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

^{1}

^{2}

^{3}

^{4}

^{*}

## Abstract

**:**

^{2}) 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 × 10

^{6}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 COD

_{Mn}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, COD

_{Mn}, 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.

## 1. Introduction

## 2. Materials and Methods

#### 2.1. Overview of the Study Area

^{2}, a length of 12 km, a maximum width of 7 km, and a maximum depth of 50 m, with a total storage capacity of 655 million m

^{3}and a normal water level of 96.48 m. In recent years, increasing attention has been paid to the water quality of Shanmei Reservoir. With the economic and social development of the catchment area, the reservoir has faced new or intensified challenges including (1) the increased pressure of eutrophication in the reservoir area, (2) the increased risk of seasonal algal blooms, and (3) TN functioning as a nutrient source. As a result, the optimization of the aquatic biological community structure and the reduction of the algal bloom risk have become key issues in the Shanmei Reservoir.

#### 2.2. Data Collection

#### 2.2.1. In Situ Data

_{Mn}(mg/L), ammonia nitrogen (NH

_{3}-N, mg/L), TN (mg/L), TP (mg/L), Chl-a concentration (mg/L), and algal density (cells/L). The monitoring frequency is once every four hours, and the monitoring data were dynamically released six times a day (0:00, 4:00, 8:00, 12:00, 16:00, and 20:00). The methods of measuring WQPs are detailed in technical specifications for automatic monitoring of surface water (HJ 915-2017) issued by the Ministry of Ecology and Environment of the People’s Republic of China (https://www.mee.gov.cn/ywgz/fgbz/bz/bzwb/jcffbz/201801/t20180108_429283.shtml, (accessed on 2 November 2021)). This study used Chl-a concentration, turbidity, and algal density data from 1 November 2020 to 26 February 2022, and the monitoring point was located at 118°24′52″ E, 25°10′54″ N. According to the introduction of Landsat 8-9 satellite and Sentinel-2 satellite, the acquisition time of both satellites is approximately 10:30 local time. Therefore, the WQPs data at 8:00 and 12:00 are selected for averaging, and the single remote sensing image pixel where the water quality monitoring station is located is selected to match the water quality data.

#### 2.2.2. Landsat 8-9 OLI Data

#### 2.2.3. Sentinel-2 MSI Data

#### 2.3. Methods

#### 2.3.1. Empirical Regression Modelling for Retrieval of WQPs

_{rs}(λ) is the reflectance corresponding to the Landsat 8 and Sentinel-2 bands, and a, b, and c are regression model constants.

#### 2.3.2. WQPs Retrieval Performance Analysis Metrics

^{2}), standard deviation (SD), standard error (SE), coefficient of variation (CV), mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE). Table 4 lists the calculation formula for each index. The flowchart in Figure 2 summarizes the overall flow of the research.

## 3. Results

#### 3.1. Behavior Parameters of In Situ WQPs at Sampling Station

^{6}cells/L, and a CV value of 141.17%. The turbidity fluctuated between 3.97 ± 2.84 (NTU) and its CV value reached 71.50%. The WT fluctuation was the smallest, with a CV value of only 20.70%, between 16.26 and 34.64 °C. In 2020–2022, the nutrient status and productivity of the reservoirs in the study area were relatively low.

#### 3.2. Landsat 8-9 OLI and Sentinel-2 MSI Reflectance (R_{rs}(λ)) Comparison

#### 3.3. WQPs Regressions from Landsat OLI and Sentinel MSI Data

#### 3.3.1. Regression of Chl-a Concentration

^{2}values. This confirmed the plausibility of the developed regression model for estimating the Chl-a concentration in the case study reservoir. For Landsat OLI data, we found the three best regression equations fits for Chl-a concentration retrieval using a combination of coastal aerosol (B1), blue (B2), green (B3), and NIR (B4) bands, and the highest value of R

^{2}is 0.70. Compared with the Sentinel MSI regression equations using the red (B4), Vegetation red edge1 (B5), and Vegetation red edge2 (B6) bands to predict Chl-a concentrations, the accuracy was improved by a minimum of 12.90%.

#### 3.3.2. Regression of Algal Density

^{2}of 0.82, and for Sentinel MSI data using the red (B4), Vegetation red edge1 (B5), and Vegetation red edge2 (B6) bands, where the R

^{2}of the best regression equation is 0.61. It can be seen that the accuracy of the regression formula of Landsat OLI data on algal density was at least 26.23% higher than that of Sentinel MSI data. The results show that although the measured algal density values have a wide range, the Landsat OLI retrieval regression equation can verify the validity of the established model and predict algal density to a certain extent.

#### 3.3.3. Regression of Turbidity

^{2}= 0.71, while the best estimate of turbidity by Sentinel MSI data was only 0.14. This demonstrates the decisive advantage of the Landsat OLI data in the retrieval of reservoir turbidity in the study area.

#### 3.4. Validation of Water Quality Prediction with In Situ Sampling

#### 3.5. Spatial Distribution of WQPs Retrieved by Landsat OLI Data

#### 3.5.1. Distribution of Chl-a Concentration

#### 3.5.2. Distribution of Algal Density

^{4}and 3.02 × 10

^{7}cells/L, and the spatial distribution was relatively consistent with the Chl-a concentration. Although the color change in the algal density classification was stronger overall than that of the Chl-a concentration, the algal density had more low values and fewer high values, so the actual numerical change was not high. The average value from 2020 to 2022 was 4.11 × 10

^{6}cells/L.

