CHLOROPHYLL CONCENTRATION IN A RESERVOIR FROM SATELLITE IMAGERY: CASE STUDY OF THE SALTO GRANDE DAM, BRAZIL

Human beings have a profound relationship with water, which is essential for life and serves a wide range of purposes. However, human interactions and activities have been degrading the quality of water bodies. Therefore, monitoring water bodies is crucial, as it enables the understanding of their spatial and temporal variability and supports decision-making and water resource management. Traditional monitoring methods, which rely on sample collection and laboratory analyses, can be costly and provide only point-based data, failing to represent the characteristics of the entire water body. Remote sensing techniques have emerged as a complementary tool to traditional monitoring, offering relatively lower costs and shorter execution times. This study aimed to use remote sensing techniques to assess the trophic state of the Salto Grande Reservoir, located in the municipality of Americana, São Paulo State, Brazil, using chlorophyll a (Chl-a) concentration as the analysis parameter. The reservoir is situated in a highly industrialized region with significant monoculture agriculture and has a history of environmental degradation. Imagery from the MSI sensor onboard the Sentinel-2 satellite was used to estimate Chl-a concentration. The images were atmospherically corrected employing the Sen2Cor software. Four algorithms based on spectral band ratios were tested to estimate Chl-a concentrations. The results were compared with in situ measurements taken on nearby dates and made publicly available. The adjusted algorithms showed coefficients of determination (R 2 ) above 0.8 and error rates below 10% and 8 µg L −1 . The NDCI algorithm by Mishra and Mishra (2012) showed the best applicability, as it returned only positive values, unlike some other algorithms that produced negative Chl-a values. The modified Carlson’s trophic state index was applied using the values from the selected algorithm, revealing a high degree of eutrophication in the reservoir. 1. INTRODUCTION Water, especially surface water, is extremely important for biodiversity, as it plays a key role in biogeochemical cycles and is essential for food and energy production (Tyler et al., 2016). In terms of water resources, Brazil holds a privileged position among countries, having almost 12% of the planet’s available freshwater. However, this resource is not evenly distributed throughout the country, being largely concentrated in sparsely populated regions (Novo, 2019; ANA, 2020). The quality of these resources is also a factor to be considered, as many water bodies have been impacted by anthropogenic activities, which can lead to water quality degradation and biodiversity loss, consequently compromising ecosystem services and functions (Dalu et al., 2015). Risks to water resources are directly associated with land use, which drives degradation not only in urban but also in agricultural areas. In urban areas, degradation is primarily due to the lack of sanitation, proper waste collection, and wastewater treatment, resulting in the discharge of untreated waste and effluents into water bodies. In agricultural zones, it is mainly due to the use of fertilizers and pesticides, as well as increased sediment transport (Cruz et al., 2019; Mello et al., 2020). The transport of sediments and nutrients from the soil to water bodies increases nutrient levels in aquatic systems—such as phosphorus and nitrogen—leading to various functional changes and potentially triggering eutrophication. Nutrient enrichment alters the development of certain biological communities and promotes the proliferation of harmful algae, with consequent toxin release into the water (Watanabe et al., 2019), thereby harming environmental biodiversity (Ullah et al., 2025). Another factor influencing the characteristics of Brazil’s water resources is its energy matrix, which is predominantly based on hydropower. This leads to a large number of artificial reservoirs along Brazilian rivers, with volumes that vary according to hydrological, economic, and climatic factors (Novo, 2019). The eutrophication process in reservoirs tends to worsen, as increased phosphorus and nitrogen levels combined with longer water retention times create favorable conditions for the development of cyanobacteria, a phytoplankton group known for its toxicity (Watanabe et al., 2019). Understanding the temporal dynamics of rivers is essential to support water resource management and decision-making aimed at mitigating anthropogenic impacts on watercourses. However, the information available on water quality is not evenly distributed due to regional differences, location, accessibility, and resource availability, and may be scarce in some cases (Mercan, 2025). Water quality is generally assessed through measurements obtained from collected samples, which allow for quantifying the physical, chemical, and biological properties of the water at a specific location and time (Novo, 2019). In situ measurements are more accurate and widespread than indirect methods such as remote sensing (Gaida et al., 2020) but require high investments of time and money and have