. Different settings for ANNS with backpropagation are important to improve precision and compatibility for different data. Furthermore, the proposed method is applied to hyperspectral remote sensing images collected using an unmanned aerial vehicle for monitoring the water quality in the Shiqi River, Zhongshan City, Guangdong Province, China. Obtained results indicate the locations of pollution sources.

Application of Drone Technologies in Surface Water Resources Monitoring and Assessment: A Systematic Review of Progress, Challenges, and Opportunities in the Global South

“…Meanwhile, Adongo et al [63] assessed the utility of undertaking bathymetric surveys combined with geographic information systems (GIS) functionalities in remotely determining the reservoir volume of nine irrigation dams in three northern regions of Ghana. On the other hand, the majority of water quality-related studies that were conducted based on drone remotely sensed data, principally mapped and monitored the chlorophyll content [30,32,33,37,38] and turbidity in lakes, ponds and dams (Figure 5b) [34][35][36]. This trend was also revealed through the bibliometric analysis illustrated in Figure 3.…”

“…As in Gholizadeh, Melesse and Reddi [7], the results of this study illustrated that most of the studies that utilised earth observation data sought to characterise water quality more than water quantity (Figure 3). The widely researched water quality parameters included conductivity [24,25], pH [25,26], Cl − [24], dissolved oxygen [27], total suspended solids (TSS) [28,29], chlorophyll [30][31][32][33], turbidity [34][35][36], K + , ammonium nitrogen (NH 4 -N), sodium (Na + ), BOD, magnesium (Mg), total phosphorous, orthophosphate (PO 4 -P), temperature and total nitrogen, iron (Fe), COD, zinc (Zn), calcium (Ca), manganese (Mn), salinity, copper (Cu), bicarbonate HCO 3− , sodium-absorbed ratio (SAR), coliform, cadmium (Cd), chromium (Cr), Ca 2+ , HCO 3− , and total hardness in order of frequency, as illustrated in Figure 4b. These parameters were mostly characterised using satellite remotely sensed data.…”

“…Specifically, only fourteen studies assessed the level of water, whereas thirty-seven studies assessed water quality parameters based on drone remotely sensed data. However, the majority of the aforementioned studies were conducted based on drone remotely sensed data, principally mapped and monitored chlorophyll content [30,32,33,37,38] and turbidity in lakes, ponds, and dams (Figure 5b) [34][35][36].…”

“…The advent of drone technologies has seen the utility of sensors, such as Nikon (NIKKOR AF-S 24-85 mm f/3.5-4.5G ED VR) and the Nikon D800 [47], GoPro Hero 4 Black Edition [48], Feiyu Mini 3D Pro [48], Sony [44], and CMOS [49] to the multispectral sensors such as the MicaSense, Parrot Sequoia [28,[50][51][52][53][54][55] Sentera [38], MicaSense RedEdge multispectral [29,56], and the hyperspectral sensors such as Headwall Photonics Inc (207 bands), Ocean Optics STS-VIS (640 bands) [27], AvaSpec-dual Gaia (640 bands) [35,57], Sky-mini Nano-Hyperspec [30], Canon EOS 5DS R, and Headwall Nano-Hyperspec (640 bands) for local-scale water remote sensing applications (Table 2). However, as the spectral resolution of drone sensors increases, the associated costs also increase linearly.…”

Applications of Unmanned Aerial Systems (UASs) in Hydrology: A Review

“…The authors point out that future work should aim to improve remote-sensing algorithms through the identification of the relationship between pigment cumulation depth and light attenuation. Zhang et al [142] mapped the concentration of eutrophication-related parameters in an urban river through the development of a self-adapting ANN based on UAS hyperspectral imagery and using the modified spectral reflectance of water measured with a ground-based analytical spectral device.…”