#### 3.5.3. Distribution of Turbidity

#### 3.6. Relationship between Reservoir WPQs

## 4. Discussion

#### 4.1. Analysis of Current Water Quality Status and Relationship between WQPs

_{Mn}dominated PC1, which explained 35.57% of the total variance, and conductivity, algal density, and WT dominated PC2, which accounted for 14.91%, indicating the importance of pH, DO, Chl-a concentration, WT, TN, COD

_{Mn}, and conductivity in estimating water quality in the study area.

_{Mn}mainly uses conventional satellite remote sensing (such as GF series satellites, Landsat series satellites) [42,43], while the retrieval of DO and TN using hyperspectral remote sensing has higher accuracy [44,45,46]. Combined with the results of the above-related studies, it can be considered that the water quality of Shanmei Reservoir can be better evaluated by measuring pH, conductivity, and WT at the monitoring station, or by establishing the regression fitting equations between Chl-a, algae density, and turbidity and DO, COD

_{Mn}, and TN.

#### 4.2. Selection and Applicability Analysis of Retrieval Band and Algorithm

#### 4.3. WQPs Regression Formula and Retrieval Error Analysis

_{Mn}, and TN, so as to ensure the good water quality of the reservoir.

#### 4.4. Research Limitations and Prospects

_{Mn}, and TN). Therefore, the WQPs of regional non-optically active water quality can be estimated through machine learning algorithms. In the ideal future, the acquisition frequency and accuracy of satellite images will be the same as that of water sample data, so as to reduce the time difference between different satellite images. At the same time, more matching data can be obtained by adding monitoring points or stations at different locations, as was performed by Curtarelli et al. who arranged them in the reservoir near the dam, in the middle of the reservoir, at the tail of the reservoir, and near the tributary [57]. In addition, the impact on the water quality of inland reservoirs can also be studied in terms of hydrological changes such as water volume and reservoir depth [58], so as to more comprehensively judge the current status and future trends of reservoir water quality. At the same time, for remote sensing data, preprocessing and adjacency affect the selection or development of corresponding algorithms for correction and the retrieval of more accurate water quality data, which can be used for water resources management and environmental protection planning.

## 5. Conclusions

^{2}and validation parameters (MAE, MSE, and RMSE) of the Landsat OLI fitting equation are better than Sentinel MSI data. Therefore, Landsat OLI data have better application potential in this study area. The 2020–2022 reservoir water quality images retrieved from Landsat OLI data show that the multi-month average values of reservoir WQPs are low. However, from the end of February 2022, the Chl-a concentration and algal density in the reservoir gradually increased, and local high values appeared. Therefore, continuous attention and corresponding water quality management measures are still needed. The results of correlation analysis and principal component analysis show that the water quality of Shanmei Reservoir can be evaluated more accurately and quickly by measuring the pH, conductivity, and WT of the monitoring station, or by establishing the regression fitting equation between Chl-a, algae density, and turbidity and DO, COD

_{Mn}, and TN. In the future, to improve the accuracy of the estimation of the overall water quality status of the reservoir, new methods can be developed to monitor, fit, and retrieve more factors that can represent the water quality status, or understand the impact of the algorithm on the different performances of remote sensing images to conduct frequent water quality assessments. Simultaneously, we can also apply the regression equation from the study area to verify the accuracy of the regression formula in adjacent waters or similar waters.