limited spatial coverage, which becomes a limiting factor for water quality monitoring (Buma and Lee, 2020). Spatiotemporal analyses of water quality parameters are essential for effective management and can be enabled through methodologies developed using remote sensing techniques (Gholizadeh et al., 2016; Rahman et al., 2025). Remote sensing approaches facilitate a better understanding of the physicochemical changes in water bodies, thus helping to identify disturbances occurring in aquatic systems (Sent et al., 2021). Chlorophyll a (Chl-a) stands out among the water quality parameters that can be assessed through remote sensing. This pigment is found in photosynthetic organisms such as phytoplankton, which includes hundreds of microalgae and cyanobacteria species, and is responsible for the apparent color of water, typically present in aquatic ecosystems such as lakes and reservoirs (Barbosa, 2019; Radin et al., 2020). Chl-a is an important parameter for assessing aquatic environments due to its relationship with trophic status, water clarity, and algal biomass (Matsushita et al., 2015). It also represents a biological response linked to nutrient enrichment and, consequently, the increase of phytoplanktonic organisms that contain this pigment (Gholizadeh et al., 2016; Qu et al., 2023). High Chl-a concentrations indicate shifts in trophic state (eutrophication level) and are associated with reduced water quality and decreased biodiversity, which destabilize ecosystem services and functions (Dalu et al., 2015; Moiseenko et al., 2024). Understanding the dynamics of Chl-a concentrations is essential for selecting appropriate management strategies, which can aid in the recovery of ecosystem functions and services. Therefore, frequent monitoring becomes fundamental to grasp such dynamics (Dalu et al., 2015). Missailidis et al. (2018) studied phosphorus accumulation in the sediments of the Salto Grande reservoir and observed a progressive eutrophication process, indicating that the reservoir has been receiving polluted water inputs for decades, a situation that has worsened with population growth. There is also a contribution from soil leaching around the reservoir, driven by the use of fertilizers. Tourism-related activities intensified in the reservoir area between the 1970s and 1980s, leading to increased civil construction along its banks and, consequently, the discharge of untreated domestic effluents into the reservoir (Martins et al., 2011). This study aimed to assess the trophic state of the Salto Grande reservoir using satellite-derived chlorophyll a concentrations as an indicator parameter and in situ data collected by the environmental monitoring agency of the state of São Paulo (CETESB) as a reference. 2. MATERIAL AND METHODS 2.1 Study area The Salto Grande reservoir is located in the eastern region of the state of São Paulo and covers an area of 11.5 km 2 (Martins et al., 2011). The reservoir is part of the Americana Hydroelectric Power Plant complex, managed by Companhia Paulista de Força e Luz (CPFL). The region where the reservoir is located (Figure 1) is predominantly characterized by urban infrastructure and large-scale monoculture of sugarcane, citrus, and pastureland (Neto, 2013). Over time, the reservoir has been experiencing increasing environmental degradation and a high degree of eutrophication, which has raised concerns among both authorities and the local population. Figure 1 – Location and land use in the surroundings of the Salto Grande reservoir. The increasing anthropogenic activity throughout the Atibaia River basin, whose outlet is the Salto Grande reservoir, has resulted in nutrient enrichment in the reservoir due to the significant discharge of industrial and domestic effluents, as well as surface runoff carrying agricultural inputs and soil via rainwater (Martins et al., 2011). According to CETESB (2020), the reservoir is organically enriched, with high phosphorus concentrations in the sediments. 2.2 Field data CETESB conducts monitoring of various water bodies within the state of São Paulo and makes the analytical results and other data from its monitoring stations publicly available through the InfoÁgua information system and annually published water quality reports. A monitoring station (Figure 2), identified as ATSG 02800, is located in the central body of the reservoir. The Chl-a concentrations used in this study were obtained via the CETESB InfoÁgua platform (https://sistemainfoaguas.cetesb.sp.gov.br/), which provides sample analysis results from this and other monitoring stations across the state of São Paulo. Figure 2 – Location of the water quality sampling station in the Salto Grande reservoir. Monitoring started in 2017 at this station, with approximately four analyses carried out per year. However, only two samplings were conducted in 2020 due to the Sars-CoV-2 (Coronavirus) pandemic, with normal monitoring activities resuming in 2021. 