Retrieval of Chemical Oxygen Demand through Modified Capsule Network Based on Hyperspectral Data

Deng, Chubo; Zhang, Lifu; Yi, Chuixiang

doi:10.3390/app9214620

Cited by 16 publications

(9 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For TN, the average R 2 in the existing publications [21, [23][24][25][26][27] was 0.613, and the performance of TN retrieval in this research was improved by 43% over the average of the past research works. For COD, the average R 2 in the past research works [17][18][19][20] was 0.722, and the performance of retrieval COD in this research was slightly higher than the upper quartile of the past research works. Based on the above analysis, the performance of water quality retrieval in this research has improved by using a combination of machine learning strategies.…”

Section: Comparison Of Retrieval Accuracymentioning

confidence: 70%

“…Over the past several decades, scholars have carried out extensive research on water quality monitoring by remote sensing, and have achieved good results in estimating optically active parameters, such as Chlorophyll-a (Chl-a), suspended solids (SS), colored dissolved organic matter (CDOM), turbidity and transparency [11][12][13][14][15][16]. There are however few studies on non-optically active parameters, such as chemical oxygen demand (COD), total phosphorus (TP), and total nitrogen (TN) [17,18,[27][28][29][19][20][21][22][23][24][25][26]. Estimating non-optically active parameters directly from spectral characteristics is difficult.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Retrieval of Non-Optically Active Parameters for Small Scale Urban Waterbodies by a Machine Learning-Based Strategy

Huang¹,

Guo²,

Chen³

et al. 2020

Preprint

View full text Add to dashboard Cite

Water quality retrieval for small urban waterbodies by remote sensing get used to be difficult due to coarse spatial resolution of the remote sensing imagery. The recently launched Sentinel-2 produces imagery with a spatial resolution of 10 m. It provides an opportunity to solve the problem of retrieving water quality for small waterbodies. Additionally, many water management issues also require fine resolution of imagery, e.g. illegal discharge to an urban waterbody. Since illegal discharges are an important issue for urban water management, chemical oxygen demand (COD), total phosphorous (TP), and total nitrogen (TN) were chosen as the target parameters for water quality retrieval in this study. COD, TP and TN, however, are non-optically active parameters. There were limited studies in the past to retrieve these parameters in comparison with optically active parameters, e.g. Chlorophyll-A etc. This study compared three machine learning models, namely Random Forest (RF), Support Vector Regression (SVR), and Neural Networks (NN), to investigate the opportunity to retrieve the above non-optically active parameters. Results showed that R 2 of TP, TN, and COD by NN, RF and SVR were 0.94, 0.88, and 0.86, respectively. The performances of water quality retrieval for these non-optically active parameters were significantly improved by the optimized machine learning models. These models hence solved the problem to use remote sensing data to retrieve these non-optically active water quality parameters and provided a new monitoring strategy for small waterbodies. Water quality mapping obtained by Sentinel-2 imagery provided a full spatial coverage of the water quality characterization for the entire water surface. Compared with water samples collecting and testing, it greatly reduced labor cost, reagents cost, and waste treatment cost. It also may help identify illegal discharges to urban waterbodies. The method developed in this research provides a new practical and efficient water quality monitoring strategy in managing water with consideration of environmental sustainability.

show abstract

Section: Comparison Of Retrieval Accuracymentioning

confidence: 70%

Section: Introductionmentioning

confidence: 99%

Retrieval of Non-Optically Active Parameters for Small Scale Urban Waterbodies by a Machine Learning-Based Strategy

Huang¹,

Guo²,

Chen³

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…In addition, Sun et al believe that non-optically active parameters may be highly correlated with optically active substances, such as Chl-a, TSM, and CDOM [55], so TN, TP, and COD can be estimated remotely [15]. At present, some scholars have developed several statistical techniques with empirical and machine learning algorithms to determine the relationship between reflectance and non-optically active parameters in inland waters with the help of hyperspectral images [46,56]. As mentioned in Section 4.1, the three WQPs in this study have a certain correlation with some non-optically active water quality parameters (such as DO, COD Mn , and TN).…”

Section: Research Limitations and Prospectsmentioning

confidence: 99%

Retrieving Inland Reservoir Water Quality Parameters Using Landsat 8-9 OLI and Sentinel-2 MSI Sensors with Empirical Multivariate Regression