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## References

- Pekel, J.; Cottam, A.; Gorelick, N.; Belward, A.S. High-resolution mapping of global surface water and its long-term chang-es. Nature
**2016**, 540, 418–422. [Google Scholar] [CrossRef] [PubMed] - Zhou, S.; Zhang, X.; Liu, J. The trend of small hydropower development in China. Renew. Energ.
**2009**, 34, 1078–1083. [Google Scholar] [CrossRef] - Khatri, N.; Tyagi, S. Influences of natural and anthropogenic factors on surface and groundwater quality in rural and urban areas. Front. Life Sci.
**2014**, 8, 23–39. [Google Scholar] [CrossRef] - Kulshreshtha, S.N. A Global Outlook for Water Resources to the Year 2025. Water Resour. Manag.
**1998**, 12, 167–184. [Google Scholar] [CrossRef] - Neves, V.H.; Pace, G.; Delegido, J.; Antunes, S.C. Chlorophyll and Suspended Solids Estimation in Portuguese Reservoirs (Aguieira and Alqueva) from Sentinel-2 Imagery. Water
**2021**, 13, 2479. [Google Scholar] [CrossRef] - Cairo, C.; Barbosa, C.; Lobo, F.; Novo, E.; Carlos, F.; Maciel, D.; Flores Júnior, R.; Silva, E.; Curtarelli, V. Hybrid Chloro-phyll-a Algorithm for Assessing Trophic States of a Tropical Brazilian Reservoir Based on MSI/Sentinel-2 Data. Remote Sens.
**2020**, 12, 40. [Google Scholar] [CrossRef][Green Version] - Watanabe, F.; Alcântara, E.; Rodrigues, T.; Imai, N.; Barbosa, C.; Rotta, L. Estimation of Chlorophyll-a Concentration and the Trophic State of the Barra Bonita Hydroelectric Reservoir Using OLI/Landsat-8 Images. Int. J. Env. Res. Pub. Health
**2015**, 12, 10391–10417. [Google Scholar] [CrossRef] - Bennett, E.M.; Carpenter, S.R.; Caraco, N.F. Human Impact on Erodable Phosphorus and Eutrophication: A Global Perspective. BioScience
**2001**, 51, 227. [Google Scholar] [CrossRef] - Ouma, Y.O.; Waga, J.; Okech, M.; Lavisa, O.; Mbuthia, D. Estimation of Reservoir Bio-Optical Water Quality Parameters Us-ing Smartphone Sensor Apps and Landsat ETM+: Review and Comparative Experimental Results. J. Sens.
**2018**, 2018, 3490757. [Google Scholar] [CrossRef] - Gordon, H.R.; Brown, O.B.; Evans, R.H.; Brown, J.W.; Smith, R.C.; Baker, K.S.; Clark, D.K. A semianalytic radiance model of ocean color. J. Geophys. Res.
**1988**, 93, 10909. [Google Scholar] [CrossRef] - Hossain, A.K.M.A.; Mathias, C.; Blanton, R. Remote Sensing of Turbidity in the Tennessee River Using Landsat 8 Satellite. Remote Sens.
**2021**, 13, 3785. [Google Scholar] [CrossRef] - Deutsch, E.S.; Fortin, M.; Cardille, J.A. Assessing the current water clarity status of ~100,000 lakes across southern Canada: A remote sensing approach. Sci. Total Environ.
**2022**, 826, 153971. [Google Scholar] [CrossRef] - Demetriou, M.; Raitsos, D.E.; Kournopoulou, A.; Mandalakis, M.; Sfenthourakis, S.; Psarra, S. Phytoplankton Phenology in the Coastal Zone of Cyprus, Based on Remote Sensing and In Situ Observations. Remote Sens.
**2022**, 14, 12. [Google Scholar] [CrossRef] - Li, Y.; Zhang, Y.; Shi, K.; Zhu, G.; Zhou, Y.; Zhang, Y.; Guo, Y. Monitoring spatiotemporal variations in nutrients in a large drinking water reservoir and their relationships with hydrological and meteorological conditions based on Landsat 8 im-agery. Sci. Total Environ.
**2017**, 599–600, 1705–1717. [Google Scholar] [CrossRef] - Guo, H.; Huang, J.J.; Chen, B.; Guo, X.; Singh, V.P. A machine learning-based strategy for estimating non-optically active water quality parameters using Sentinel-2 imagery. Int. J. Remote Sens.
**2021**, 42, 1841–1866. [Google Scholar] [CrossRef] - Andrade, C.; Alcântara, E.; Bernardo, N.; Kampel, M. An assessment of semi-analytical models based on the absorption co-efficient in retrieving the chlorophyll-a concentration from a reservoir. Adv. Space Res.
**2019**, 63, 2175–2188. [Google Scholar] [CrossRef] - Ouma, Y.O.; Noor, K.; Herbert, K. Modelling Reservoir Chlorophyll-a, TSS, and Turbidity Using Sentinel-2A MSI and Landsat-8 OLI Satellite Sensors with Empirical Multivariate Regression. J. Sens.
**2020**, 2020, 8858408. [Google Scholar] [CrossRef] - Lee, Z.; Carder, K.L.; Arnone, R.A. Deriving inherent optical properties from water color: A multiband quasi-analytical al-gorithm for optically deep waters. Appl. Opt.
**2002**, 41, 5755–5772. [Google Scholar] [CrossRef] - Campbell, G.; Phinn, S.R.; Dekker, A.G.; Brando, V.E. Remote sensing of water quality in an Australian tropical freshwater impoundment using matrix inversion and MERIS images. Remote Sens. Environ.
**2011**, 115, 2402–2414. [Google Scholar] [CrossRef][Green Version] - Hoge, F.E.; Lyon, P.E. Satellite retrieval of inherent optical properties by linear matrix inversion of oceanic radi-ance models: An analysis of model and radiance measurement errors. J. Geophys. Res.