2.3 Image acquisition and atmospheric correction The estimation of Chl-a concentrations in the reservoir was performed using adjusted algorithms based on images from the Sentinel-2 satellite. The results obtained from these algorithms were compared with in situ measurements by CETESB, which served as the reference data. Images were selected from dates close to CETESB’s sampling campaigns, within a three-day window before and after the sampling date. Preference was given to images with the least cloud interference over or near the study area (Table 1). Table 1 – CETESB sampling dates in the Salto Grande reservoir and corresponding chlorophyll a values. Tabela 1: Dados de amostragem da CETESB e respectivas datas de imageamento. | 22 Feb. 2018 | 12:58 | 20.58 | Yes | 24 Feb. 2018 | | 16 Aug. 2018 | 16:28 | 27 | No | 18 Aug. 2018 | | 06 Dec. 2018 | 12:02 | 12.43 | No | 06 Dec. 2018 | | 21 Feb. 2019 | 13:25 | 27.45 | Yes | 24 Feb. 2019 | | 26 June 2019 | 12:44 | 11.49 | No | 24 June 2019 | | 29 Aug. 2019 | 11:51 | 123.63 | No | 28 Aug. 2019 | | 12 Feb. 2020 | 11:17 | 33.86 | Yes | 14 Feb. 2020 | | 08 Feb. 2021 | 10:34 | 36.98 | No | 08 Feb. 2021 | The satellite images used in this study were captured by the twin Sentinel-2A and 2B satellites, which are equipped with the MSI (Multispectral Imager) instrument. The Sentinel mission was developed by the European Space Agency (ESA) and is part of the Copernicus Program, which comprises several pairs of satellites. The images were obtained from the Earth Explorer database of the United States Geological Survey (USGS), with a spatial resolution of 20 m. They are provided as Level-1C products, meaning that pixel-based radiometric measurements are given in top of atmosphere (TOA) reflectance, with UTM/WGS84 projection (ESA, 2020). The images underwent atmospheric correction using the Sen2Cor algorithm, available in the SNAP (Sentinel Application Platform) software, using the default configuration. This correction process converts Level-1C images into Level-2A products, which provide bottom of atmosphere (BOA) reflectance, also referred to as surface reflectance. Surface reflectance was divided by π to convert the data into remote sensing reflectance (Rrs), which is the input data used in the selected algorithms. 2.4 Estimation of chlorophyll a concentration Four algorithms proposed by other authors were applied to estimate chlorophyll a (Chl-a) concentration. These algorithms use spectral band ratios, with input data based on the remote sensing reflectance of water (Rrs(λi)). The applied algorithms consisted of (i) the two-band algorithm, proposed by Dall’Olmo and Gitelson (2025); (ii) the three-band algorithm, proposed by Dall’Olmo and Gitelson (2005); (iii) the NDCI algorithm, proposed by Mishra and Mishra (2012); and (iv) the SLOPE algorithm, proposed by Mishra and Mishra (2010). The algorithms were designated as DG2B, DG3B, NDCI, and SLOPE, respectively, for identification purposes. The DG3B algorithm was originally developed to estimate pigment content in terrestrial vegetation. However, Dall’Olmo and Gitelson (2005) observed that it could also be used to assess Chl-a concentration in complex waters: Chl-a ∝ [Rrs-1 (λ1) − Rrs-1 (λ2)] * Rrs (λ3) (1) As an alternative to the three-band model, Dall’Olmo and Gitelson (2005) also proposed a two-band model (DG2B), which uses the ratio between the red and near-infrared (NIR) bands: Chl-a ∝ Rrs (λ2) / Rrs (λ1) (2) The NDCI (Normalized Difference Chlorophyll Index) is an algorithm proposed by Mishra and Mishra (2012), which also uses the ratio between red and NIR bands to avoid interference in the reflectance spectra of water at shorter wavelengths: Chl-a ∝ [Rrs(λ2) − Rrs(λ1)] / [Rrs(λ2) + Rrs(λ1)] (3) The fourth algorithm tested, named SLOPE and proposed by Mishra and Mishra (2010), is based on the relationship between a scattering-sensitive band and the band with maximum chlorophyll absorption: Chl-a ∝ [Rrs(λ2) − Rrs(λ1)] / (λ2 − λ1) (4) Remote sensing reflectance values were used for bands corresponding to wavelengths near 700 nm and 670 nm, as Rrs(λ2) and Rrs(λ1), respectively. These wavelengths have been widely used to estimate chlorophyll concentrations because the reflectance peaks in these regions are maximally sensitive to variations in Chl-a concentrations in water and the maximum Chl-a pigment absorption, respectively (Matthews, 2011; Mishra and Mishra, 2012). In addition, the DG3B algorithm incorporates a third wavelength, near 750 nm (Rrs(λ3)), since reflectance values at this wavelength are less affected by absorption (Gitelson et al., 2008). Considering the algorithms and respective wavelengths proposed by the authors and comparing these values with the bands imaged by the MSI sensor, the images corresponding to Bands 4 (Red), 5 (Red Edge), and 6 (Red Edge) were used as Rrs (λ1), Rrs (λ2), and Rrs (λ3), respectively. After applying the four algorithms to the satellite images, the resulting values were extracted from the pixel corresponding to the exact coordinates of the CETESB sampling point and surrounding pixels, forming a 3 x 3 grid, totaling nine pixels, and the mean of these values was then calculated. Regressions were fitted to compare the mean values calculated from each algorithm with the reference values measured by CETESB. Then, five statistical parameters were calculated to evaluate the fitting: coefficient of determination (R 2 ), root mean squared error (RMSE), normalized root mean squared error (NRMSE), mean absolute percentage error (MAPE), and bias. The function that showed the highest correlation along with the lowest percentage error (MAPE) was selected and applied to the remaining images, which together covered the entire reservoir area. Areas with cloud cover or macrophyte coverage were removed to avoid misinterpretation of the results. For this procedure, a mask was created to delineate these areas, which were then excluded from the total reservoir area. 3. RESULTS AND DISCUSSION 3.1 Field-collected Chl-a data The results of the sample analyses conducted by CETESB regarding Chl-a concentrations at the sampling point in the Salto Grande reservoir (Figure 3) showed a significant variability in Chl-a concentrations over the analyzed historical series. These concentrations ranged from 10 to 130 µg L −1 . Figure 3 – Variation in chlorophyll a concentration measured at the sampling point over time. The highest Chl-a concentrations were recorded in 2017, with values exceeding 60 µg L −1, whereas the following period, between 2018 and the first half of 2019, showed the lowest Chl-a values. June and August of 2019 had the most significant differences in Chl-a concentrations in sequential samples. The Chl-a concentration in June was 11.49 µg L −1, the lowest value recorded during the entire period. Furthermore, the Chl-a concentration increased significantly in August, reaching 123.63 µg L −1 . This period coincides with the winter school holidays in July, which may have influenced the presence of residents or vacationers staying in houses near the reservoir. Another important factor is that this period also corresponds to the dry season in the region, which may have contributed to the increase in Chl-a concentration due to a combination of factors such as reduced water volume and greater nutrient availability, thereby promoting the growth of algae and cyanobacteria (Moura et al., 2017). Additionally, both quantitative and qualitative variations are influenced by multiple factors, including the type of ecosystem (Zakeyuddin et al., 2016), the dynamics of aquatic systems, and the concentrations and interactions of their components, which are affected by biotic and abiotic factors. The variability of water quality was assessed by compiling remote sensing reflectance (Rrs) values from the eight selected images for the CETESB sampling point (Figure 4) at different wavelengths, according to the spectral bands of the Sentinel-2 MSI sensor. Figure 4 – Reflectance of Sentinel-2 imagery at the sampling point on different dates. Figure 4 shows evidence of the presence of Chl-a due to the high absorption in the blue and red regions, the high reflectance in the green region (~550 nm), and, mainly, the reflectance peak near 700 nm, which is characteristic of this pigment. However, some of these peaks are influenced by other components present in the water, such as those near 560 nm and 700 nm, in which contributions from solids suspended in the water may occur (Santos et al., 2019). The reflectance spectrum for 28 August 2019 shows well-defined peaks in the visible range and an increase in reflectance beyond 700 nm, being the only spectrum with higher reflectance in this region than in the visible range. These peaks may indicate high concentrations of suspended solids and elevated phytoplankton biomass. Data on turbidity, total solids, and total dissolved solids were compiled from the InfoÁguas platform (Table 2) to investigate the presence of suspended solids in the reservoir on the sampling dates, corresponding to the satellite image dates. The suspended solids values were calculated by subtracting total dissolved solids from total solids. Table 2 – Chlorophyll a (Chl-a), total solids, total dissolved solids, suspended solids, and turbidity at the sampling point. | 22/02/18 | 24/02/18 | 20.58 | 132 | 128 | 4 | 6.6 | | 07/06/18 | 09/06/18 | 43.44 | 218 | 208 | 10 | 5.7 | | 16/08/18 | 18/08/18 | 27 | 202 | 196 | 6 | 4.5 | | 06/12/18 | 06/12/18 | 12.43 | 128 | 126 | 2 | 15 | | 21/02/19 | 24/02/19 | 27.45 | 162 | 148 | 14 | 4.34 | | 26/06/19 | 24/06/19 | 11.49 | 164 | 148 | 16 | 2.33 | | 29/08/19 | 28/08/19 | 123.63 | 208 | 188 | 20 | 34.2 | | 12/02/20 | 14/02/20 | 33.86 | 192 | 184 | 8 | 11.2 | | 08/02/21 | 08/02/21 | 36.98 | 112 | 100 | 12 | 9.54 | The sampling date of 29 August 2019, corresponding to the 28 August 2019 image, shows not only the highest Chl-a value but also the highest suspended solids content and turbidity among all samples. Turbidity is directly proportional to the presence of suspended solids, as it is a measure of the difficulty for a light beam to penetrate the water due to suspended materials. 