Meng

Zhang

Zhang³

2022

IJERPH

View full text Add to dashboard Cite

Improving water quality is one of the top priorities in the global agenda endorsed by the United Nations. To ensure the achievement of this goal, governments have developed plans to continuously monitor the status of inland waters. Remote sensing provides a low-cost, high-frequency, and practical complement to monitoring systems that can cover a large area. However, it is crucial to evaluate the suitability of sensors for retrieving water quality parameters (WQPs), owing to differences in spatial and spectral sampling from different satellites. Taking Shanmei Reservoir in Fuzhou City, Fujian Province as a case study, this study collected and sorted the water quality data measured at the site in 2020 to 2022 and Landsat 8-9 OLI and Sentinel-2 MSI images, simulated the chlorophyll-a (Chl-a) concentration, algae density, and turbidity using empirical multivariate regression, and explored the relationship between different WQPs using correlation analysis and principal component analysis (PCA). The results showed that the fitting effect of Landsat OLI data was better than that of the Sentinel-2 MSI data. The coefficient of determination (R2) values of Chl-a, algal density, and turbidity simulated by Landsat OLI data were 0.70, 0.81, and 0.80, respectively. Furthermore, the parameters of its validation equation were also smaller than those of Sentinel MSI data. The spatial distribution of three key WQPs retrieved from Landsat OLI data shows their values were generally low, with the mean values of the Chl-a concentration, algal density, and turbidity being 4.25 μg/L, 4.11 × 106 cells/L, and 1.86 NTU, respectively. However, from the end of February 2022, the values of the Chl-a concentration and algae density in the reservoir gradually increase, and the risk of water eutrophication also increases. Therefore, it is still necessary to pay continuous attention and formulate corresponding water quality management measures. The correlation analysis shows that the three key WQPs in this study have a high correlation with pH, water temperature (WT), and dissolved oxygen (DO). The results of PCA showed that pH, DO, Chl-a concentration, WT, TN, and CODMn were dominant in PC1, explaining 35.57% of the total variation, and conductivity, algal density, and WT were dominant in PC2, explaining 13.34% of the total variation. Therefore, the water quality of the Shanmei Reservoir can be better evaluated by measuring pH, conductivity, and WT at the monitoring station, or by establishing the regression fitting equations between DO, CODMn, and TN. The regression algorithm used in this study can identify the most important water quality features in the Shanmei Reservoir, which can be used to monitor the nutritional status of the reservoir and provide a reference for other similar inland water bodies.

show abstract

“…The presence of industrial wastewater, domestic sewage, and domestic waste has negatively impacted the ecological environment in some areas of Lake Baiyangdian [37,40,41], posing a considerable threat to the water environment's security. Lake Baiyangdian is facing serious eutrophication, its water quality urgently requires monitoring, and several studies have used remote sensing technology to assess the lake's water quality [32,33,42]. However, as an important indicator of water eutrophication, Lake Baiyangdian's TP remains insufficiently investigated.…”

Section: Introductionmentioning

confidence: 99%

Prediction of Total Phosphorus Concentration in Macrophytic Lakes Using Chlorophyll-Sensitive Bands: A Case Study of Lake Baiyangdian

Zhang

Cen

et al. 2022

Remote Sensing

View full text Add to dashboard Cite

Total phosphorus (TP) is a significant indicator of water eutrophication. As a typical macrophytic lake, Lake Baiyangdian is of considerable importance to the North China Plain’s ecosystem. However, the lake’s eutrophication is severe, threatening the local ecological environment. The correlation between chlorophyll and TP provides a mechanism for TP prediction. In view of the absorption and reflection characteristics of the chlorophyll concentrations in inland water, we propose a method to predict TP concentration in a macrophytic lake with spectral characteristics dominated by chlorophyll. In this study, water spectra noise is removed by discrete wavelet transform (DWT), and chlorophyll-sensitive bands are selected by gray correlation analysis (GRA). To verify the effectiveness of the chlorophyll-sensitive bands for TP concentration prediction, three different machine learning (ML) algorithms were used to build prediction models, including partial least squares (PLS), random forest (RF) and adaptive boosting (AdaBoost). The results indicate that the PLS model performs well in terms of TP concentration prediction, with the least time consumption: the coefficient of determination (R2) and root mean square error (RMSE) are 0.821 and 0.028 mg/L in the training dataset, and 0.741 and 0.029 mg/L in the testing dataset, respectively. Compared with the empirical model, the method proposed herein considers the correlation between chlorophyll and TP concentration, as well as a higher accuracy. The results indicate that chlorophyll-sensitive bands are effective for predicting TP concentration.

show abstract

Retrieval of Chemical Oxygen Demand through Modified Capsule Network Based on Hyperspectral Data

Cited by 16 publications

References 17 publications

Retrieval of Non-Optically Active Parameters for Small Scale Urban Waterbodies by a Machine Learning-Based Strategy

Retrieval of Non-Optically Active Parameters for Small Scale Urban Waterbodies by a Machine Learning-Based Strategy

Retrieving Inland Reservoir Water Quality Parameters Using Landsat 8-9 OLI and Sentinel-2 MSI Sensors with Empirical Multivariate Regression

Prediction of Total Phosphorus Concentration in Macrophytic Lakes Using Chlorophyll-Sensitive Bands: A Case Study of Lake Baiyangdian

Contact Info

Product

Resources

About