**1996**, 101, 16631–16648. [Google Scholar] [CrossRef] - Toming, K.; Kutser, T.; Laas, A.; Sepp, M.; Paavel, B.; Nõges, T. First Experiences in Mapping Lake Water Quality Parameters with Sentinel-2 MSI Imagery. Remote Sens.
**2016**, 8, 640. [Google Scholar] [CrossRef][Green Version] - Potes, M.; Rodrigues, G.; Penha, A.M.; Novais, M.H.; Costa, M.J.; Salgado, R.; Morais, M.M. Use of Sentinel 2–MSI for water quality monitoring at Alqueva reservoir, Portugal. Proc. Int. Assoc. Hydrol. Sci.
**2018**, 380, 73–79. [Google Scholar] [CrossRef] - El Behairy, R.A.; El Baroudy, A.A.; Ibrahim, M.M.; Kheir, A.M.S.; Shokr, M.S. Modelling and Assessment of Irrigation Water Quality Index Using GIS in Semi-arid Region for Sustainable Agriculture. Water Air Soil Pollut.
**2021**, 232, 352. [Google Scholar] [CrossRef] - Baroudy, A.A.E.; Ali, A.M.; Mohamed, E.S.; Moghanm, F.S.; Shokr, M.S.; Savin, I.; Poddubsky, A.; Ding, Z.; Kheir, A.M.S.; Aldosari, A.A.; et al. Modeling Land Suitability for Rice Crop Using Remote Sensing and Soil Quality Indicators: The Case Study of the Nile Delta. Sustainability
**2020**, 12, 9653. [Google Scholar] [CrossRef] - Hammam, A.A.; Mohamed, W.S.; Sayed, S.E.; Kucher, D.E.; Mohamed, E.S. Assessment of Soil Contamination Using GIS and Multi-Variate Analysis: A Case Study in El-Minia Governorate, Egypt. Agronomy
**2022**, 12, 1197. [Google Scholar] [CrossRef] - Topp, S.N.; Pavelsky, T.M.; Jensen, D.; Simard, M.; Ross, M.R.V. Research Trends in the Use of Remote Sensing for Inland Water Quality Science: Moving Towards Multidisciplinary Applications. Water
**2020**, 12, 169. [Google Scholar] [CrossRef][Green Version] - Ma, C.; Zhao, J.; Ai, B.; Sun, S.; Yang, Z. Machine Learning Based Long-Term Water Quality in the Turbid Pearl River Estuary, China. J. Geophys. Res. Ocean.
**2022**, 127, e2021JC018017. [Google Scholar] [CrossRef] - Schaeffer, B.; Salls, W.; Coffer, M.; Lebreton, C.; Werther, M.; Stelzer, K.; Urquhart, E.; Gurlin, D. Merging of the Case 2 Regional Coast Colour and Maximum-Peak Height chlorophyll-a algorithms: Validation and demonstration of satel-lite-derived retrievals across US lakes. Environ. Monit. Assess.
**2022**, 194, 179. [Google Scholar] [CrossRef] - Masoud, A.A. On the Retrieval of the Water Quality Parameters from Sentinel-3/2 and Landsat-8 OLI in the Nile Delta’s Coastal and Inland Waters. Water
**2022**, 14, 593. [Google Scholar] [CrossRef] - Wang, X.; Gong, Z.; Pu, R. Estimation of chlorophyll a content in inland turbidity waters using WorldView-2 imagery: A case study of the Guanting Reservoir, Beijing, China. Environ. Monit. Assess.
**2018**, 190, 620. [Google Scholar] [CrossRef] - Rodríguez-López, L.; Duran-Llacer, I.; González-Rodríguez, L.; Cardenas, R.; Urrutia, R. Retrieving Water Turbidity in Ar-aucanian Lakes (South-Central Chile) Based on Multispectral Landsat Imagery. Remote Sens.
**2021**, 13, 3133. [Google Scholar] [CrossRef] - Rodríguez-López, L.; Duran-Llacer, I.; González-Rodríguez, L.; Abarca-del-Rio, R.; Cárdenas, R.; Parra, O.; Mar-tínez-Retureta, R.; Urrutia, R. Spectral analysis using LANDSAT images to monitor the chlorophyll-a concentration in Lake Laja in Chile. Ecol. Inform.
**2020**, 60, 101183. [Google Scholar] [CrossRef] - Liu, M.; Chen, X.; Chen, Y.; Gao, L.; Deng, H. Nitrogen Retention Effects under Reservoir Regulation at Multiple Time Scales in a Subtropical River Basin. Water
**2019**, 11, 1685. [Google Scholar] [CrossRef][Green Version] - Wu, J.; Chen, X.; Yao, H.; Gao, L.; Chen, Y.; Liu, M. Non-linear relationship of hydrological drought responding to meteor-ological drought and impact of a large reservoir. J. Hydrol.
**2017**, 551, 495–507. [Google Scholar] [CrossRef] - Ogashawara, I.; Mishra, D.R.; Gitelson, A.A. Remote Sensing of Inland Waters. In Bio-Optical Modeling and Remote Sensing of In-land Waters; Elsevier: Amsterdam, The Netherlands, 2017; pp. 1–24. [Google Scholar]
- Fan, D.; Huang, Y.; Song, L.; Liu, D.; Zhang, G.; Zhang, B. Prediction of chlorophyll a concentration using HJ-1 satellite im-agery for Xiangxi Bay in Three Gorges Reservoir. Water Sci. Eng.
**2014**, 7, 70–80. [Google Scholar] - Boucher, J.; Weathers, K.C.; Norouzi, H.; Steele, B. Assessing the effectiveness of Landsat 8 chlorophylla re-trieval algorithms for regional freshwater monitoring. Ecol. Appl.
**2018**, 28, 1044–1054. [Google Scholar] [CrossRef][Green Version] - Elhag, M.; Gitas, I.; Othman, A.; Bahrawi, J.; Gikas, P. Assessment of Water Quality Parameters Using Temporal Remote Sensing Spectral Reflectance in Arid Environments, Saudi Arabia. Water
**2019**, 11, 556. [Google Scholar] [CrossRef][Green Version] - Brivio, P.A.; Giardino, C.; Zilioli, E. Determination of chlorophyll concentration changes in Lake Garda using an im-age-based radiative transfer code for Landsat TM images. Int. J. Remote Sens.