3.2 Algorithm application The coefficient of determination (R 2 ) was calculated between the values estimated by the equations and the sample values measured from the application of the four algorithms used and the fitting of several linear and nonlinear trend lines. The regression models that showed the highest correlations and lowest error rates for each algorithm were selected and are shown in Figure 5. Figure 5 – Best-fit regression equations, coefficient of determination (R 2 ), and errors between measured and estimated values for the four applied algorithms. Most of the data points are concentrated in the range from 20 to 40 mg m −3, with only one data point showing a high Chl-a value above 120 mg m −3 and two data points below 20 mg m −3 . Additionally, the trend lines with the highest R 2 values were the second-degree polynomial and linear regressions for all algorithms. The correlation and error values for both linear and second-degree polynomial regressions were very similar in the DG2B algorithm. However, the linear regression was selected as the most appropriate model because it exhibited the lowest MAPE and the smallest bias. The SLOPE algorithm had the highest R 2 and the lowest errors, except for the MAPE values, for which the linear regression of the NDCI algorithm yielded the lowest error. The other algorithms produced similar results, with R 2 values around 0.95, RMSE close to 7 mg m −3, NRMSE around 6%, and MAPE ranging between 9% and 11%. The second-best algorithm was DG2B, as it produced better R 2, RMSE, and NRMSE values than the others, followed by NDCI. The DG3B algorithm ranked lowest, with the highest error values and the lowest R 2 . Similar results have been reported in the literature. For example, Neil et al. (2019) evaluated the performance of 48 Chl-a estimation algorithms with different structural models, using datasets collected from 185 inland and coastal water systems, encompassing 13 optically diverse water types. Four algorithms stood out among them as the most appropriate and accurate, including DG2B and NDCI. Cairo et al. (2019) studied the Ibitinga reservoir in Brazil and classified Chl-a concentrations into ranges and tested several algorithms for each interval. They found that the SLOPE algorithm was the most suitable for waters with Chl-a concentrations between 19.51 and 87.63 mg m −3, a range that also includes most of the data used in the present study, supporting the observed results. However, applying the SLOPE algorithm with a second-degree polynomial fit resulted in negative SLOPE values for low Chl-a concentrations, a pattern also observed in the linear fit. Similarly, it was found for the DG2B algorithm when using a second-degree polynomial regression. Watanabe et al. (2019) investigated the trophic gradient in three cascade reservoirs along the Tietê River, Brazil, and described a similar behavior. As a result, they chose an algorithm that did not yield negative values, even if it had less favorable correlation and error metrics, ultimately selecting the NDCI algorithm with a linear fit. Likewise, an algorithm that did not generate negative values was selected in the present study. Thus, the NDCI algorithm with a second-degree polynomial fit was used. Although it had the third-best performance overall, it did not associate negative values with low Chl-a concentrations. According to Mishra and Mishra (2012), one of the advantages of NDCI is that its output range varies between −1 and +1, making qualitative Chl-a mapping and vegetation bloom detection via remote sensing feasible even in remote areas where field data may be unavailable or unusable. Several other algorithms can be used to estimate Chl-a, as demonstrated by Barraza-Moraga et al. (2022), Barreneche et al. (2023), and Chusnah et al. (2023), who reported coefficients of determination similar to those found in this study and classified the method as robust. Therefore, the algorithms are appropriate for estimating Chl-a concentrations in the Salto Grande reservoir. Importantly, although the correlation coefficients obtained in this study were high, sometimes even higher than those reported by other authors, we aimed to use publicly available data, without the collection of new in situ measurements. This resulted in a smaller number of samples compared to other studies. Certain Chl-a concentration ranges were not considered due to the limited dataset, which may lead to estimation errors. Also, the waterbody has only one monitoring point, and its characteristics are assumed to be representative of the entire reservoir, which may not be accurate, given that the trophic status likely varies spatially throughout the reservoir depending on the location and site-specific characteristics. Therefore, new data collection is recommended in future research to improve the accuracy of results under different conditions. However, no in situ sampling was conducted because one of the objectives of this study was to rely exclusively on published monitoring data. 3.3 Spatiotemporal distribution of Chl-a The NDCI algorithm, fitted using a second-degree polynomial function, was applied to all satellite images, resulting in spatially distributed Chl-a concentrations (Figure 6), expressed in mg m −3 . Figure 6 – Spatial distribution of chlorophyll-a concentrations estimated using the NDCI algorithm. Chl-a concentrations ranged from values below 25 mg m −3 to 150 mg m −3, with a maximum value of 152.3 mg m −3 and a minimum of 8.80 mg m −3 . In general, the lowest concentrations were found near the inflow areas of other watercourses. Chl-a concentrations tend to increase along the reservoir, with the highest concentrations detected near the dam wall and along the banks. This pattern was