**2001**, 22, 487–502. [Google Scholar] [CrossRef] - Mokarram, M.; Sathyamoorthy, D. Relationship between landform classification and vegetation (case study: Southwest of Fars province, Iran). Open Geosci.
**2016**, 8, 302–309. [Google Scholar] [CrossRef][Green Version] - Grendaitė, D.; Stonevičius, E.; Karosienė, J.; Savadova, K.; Kasperovičienė, J. Chlorophyll-a concentration retrieval in eu-trophic lakes in Lithuania from Sentinel-2 data. Geologija. Geografija
**2018**, 4, 15–28. [Google Scholar] [CrossRef] - Feng, Y.B.; He, Y.Q.; Fu, Q.H.; Liu, C.Q.; Pan, H.Z.; Yin, B. Estimation Cod&MN; IN Guangzhou section of pearl river based on gf-1 images. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci.
**2018**, XLII-3, 349–354. [Google Scholar] - He, Y.; Gong, Z.; Zheng, Y.; Zhang, Y. Inland Reservoir Water Quality Inversion and Eutrophication Evaluation Using BP Neural Network and Remote Sensing Imagery: A Case Study of Dashahe Reservoir. Water
**2021**, 13, 2844. [Google Scholar] [CrossRef] - Kwon, Y.S.; Pyo, J.; Kwon, Y.; Duan, H.; Cho, K.H.; Park, Y. Drone-based hyperspectral remote sensing of cyanobacteria using vertical cumulative pigment concentration in a deep reservoir. Remote Sens. Environ.
**2020**, 236, 111517. [Google Scholar] [CrossRef] - Rostom, N.G.; Shalaby, A.A.; Issa, Y.M.; Afifi, A.A. Evaluation of Mariut Lake water quality using Hyperspectral Remote Sensing and laboratory works. Egypt. J. Remote. Sens. Space Sci.
**2017**, 20, S39–S48. [Google Scholar] [CrossRef][Green Version] - Qun’Ou, J.; Lidan, X.; Siyang, S.; Meilin, W.; Huijie, X. Retrieval model for total nitrogen concentration based on UAV hyper spectral remote sensing data and machine learning algorithms–A case study in the Miyun Reservoir, China. Ecol. Indic.
**2021**, 124, 107356. [Google Scholar] [CrossRef] - Kapalanga, T.S.; Hoko, Z.; Gumindoga, W.; Chikwiramakomo, L. Remote-sensing-based algorithms for water quality mon-itoring in Olushandja Dam, north-central Namibia. Water Supply
**2021**, 21, 1878–1894. [Google Scholar] [CrossRef] - Buma, W.G.; Lee, S. Evaluation of Sentinel-2 and Landsat 8 Images for Estimating Chlorophyll-a Concentrations in Lake Chad, Africa. Remote Sens.
**2020**, 12, 2437. [Google Scholar] [CrossRef] - Lai, Y.; Zhang, J.; Song, Y.; Gong, Z. Retrieval and Evaluation of Chlorophyll-a Concentration in Reservoirs with Main Wa-ter Supply Function in Beijing, China, Based on Landsat Satellite Images. Int. J. Env. Res. Pub. Health
**2021**, 18, 4419. [Google Scholar] [CrossRef] - Johansen, R.; Beck, R.; Nowosad, J.; Nietch, C.; Xu, M.; Shu, S.; Yang, B.; Liu, H.; Emery, E.; Reif, M.; et al. Evaluating the portability of satellite derived chlorophyll-a algorithms for temperate inland lakes using airborne hyperspectral imagery and dense surface observations. Harmful Algae.
**2018**, 76, 35–46. [Google Scholar] [CrossRef] - Moses, W.J.; Sterckx, S.; Montes, M.J.; Keukelaere, L.D.; Knaeps, E. Atmospheric Correction for Inland Waters-ScienceDirect. In Bio-Optical Modeling and Remote Sensing of In-Land Waters; Elsevier: Amsterdam, The Netherlands, 2017; pp. 69–100. [Google Scholar]
- Díaz-Torres, O.; de Anda, J.; Lugo-Melchor, O.Y.; Pacheco, A.; Orozco-Nunnelly, D.A.; Shear, H.; Senés-Guerrero, C.; Gradil-la-Hernández, M.S. Rapid Changes in the Phytoplankton Community of a Subtropical, Shallow, Hypereutrophic Lake Dur-ing the Rainy Season. Front. Microbiol.
**2021**, 12, 415. [Google Scholar] [CrossRef] - Bergen, T.J.H.M.; Barros, N.; Mendonça, R.; Aben, R.C.H.; Althuizen, I.H.J.; Huszar, V.; Lamers, L.P.M.; Lürling, M.; Roland, F.; Kosten, S. Seasonal and diel variation in greenhouse gas emissions from an urban pond and its major drivers. Limnol. Oceanogr.
**2019**, 64, 2129–2139. [Google Scholar] [CrossRef] - Valipour, R.; León, L.F.; Howell, T.; Dove, A.; Rao, Y.R. Episodic nearshore-offshore exchanges of hypoxic waters along the north shore of Lake Erie. J. Great Lakes Res.
**2021**, 47, 419–436. [Google Scholar] [CrossRef] - Sun, X.; Zhang, Y.; Shi, K.; Zhang, Y.; Li, N.; Wang, W.; Huang, X.; Qin, B. Monitoring water quality using proximal remote sensing technology. Sci. Total Environ.
**2022**, 803, 149805. [Google Scholar] [CrossRef] - Deng, C.; Zhang, L.; Cen, Y. Retrieval of Chemical Oxygen Demand through Modified Capsule Network Based on Hyperspectral Data. Appl. Sci.
**2019**, 9, 4620. [Google Scholar] [CrossRef][Green Version] - Curtarelli, M.; Neto, E.; de Siqueira, F.; Yopan, F.; Soares, G.; Pauli, G.; de Souza, J.; Silva, L.; Sagaz, M.; Demay, M.; et al. QDA-System: A Cloud-Based System for Monitoring Water Quality in Brazilian Hydroelectric Reservoirs from Space. Remote Sens.
**2022**, 14, 1541. [Google Scholar] [CrossRef] - Tamm, T.; Nõges, T.; Järvet, A.; Bouraoui, F. Contributions of DOC from surface and groundflow into Lake Võrtsjärv (Esto-nia). Hydrobiologia
**2008**, 599, 213–220. [Google Scholar] [CrossRef]

**Figure 1.**(

**a**) Fujian Province in the yellow section is located on the southeastern coast of China, (

**b**) Quanzhou City in the pink section is located at the eastern end of central Fujian Province, and (

**c**) the boundary of Shanmei Reservoir. The green point is the location of the monitoring station.

**Figure 3.**Reflectance variation in different bands for (

**a**) Landsat OLI data and (

**b**) Sentinel MSI data.

**Figure 5.**In situ-derived Chl-a concentration and Landsat OLI bands reflectance from training datasets.

**Figure 11.**Correlation coefficient diagram between WQPs. Con, conductivity, Tur, turbidity, AD, algal density. The diagonal line gives the distribution, histogram, and density curve of WQPs. The lower triangle (the lower-left corner of the diagonal) gives the scatter diagram and the upper triangle (the upper-right corner of the diagonal) between the two WQPs. The value represents the correlation coefficient of the two variables. The larger the value, the greater the correlation degree; the asterisk indicates the degree of significance, * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001.

Number | Year | Image Date | Number | Year | Image Date |
---|---|---|---|---|---|

1 | 2020 | 11 November | 8 | 2021 | 25 July |

2 | 2020 | 27 November | 9 | 2021 | 26 August |

3 | 2020 | 29 December | 10 | 2021 | 11 September |

4 | 2021 | 30 January | 11 | 2021 | 27 September |

5 | 2021 | 15 February | 12 | 2021 | 16 December |

6 | 2021 | 6 May | 13 | 2022 | 1 January |

7 | 2021 | 9 July | 14 | 2022 | 26 February |

Number | Year | Image Date | Number | Year | Image Date |
---|---|---|---|---|---|

1 | 2020 | 9 November | 15 | 2021 | 16 August |

2 | 2021 | 29 December | 16 | 2021 | 26 August |

3 | 2021 | 13 January | 17 | 2021 | 5 September |

4 | 2021 | 18 January | 18 | 2021 | 25 September |

5 | 2021 | 23 January | 19 | 2021 | 5 October |

6 | 2021 | 2 February | 20 | 2021 | 14 November |

7 | 2021 | 17 February | 21 | 2021 | 24 November |

8 | 2021 | 9 March | 22 | 2021 | 04 December |

9 | 2021 | 24 March | 23 | 2021 | 09 December |

10 | 2021 | 12 June | 24 | 2021 | 14 December |

11 | 2021 | 17 June | 25 | 2021 | 19 December |

12 | 2021 | 7 July | 26 | 2021 | 29 December |

13 | 2021 | 22 July | 27 | 2022 | 3 January |

14 | 2021 | 27 July | 28 | 2022 | 8 January |

Band Combinations | Landsat OLI Data | Sentinel MSI Data |
---|---|---|

Single bands | B_{L8i} | B_{S2i} |

Linear band combination | B_{L8i} + B_{L8j} | B_{S2i} + B_{S2j} |

Band ratios | B_{L8i}/B_{L8j} | B_{S2i}/B_{S2j} |

Mixed band combinations | (B_{L8i} + B_{L8j})/B_{L8k} | (B_{S2i} + B_{S2j})/B_{S2k} |

Name | Equation |
---|---|

Correlation coefficient | $r=\frac{{\displaystyle \sum _{i=1}^{n}\left(\left({X}_{i}^{Measured}-{\overline{X}}^{Measured}\right)\left({Y}_{i}^{Estimated}-{\overline{Y}}^{Estimated}\right)\right)}}{\sqrt{{\displaystyle \sum _{i=1}^{n}{\left({X}_{i}^{Measured}-{\overline{X}}^{Measured}\right)}^{2}}}\sqrt{{\displaystyle \sum _{i=1}^{n}({Y}_{i}^{Estimated}-{\overline{Y}}^{Estimated}{)}^{2}}}}$ |

Determination coefficient | ${R}^{2}=1-\frac{{\displaystyle \sum _{i=1}^{n}\left({X}_{i}^{Measured}-{X}_{i}^{E\mathrm{stimated}}\right)}}{{\displaystyle \sum _{i=1}^{n}\left({X}_{i}^{Measured}-{\overline{X}}_{}^{E\mathrm{stimated}}\right)}}$ |

Standard deviation (SD) | $\mathrm{SD}=\sqrt{\frac{1}{n}\times {\displaystyle \sum _{i}^{\mathrm{n}}{({X}_{i}-\overline{X})}^{2}}}$ |

Standard error (SE) | $\mathrm{SE}=\frac{\mathrm{SD}}{\sqrt{n}}$ |

Coefficient of variation (CV) | $\mathrm{CV}=\frac{\mathrm{SD}}{\overline{X}}$ |

Mean absolute error (MAE) | $\mathrm{MAE}=\frac{1}{n}\times {\displaystyle \sum _{i=1}^{n}\left|{X}_{i}^{Estimated}-{X}_{i}^{Measured}\right|}$ |

Mean square error (MSE) | $\mathrm{MSE}=\frac{1}{n}\times {{\displaystyle \sum _{i=1}^{n}\left({X}_{i}^{Estimated}-{X}_{i}^{Measured}\right)}}^{2}$ |

Root mean square error (RMSE) | $\mathrm{RMSE}=\sqrt{\mathrm{MSE}}$ |

Mean absolute percentage error (MAPE) | $\mathrm{MAPE}=\frac{100}{n}\times {\displaystyle \sum _{i=1}^{n}\left(\frac{{X}_{i}^{Estimated}-{X}_{i}^{Measured}}{{X}_{i}^{Estimated}}\right)}$ |