expected, given that the dam structure restricts water flow, causing nutrients and particles to accumulate in this area. Areas affected by cloud cover and macrophyte presence were removed and represented by masks. These plants interfere with the results because the reflectance data corresponds to the vegetation and not to the water surface. However, the presence of these plant communities is also considered an indicator of high nutrient concentrations in the environment, which favor their growth. Importantly, these communities were located near the dam wall in most of the satellite images. Macrophyte presence was also detected near the sampling point on 24 February 2019 and 24 June 2019. Moreover, elevated Chl-a concentrations were observed on 28 August 2019, along the reservoir’s banks, particularly in the vicinity of the sampling location. This location exhibits a high density of buildings, predominantly residential, and is where the local Yacht Club is situated. The rate of human presence and occupancy in this area may contribute to increased domestic sewage discharge and waste inputs, which influence nutrient loading and the eutrophication process in this region. The trophic state of aquatic systems can also be assessed using trophic state indices (TSI) (Novo et al., 2013). The TSI of the area (Figure 7) can be estimated based on chlorophyll concentrations. For this, the modified Carlson’s trophic state index, the same index adopted by CETESB, was used. This index is based on three indicators: Secchi disk depth, total phosphorus, and Chl-a. In this study, only Chl-a values were used for classification. The TSI includes six categories, classified according to Chl-a concentration as ultraoligotrophic (Chl-a ≤ 1.17 mg m −3 ), oligotrophic (1.17 < Chl-a ≤ 3.24 mg m −3 ), mesotrophic (3.24 < Chl-a ≤ 11.03 mg m −3 ), eutrophic (11.03 < Chl-a ≤ 30.55 mg m −3 ), supereutrophic (30.55 < Chl-a ≤ 69.05 mg m −3 ), and hypereutrophic (Chl-a ≥ 69.05 mg m −3 ). Most of the reservoir area, excluding regions covered by clouds or macrophytes, falls into the eutrophic category or higher. The lowest estimated value was 8.8 mg m −3 . Therefore, the lowest trophic category identified in the reservoir was mesotrophic, according to CETESB’s classification. Figure 7 – Trophic state index of the Salto Grande reservoir estimated from Chl-a values. The mesotrophic category shows the lowest occurrence, occupying an average of 0.41% of the area, with no areas falling under this category in the images from 24 February 2018 and 28 August 2019. In contrast, the eutrophic category has the highest occurrence, covering approximately 63% of the area. In general, areas classified as hypereutrophic are found near the banks and the dam. The image from 28 August 2019 exhibits the largest area classified as hypereutrophic, with the remainder of the reservoir mostly categorized as supereutrophic. Data from the region closest to the dam structure were not used due to the presence of macrophytes. However, this area may exhibit high nutrient concentrations and, consequently, high eutrophication levels due to the retention of water and associated compounds, which causes nutrients and other substances to accumulate near this location. The upper part of the reservoir is classified as supereutrophic in the image from 24 June 2019, while the second region is classified as eutrophic, the opposite pattern observed in the other images, where higher Chl-a concentrations are typically found closer to the dam. Figure 8 shows the percentage of areas occupied by each trophic class, excluding areas with macrophytes or cloud cover. Figure 8 – Percentage of area occupied by each of the four trophic state index (TSI) classes in the Salto Grande reservoir. Most of the reservoir’s surface area is classified as eutrophic or higher on all analyzed dates, with the supereutrophic or higher classification accounting for over 50% of the total reservoir area on 24 June 2019, 28 August 2019, and 14 February 2020. It indicates that the majority of the reservoir is classified as eutrophic, supereutrophic, or hypereutrophic. All these analyses highlight the concerning condition of this reservoir, which exhibits a high level of eutrophication, affecting the biogeochemical cycles of the environment and potentially leading to biodiversity loss and adverse impacts on surrounding communities and people who come into contact with the water or consume organisms from this environment. 