Parameters | Statistical Indicators | Numerical Value |
---|---|---|

Chl-a concentration | min-max (μg/L) | 1.00–23.11 |

Average ± σ ^{1} (μg/L) | 5.24 ± 3.04 | |

CV(%) | 58.12 | |

n ^{1} | 2819 | |

Algal Density | min-max (10^{6} cells/L) | 0.06–102.11 |

Average ± σ (10^{6} cells/L) | 4.46 ± 6.29 | |

CV (%) | 141.17 | |

n | 2819 | |

Turbidity | min-max (NTU) | 0.62–26.36 |

Average ± σ (NTU) | 3.97 ± 2.84 | |

CV(%) | 71.50 | |

n | 2819 | |

WT | min-max (° C) | 16.26–34.64 |

Average ± σ (° C) | 24.52 ± 5.08 | |

CV (%) | 20.70 | |

n | 2819 |

^{1}n, data number, σ, standard deviation.

**Table 6.**Regression model for the retrieval of Chl-a concentration using Landsat OLI training datasets.

No. | Landsat OLI Data Regression Model Equation for Chl-a Concentration Estimation | Band Combination (=x) | r | R^{2} |
---|---|---|---|---|

1 | y = −14.61 × x + 10.98 | B2/(B3 + B4) | −0.84 | 0.70 |

2 | y = −11.50 × x + 2.50 | (B2 − B3)/(B2 + B3) | −0.80 | 0.65 |

3 | y = −10.40 × x + 7.93 | B1/(B3 + B4) | −0.79 | 0.62 |

**Table 7.**Regression model for the retrieval of Chl-a concentration using Sentinel MSI training datasets.

No. | Sentinel MSI Data Regression Model Equation for Chl-a Concentration Estimation | Band Combination (=x) | r | R^{2} |
---|---|---|---|---|

1 | y = 1183.26 × x + 2.62 | B5 − (B4 + B6)/2 | 0.79 | 0.62 |

2 | y = 2.57 × e^{284.24×x} | 0.59 |

**Table 8.**Regression model for the retrieval of algal density concentration using Landsat OLI training datasets.

No. | Landsat OLI Data Regression Model Equation for Algal Density Estimation | Band Combination (=x) | r | R^{2} |
---|---|---|---|---|

1 | y = −19,789,532.32 × x + 8,686,780.32 | B1/(B2 + B3 + B4) | −0.91 | 0.82 |

2 | y = −12,502,445.57 × x + 1,628,888.71 | (B2 − B3)/(B2 + B3) | −0.90 | 0.81 |

3 | y = 3,292,817.47 × x − 1,084,109.44 | (B1 − B3)/(B1 + B3) | 0.88 | 0.77 |

**Table 9.**Regression model for the retrieval of algal density concentration using Sentinel MSI training datasets.

No. | Sentinel MSI Data Regression Model Equation for Algal Density Estimation | Band Combination (=x) | r | R^{2} |
---|---|---|---|---|

1 | y = 1,227,118,587.08 × x + 2,142,593.11 | B5 − (B4 + B6)/2 | 0.78 | 0.61 |

2 | y = 1,981,923.63 × e^{351.40×x} | 0.53 |

**Table 10.**Regression model for the retrieval of turbidity concentration using Landsat OLI training datasets.

No. | Landsat OLI Data Regression Model Equation for Turbidity Estimation | Band Combination (=x) | r | R^{2} |
---|---|---|---|---|

1 | y = −2.78 × x + 2.30 | (B2 − B4)/B3 | −0.84 | 0.80 |

2 | y = −7.08 × x + 4.93 | B2/(B1 + B3) | −0.78 | 0.71 |

3 | y = −3.10 × x + 2.24 | (B2 − B4)/(B2 + B4) | −0.73 | 0.61 |

**Table 11.**Regression model for the retrieval of turbidity concentration using Sentinel MSI training datasets.

No. | Sentinel MSI Data Regression Model Equation for Turbidity Estimation | Band Combination (=x) | r | R^{2} |
---|---|---|---|---|

1 | y = −10.42 × x + 2.76 | B1 | −0.37 | 0.14 |

2 | y = 3.44 × x + 0.46 | B3/(B1 + B2) | 0.35 | 0.12 |

Water Quality Parameter | Data Source | No. | Min | Max | Average | SD | CV(%) | SE |
---|---|---|---|---|---|---|---|---|

Chl-a concentration (μg/L) | Landsat OLI | 1 | 2.26 | 3.15 | 2.76 | 0.40 | 14.35 | 0.20 |

2 | 2.48 | 3.45 | 2.91 | 0.40 | 13.74 | 0.20 | ||

3 | 1.97 | 3.18 | 2.56 | 0.52 | 20.17 | 0.26 | ||

In situ data for L-O validation | / | 1.99 | 8.17 | 3.72 | 2.58 | 69.53 | 1.29 | |

Sentinel MSI | 1 | 2.86 | 7.23 | 4.15 | 1.30 | 31.38 | 0.65 | |

2 | 2.72 | 7.79 | 3.92 | 1.51 | 38.59 | 0.76 | ||

In situ data for S-M validation | / | 1.93 | 3.91 | 2.64 | 0.60 | 22.75 | 0.30 | |

Algae density | Landsat OLI | 1 | 1.57 × 10^{6} | 2.80 × 10^{6} | 2.15 × 10^{6} | 5.21 × 10^{5} | 24.24 | 2.60 × 10^{5} |