4. CONCLUSION MSI sensor imagery acquired by the Sentinel-2 satellites and the data contained within allowed the estimation of Chl-a through the application of algorithms. Importantly, these images possess essential characteristics for such applications, including high temporal resolution, a greater number of spectral bands, and good spatial resolution. The employed algorithms were effective in estimating Chl-a in the Salto Grande reservoir, enabling the analysis of the distribution of its concentration in space and time. Linear and second-degree polynomial fits presented the best performance among all tested regression equations, comparing field-measured values with those estimated by the algorithms. The Normalized Difference Chlorophyll Index (NDCI) algorithm was the most applicable and efficient for this study area, primarily due to its characteristic range between −1 and +1, which ensured all estimated Chl-a values were positive, an advantage already noted in previous studies employing this algorithm. This algorithm, when applied to the second-degree polynomial fit, achieved a coefficient of determination of 0.95, with error indices below 10%, with error values of 7.36 mg m −3 and 6.6% to 9.4%. The study demonstrated the potential of satellite imagery to monitor the Salto Grande reservoir. Nevertheless, further research is recommended, involving the collection of new in situ data and the establishment of new sampling points throughout the reservoir to better calibrate and validate the algorithms and enhance accuracy across varying environmental conditions. One major source of interference in Chl-a estimation was the presence of aquatic macrophyte communities in the region, which occupy a significant portion of the water surface. However, despite interfering with the results and algorithm applicability, the presence of macrophytes can also be considered an indicator of the trophic state of the water, pointing to high nutrient concentrations. The application of algorithms, satellite imagery, and data provided by CETESB allowed us to conclude that the Salto Grande reservoir is in a critical eutrophication state. Implementing continuous monitoring, enhancing regulatory enforcement, and adopting public policies focused on sanitation and water resource management are indicated to mitigate this condition. In addition, further research and the application of environmentally appropriate management techniques would reduce pollution and ecological damage in the reservoir. DECLARATION OF COMPETING INTEREST The authors declare that they have no conflict of interest. ETHICAL STATEMENT The authors state that the research was conducted according to ethical standards. FUNDING BODY The Coordination for the Improvement of Higher Education Personnel – CAPES, Brazil. REFERENCIAS ANA - AGÊNCIA NACIONAL DE ÁGUAS E SANEAMENTO BÁSICO. Panorama das águas: quantidade de água . Brasília, 2020. Disponível em: . Acesso em: 10 de set de 2020. Barbosa, C. C. F. Princípios Físicos do Sensoriamento Remoto Aquático. 2019. In: Barbosa, C.C.F.; Novo, E.M.L.M.; Martins, V.S . Introdução ao Sensoriamento Remoto de Sistemas Aquáticos: princípios e aplicações. 1ª edição. Instituto Nacional de Pesquisas Espaciais. São José dos Campos. 161p. 2019. p. 23-54. Barraza-Moraga, F., H. Alcayaga, A. Pizzaro, et al. 2022. Estimation of chlorophyll-a concentrations in Lanalhue Lake using Sentinel-2 MSI satellite images. Remote Sensing 14, no. 22, 5647. Barreneche, J. M., B. Guigou, F. Gallego, et al. 2023. Monitoring Uruguay’s freshwaters from space: An assessment of different satellite image processing schemes for chlorophyll-a estimation. Remote Sensing Applications: Society and Environment 29, 100891. Bouvy, M.; S. M. Nascimento, R. J. Molica, et al. 2003 Limnological features in Tapacurá reservoir (northeast Brazil) during a severe drought . Hydrobiologia 493, no. 1, 115-130. Buma, W. G. and S. Lee. Evaluation of sentinel-2 and landsat 8 images for estimating chlorophyll-a concentrations in lake Chad, Africa. 2020. Remote Sensing 12, no. 15, 2437. Cairo, C.; C. Barbosa, F. Lobo, et al. 2019. Hybrid chlorophyll-a algorithm for assessing trophic states of a tropical brazilian reservoir based on msi/sentinel-2 data. Remote Sensing 12, no. 1, 40. Chusnah, W. N., H. J. Chu, T. Jaelani and L. M. Jaelani. 2023. Machine-learning-estimation of high-spatiotemporal-resolution chlorophyll-a concentration using multi-satellite imagery. Sustainable Environment Research 33, no. 1, 11. COMPANHIA AMBIENTAL DO ESTADO DE SÃO PAULO - CETESB. 2020. Relatório de Qualidade das Águas Interiores do Estado de São Paulo 2019. 336 p. São Paulo. Cruz, M. A. S., A. A. Gonçalves, R. Aragão, R., et al. 2019. Spatial and seasonal variability of the water quality characteristics of a river in Northeast Brazil. Environmental Earth Sciences 78, no. 3, 1–11. Dall’Olmo, G., A. A. Gitelson. 2005. Effect of bio-optical parameter variability on the remote estimation of chlorophyll-a concentration in turbid productive waters: experimental results. Applied Optics 44, no. 3, 412-422. Dalu, T., T. Dube, P. W. Froneman, et al. 2015. An assessment of chlorophyll-a concentration spatio-temporal variation using Landsat satellite data, in a small tropical reservoir. Geocarto International 30, no. 10, 1130-1143. Gaida, W.; F. M.Breunig; L. S. Galvão, F. J. Ponzoni, 2020. Correção Atmosférica em Sensoriamento Remoto: Uma Revisão. Revista Brasileira de Geografia Física 13, no. 1, 229-248. Gholizadeh, M. H., A. M. Melesse, L. Reddi. 