2 | 2.10 × 10^{6} | 3.15 × 10^{6} | 2.56 × 10^{6} | 4.35 × 10^{5} | 16.95 | 2.17 × 10^{5} | ||

3 | 1.94 × 10^{6} | 3.03 × 10^{6} | 2.41 × 10^{6} | 4.50 × 10^{5} | 18.68 | 2.25 × 10^{5} | ||

In situ data for L-O validation | / | 1.80 × 10^{6} | 6.54 × 10^{6} | 4.08 × 10^{6} | 2.17 × 10^{5} | 53.29 | 1.09 × 10^{5} | |

Sentinel MSI | 1 | 2.39 × 10^{6} | 6.93 × 10^{6} | 3.72 × 10^{6} | 1.35 × 10^{5} | 36.23 | 6.75 × 10^{5} | |

2 | 2.13 × 10^{6} | 7.80 × 10^{6} | 3.41 × 10^{6} | 1.70 × 10^{6} | 49.89 | 8.49 × 10^{5} | ||

In situ data for S-M validation | / | 1.59 × 10^{6} | 3.50 × 10^{6} | 2.54 × 10^{6} | 5.74 × 10^{5} | 22.56 | 2.87 × 10^{5} | |

Turbidity (NTU) | Landsat OLI | 1 | 1.28 | 1.52 | 1.39 | 0.09 | 6.12 | 0.04 |

2 | 1.47 | 1.61 | 1.53 | 0.06 | 3.76 | 0.03 | ||

3 | 1.44 | 1.60 | 1.51 | 0.06 | 3.99 | 0.03 | ||

In situ data for L-O validation | / | 2.34 | 6.44 | 4.69 | 1.52 | 32.37 | 0.76 | |

Sentinel MSI | 1 | 2.32 | 2.60 | 2.51 | 0.08 | 3.19 | 0.04 | |

2 | 1.96 | 2.53 | 2.25 | 0.18 | 8.05 | 0.09 | ||

In situ data for S-M validation | / | 4.48 | 8.02 | 5.84 | 1.19 | 20.37 | 0.59 |

**Table 13.**Comparison of evaluation indexes of WQPs regression formula (brought into training and testing dataset together).

WQPs | Data Type | Band Combination (=x) | Regression Model Equation | MAE | MSE | RMSE |
---|---|---|---|---|---|---|

Chl-a concentration | Landsat OLI | B2/(B3 + B4) | y = −14.61 × x + 10.98 | 1.19 | 2.75 | 1.66 |

Sentinel MSI | B5 − (B4 + B6)/2 | y = 2.57 × e ^{284.24×x} | 1.12 | 2.92 | 1.71 | |

Algal density | Landsat OLI | (B2 − B3)/(B2 + B3) | y = −12,502,445.57 × x + 1,628,888.71 | 9.19 × 10^{5} | 2.04 × 10^{12} | 1.43 × 10^{6} |

Sentinel MSI | B5 − (B4 + B6)/2 | y = 1,227,118,587.08 × x + 2,142,593.11 | 1.13 × 10^{6} | 2.77 × 10^{12} | 1.66 × 10^{6} | |

Turbidity | Landsat OLI | (B2 − B4)/B3 | y = −2.78 × x + 2.30 | 1.08 | 3.80 | 1.95 |

Sentinel MSI | B1 | y = −10.42 × x + 2.76 | 1.44 | 4.36 | 2.09 |

PC1 | PC 2 | PC 3 | PC 4 | |
---|---|---|---|---|

pH | 0.958 | −0.023 | −0.139 | −0.016 |

DO | 0.844 | 0.173 | −0.293 | 0.034 |

Chl-a concentration | 0.800 | −0.283 | 0.267 | 0.044 |

WT | 0.660 | −0.471 | 0.216 | −0.041 |

TN | 0.573 | 0.427 | 0.183 | −0.074 |

COD_{Mn} | 0.514 | 0.428 | 0.008 | 0.138 |

Conductivity | 0.478 | 0.572 | 0.375 | −0.039 |

Algal density | 0.370 | −0.483 | 0.358 | 0.131 |

TP | −0.365 | 0.452 | 0.322 | 0.187 |

Turbidity | −0.341 | −0.057 | 0.818 | 0.004 |

NH_{3}-N | −0.011 | −0.040 | −0.082 | 0.958 |

Variability (%) | 35.569 | 13.341 | 11.870 | 9.103 |

Cumulative (%) | 35.569 | 48.910 | 60.780 | 69.883 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Meng, H.; Zhang, J.; Zheng, Z. Retrieving Inland Reservoir Water Quality Parameters Using Landsat 8-9 OLI and Sentinel-2 MSI Sensors with Empirical Multivariate Regression. *Int. J. Environ. Res. Public Health* **2022**, *19*, 7725.
https://doi.org/10.3390/ijerph19137725

**AMA Style**

Meng H, Zhang J, Zheng Z. Retrieving Inland Reservoir Water Quality Parameters Using Landsat 8-9 OLI and Sentinel-2 MSI Sensors with Empirical Multivariate Regression. *International Journal of Environmental Research and Public Health*. 2022; 19(13):7725.
https://doi.org/10.3390/ijerph19137725

**Chicago/Turabian Style**

Meng, Haobin, Jing Zhang, and Zhen Zheng. 2022. "Retrieving Inland Reservoir Water Quality Parameters Using Landsat 8-9 OLI and Sentinel-2 MSI Sensors with Empirical Multivariate Regression" *International Journal of Environmental Research and Public Health* 19, no. 13: 7725.
https://doi.org/10.3390/ijerph19137725