2016. A comprehensive review on water quality parameters estimation using remote sensing techniques. Sensors 16, no. 8, 1298. Gitelson, A. A., G. Dall’Olmo, W. Moses, et al. 2008. A simple semi-analytical model for remote estimation of chlorophyll-a in turbid waters: Validation. Remote Sensing of Environment 112, no. 9, 3582–3593. Martins, D., S. R. Marchi, N. V. Costa, et al. 2011. Aquatic plant survey in Salto grande reservoir in Americana-SP, Brazil. Planta Daninha 29, no. 1, 231–236. Matsushita, B., W. Yang, G. Yu, et al. 2015. A hybrid algorithm for estimating the chlorophyll-a concentration across different trophic states in Asian inland waters. Journal of Photogrammetry and Remote Sensing 102, 28-37. Mattheuws, M. W. 2011. A current review of empirical procedures of remote sensing in inland and near-coastal transitional waters. International Journal of Remote Sensing 32, no. 21, 6855-6899. Mello, K., R. H. Taniwaki, F. R. Paula, et al. 2020. Multiscale land use impacts on water quality: Assessment, planning, and future perspectives in Brazil. Journal of Environmental Management 270, no. 6, 110879. Mercan, D. 2025. Approaches to pollution in lake ecosystems, the library of life: Na example from Uzungöl (Trabzon, Türkiye). Ecohydrology 18, 2, e70014. Misailidis, M. L., N. M. Strikis, R. C. Figueira, et al. 2018. Anthropogenic factors driving phosphorus contents in Salto Grande reservoir sediments, São Paulo state (SE Brazil). Journal of Sedimentary Environments 3, no. 3, 166-175. Mishra, D.R., S. Mishra. 2010. Plume and bloom: effect of the Mississippi River diversion on the water quality of Lake Pontchartrain. Geocarto International 25, no. 7, 555-568. Mishra, S., D. R. Mishra. 2012. Normalized difference chlorophyll index: A novel model for remote estimation of chlorophyll-a concentration in turbid productive waters. Remote Sensing of Environment 117, 394–406. Moiseenko, T. I., M. M. Bazova. 2024. Eutrophication of lakes: From global process to regional implecation in the Kola Artic Region. Ecohydrology 17, 8, e2713. Moura, M. E. P. D., L. D. S. Rocha, J. C. Nabout. 2017. Effects of global climate change on chlorophyll-a concentrations in a tropical aquatic system during a cyanobacterial bloom: a microcosm study. Revista Ambiente & Água 12, no.3, 390-404. Neil, C., E. Spyrakos, P. D. Hunter, A. N. Tyler. 2019. A global approach for chlorophyll-a retrieval across optically complex inland waters based on optical water types. Remote Sensing of Environment 229, 59-178. Neto, R. M. 2013. Zoogeografia em ambientes de represas: estudo na área sob influência do reservatório de Salto Grande–Rio Atibaia, Depressão Periférica Paulista. Revista de Geografia-PPGEO-UFJF 3, no. 2, 1-10. Novo, E. M. L. D. M., L. D. R. Londe, C. Barbosa, et al. 2013. Proposal for a remote sensing trophic state index based upon Thematic Mapper/Landsat images. Revista Ambiente & Água 8, 65-82. Novo, E. M. L. M. Sistemas Aquáticos Continentais: Definição e Características. 2019. In: Barbosa, C.C.F.; Novo, E.M.L.M.; Martins, V.S. Introdução ao Sensoriamento Remoto de Sistemas Aquáticos: princípios e aplicações. 1ª edição. Instituto Nacional de Pesquisas Espaciais. São José dos Campos. 161p. p. 9 – 22. Qu, S., J. Wang, A. Kumar, et al. 2023. Identification of driving factors for chlorophylla in multi-stable shallow lakes of China employing machine learning methods. Ecohydrology 16, 8, e2590. Radin, C., X. Soria-Perpiñà, J. Delegido. 2020. Estudio multitemporal de calidad del agua del embalse de Sitjar (Castelló, España) utilizando imágenes Sentinel-2. Revista Española de Teledetección 56, 117-130. Rahman, S. U., Y. Chey, Y. Su, et al. 2025. PGIS as a tool for reservoir health assessment: Community insights validates bt laboratory analysis and remote sensing. Ecohydrology 18, 2, e70005. Santos, L. D. S., H. V. Domingos, R. C. Lins, et al. 2019. Avaliação de modelos semi-empíricos para estimativa da concentração de clorofila-a baseado em bandas simuladas de satélites em um sistema estuarino lagunar. Aqua 1, no. 500, 469-555. Sent, G., B. Biguino, L. Favareto, et al. 2021. Deriving water quality parameters using Sentinel-2 imagery: A case study in the Sado estuary, Portugal. Remote Sensing 13, no. 5, 1043. Tyler, A. N., P. D. Hunter, E. Spyrakos, et al. 2016. Developments in Earth observation for the assessment and monitoring of inland, transitional, coastal and shelf-sea waters. Science of the Total Environment 572, 1307-1321. Ullah, N., F. Hussain, A. Nawaz, et al. 2025. Integrated ecohydrology and systems biology: Algae morphotypes and their role in freshwater ecosystem functioning. Ecohydrology 18, 3, e70027. Watanabe, F.; E. Alcântara, N. Bernardo et al. 2019. Mapping the chlorophyll-a horizontal gradient in a cascading reservoirs system using MSI Sentinel-2A images. Advances in Space Research 64, no. 3, 581-590. Zakeyuddin, M. S., A. S. M. Shah, M. S. Mohammad, et al. 2016. Spatial and temporal variations of water quality and trophic status in Bukit Merah Reservoir, Perak. Sains Malaysiana 45, no. 6, 853-863. Information & Authors Information Version history Copyright This work is licensed under a Non Exclusive No Reuse License.

Authors Metrics & Citations Metrics Article Usage 220views 192downloads Citations Download citation Vivian Fernanda Wiesenhutter Rodrigues, Rafael Kenji Koike Shimabukuro, Ronalton Evandro Machado, et al. CHLOROPHYLL CONCENTRATION IN A RESERVOIR FROM SATELLITE IMAGERY: CASE STUDY OF THE SALTO GRANDE DAM, BRAZIL. Authorea. 08 July 2025. DOI: https://doi.org/10.22541/au.175195825.50260336/v1 DOI: https://doi.org/10.22541/au.175195825.50260336/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu.