Estimation of the level of eutrophication of coastal waters of the Baltic Sea on the basis of Earth remote sensing data

. One of the serious environmental problems is the eutrophication of water bodies, so monitoring of this process is necessary. The purpose of the study is to assess the level of eutrophication of coastal waters based on the analysis of satellite imagery data using NDVI. The NDVI value and the water area where NDVI is greater than zero were used as indicators of eutrophication. The study was carried out in the Gdansk and Kaliningrad bays of the Baltic Sea. Computer processing of images in .tif (NDVI) and .png (RGB images) formats from satellites of the Sentinel-2 and Landsat family for 2013-2021 was performed. As a result of the study, it was found that the calculation of the NVDI index and the area where this index is greater than zero is an adequate method for monitoring eutrophication, since it demonstrates areas of the water area where a high level of vegetation is observed. Compared to the calculation of chlorophyll content indices (which are very difficult to establish standard values), the calculation of NDVI is easier to interpret. The study shows that the level of eutrophication in the bays of the Baltic Sea depends on seasonality, it usually increases from May to August, and decreases in September. Annual fluctuations are also observed, in particular, a decrease in eutrophication in 2020-2021 due to the decline in economic activity and anthropogenic pollution. Sharp and rapid increases in the level of eutrophication were found, but in the study period, the ecosystems of the bays were capable of self-healing.


Introduction
The paradigm of sustainable development implies a responsible attitude towards the environment, in particular, the preservation of clean air and water [1]. Of great importance for modern mankind is the state of coastal sea waters, in particular, bays [2]. This is due to the active economic and recreational use of the coasts. At the same time, as practice shows, it is the bays that often suffer from severe pollution by sewage and other waste. Suffice it to recall the example of the Gulf of Mexico, which was badly affected by pollution even before the accident on the drilling platform and the spill of a large amount of oil in 2010. By the beginning of the 2000 it has become a zone devoid of life due to the runoff of fertilizers, chemicals, wastewater [3]. One of the main environmental problems of coastal waters is eutrophication. It is understood as such a type of pollution, when an excessive amount of nutrients for plants accumulate in the water, algae actively multiply (the socalled "water bloom" occurs) [4]. In the course of reproduction and subsequent death of a large number of aquatic plants, oxygen reserves are rapidly depleted. In addition, a large amount of vegetation limits the ability of sunlight to penetrate the water column. As a result, there is a massive death of aquatic organisms, which further enhances eutrophication and deprives the reservoir of oxygen. Also, in the process of eutrophication, the natural properties of water are lost, its composition changes [5].
If we are talking about a freshwater reservoir, then eutrophication makes it difficult to use water as drinking water, since it is extremely difficult to purify water with a large number of dead plants [6]. In addition, the reservoir is losing its importance as a recreational facility. Poisoning of animals and humans with algae waste products, such as toxoid, is also possible. In extreme cases, eutrophication causes the water body to turn into a swamp or disappear completely. It is now known that eutrophication can be caused by both natural and anthropogenic factors. In the first case, excessive reproduction of algae is stimulated by high temperature, sunlight, and calm [7]. Anthropogenic eutrophication occurs when nitrogen and phosphorus enter the water -nutrients that cause rapid reproduction of aquatic plants. The source of nitrogen and phosphorus, in turn, are mineral fertilizers from the fields, urban wastewater containing phosphates, and industrial effluents [8]. The regulation of eutrophication processes requires monitoring the state of waters in order to identify problems in a timely manner, as well as take measures to prevent negative processes. Initially, manual water sampling was used to monitor eutrophication, followed by visual assessment and laboratory analysis. This method was implemented in the study by A. Damar et al. to study the state of the microbiota in the Jakarta Bay of the Java Sea [9]. However, such studies require large financial and time costs. In addition, its application necessarily requires a specialized research laboratory, which is not available to all coastal countries and regions.
E. Neverova-Dziopak, Z. Kowalewski and M. Preisner note other shortcomings of laboratory water analysis as a method for assessing eutrophication. In particular, the measurement of the content of phosphorus, nitrogen, chlorophyll in water, the study of transparency do not provide information on the state of the biotic balance in dynamics. Along with this, the actual values of the relevant indicators are difficult to interpret in terms of compliance. So, under different conditions, the normal content of chlorophyll is recognized from 6 µg/L and 100 µg/L [10]. In the digital economy, many operations that were previously carried out by a person with limited accuracy and reliability can be performed by unmanned vehicles, a bot, a neural network quickly, at minimal cost and obtaining the most accurate information [11]. Therefore, to study eutrophication, it is promising to use digital technologies; in particular, computer processing of Earth remote sensing data obtained from artificial Earth satellites and unmanned aerial vehicles.
An analysis of the literature shows that various methods for analysing information from satellite images have already been used to survey water bodies. In particular, M. Peppa, C. Vasilakos, D. Kavroudakis used the analysis of images from satellites of the Sentinel-2 family for eutrophication monitoring for lake Pamvotis, Greece [12]. These researchers determined the content of phytoplankton in the water using chl-a detection algorithms. As a result, zones of the lake subject to eutrophication were found. H. Liu et al. also used images from satellites of the Sentinel-2 family to monitor the eutrophication of lakes in the vicinity of the city of Wuhan (China). The trophic level index, which reflects the content of chlorophyll in water, was determined in the cited work according to satellite images. It made it possible to reveal patterns of eutrophication by months and by types of lakes [13].
In [14], a unified system for global monitoring of the chlorophyll content in water is proposed, which is important for supporting the Sustainable Development Goals.
However, the assessment of the level of eutrophication by the content of chlorophyll is not entirely accurate, since the corresponding indicator strongly depends on the season of the year, climatic conditions, and specific properties of the reservoir. In addition, the determination of the chlorophyll content, as noted above, does not characterize the biotic balance. According to the authors, a more accurate assessment of eutrophication can be given by the use of the NDVI index, a classical metric of the state of the vegetation cover of an area. It is used to assess the state of agricultural land, forests, but has not yet been used to study the eutrophication of waters. Thus, the aim of the study is to estimate the level of eutrophication of coastal waters based on the analysis of satellite imagery data using NDVI. NDVI can take values from minus one to one, but in the study we were interested in values greater than zero, since negative values mean the absence of vegetation. Along with the .tif images, traditional RGB images were also used to improve the accuracy of the assessment, which provide information about the blue-green glow (typical for clusters of blue-green bacteria and algae). When the two types of images are superimposed, clumps of vegetation are more clearly identified due to the correlation between the blue-green glow and the NDVI index. To calculate the average NDVI value and the area with a positive NDVI value, the "QGIS" cross-platform geoinformation system was used. Figure 1 shows an example of a combined satellite image used in the study. Overlaying NDVI and RGB images allows you to see that the green glow in the RGB image (colour code #d9ef8b) corresponds to the value 0.5 <NDVI ≤ 0.6 (according to the colour legend presented in Table 1). This means that the vegetation is developed enough, but not excessively. A total of 33 images were analysed. An example of one of the images used to calculate NDVI is shown in Figure 2. This image is from July 2017. There are visually distinct patches of different colours. Larger accumulations of a lighter shade are closer to the shores of the bays; therefore, eutrophication is stronger in these parts of the sea. In the course of computer processing of this image, it was found that the average NDVI value is 0.30, and the area where this index is greater than zero (that is, there is vegetation) is 2784.6 thousand square meters. The results of processing all images are shown in Table 2. It should be noted that it was not possible to achieve full coverage of all months of the summer period due to the unavailability of some of the images. The data presented in Table 2 make it possible to assess and analyse the eutrophication of the study area, to identify trends and cause-and-effect relationships. In particular, the dependence of the level of eutrophication on the season of the year was found ( Figure 3). Figure 3 clearly shows seasonal changes in the level of eutrophication. The minimum values of NDVI are observed, as a rule, in May, and the maximum -in August, which correlates with the water temperature and the level of solar radiation, which contribute to the development of vegetation. This fact confirms the correctness of the obtained estimates of eutrophication and corresponds to recognized scientific ideas.

Results
When analysing annual data, a pronounced multidirectional dynamics is revealed: in 2014, the level of eutrophication of the water area was maximum, by 2017 it decreased, in 2019 it increased again, and then it decreased.
Changes in NDVI over the years are due to both biological and climatic factors (different temperature conditions in different years) and changes in the anthropogenic load. In particular, in 2020-2021 due to the COVID-19 pandemic, there was a drop in shipping activity, a decrease in sewage discharge, which caused a decrease in NDVI. The dynamics of the area of the studied water area with a NDVI value above zero is shown in Figure 4. The data in Figure 4 show that the eutrophication of the study area is characterized by sharp fluctuations when there is a rapid increase in the area occupied by vegetation (for example, this is May 2016 and August 2019). This is due to the formation of favourable weather conditions, in particular, high solar insolation and air temperature. At the same time, the rapid reproduction of algae, in accordance with modern biological concepts, causes an equally rapid subsequent death due to a lack of oxygen. Thus, in the course of the study, specific quantitative estimates of eutrophication in the Gdansk and Kaliningrad bays of the Baltic Sea were obtained, which correspond to known ideas about its dynamics and factors.

Discussion
In this study, the authors sought to make a certain contribution to identifying the features of eutrophication of the southeastern coast of the Baltic Sea and to demonstrate the potential of satellite imagery analysis for a quick, accurate and cheap assessment of the state of vegetation in water bodies. In the course of the study, by desk analysis of satellite images in the free (that is, free) cross-platform geoinformation system "QGIS", it was possible to accurately determine the area and boundaries of the water area with developed vegetation, to measure the intensity of its growth in dynamics.
At the same time, unlike existing studies, the NDVI index was used for the analysis of marine areas, information about which can be obtained from existing satellite images. Compared to studies where chlorophyll indices were measured, the NDVI assessment in this study allows you to immediately demonstrate which part of the water area has a high level of vegetation development (regardless of the characteristics of the object under study).
Thus, the possibility of estimating the level of eutrophication without field studies with water sampling or using lidars has been shown. At the same time, a minimum level of costs and significant measurement accuracy are achieved. Although in some cases field studies are necessary for a more complete assessment of the situation in small water bodies or sections of large water bodies, the methodology presented in the study is well suited for the purposes of global or national monitoring of water bodies.
In the context of existing studies on the eutrophication of the Baltic Sea, this work confirms the presence of a problem in its southeastern part (Baltic proper). The known works did not analyze coastal waters, which are usually characterized by higher levels of eutrophication. At the same time, it was shown that the situation is not catastrophic, and a sharp increase in NDVI and the area of the water area with a high level of NDVI is gradually compensated by the self-regulation of the aquatic ecosystem. The obtained specific estimates of NDVI and areas of water with a high level of NDVI open up opportunities for further deeper analysis and modeling of cause-and-effect relationships for eutrophication processes in the southeastern coast of the Baltic Sea.

Conclusions
In this study, the authors sought to make a certain contribution to identifying the features of eutrophication of the southeastern coast of the Baltic Sea and to demonstrate the potential of satellite imagery analysis for a quick, accurate and cheap assessment of the state of vegetation in water bodies. In the course of the study, by desk analysis of satellite images in the free (that is, free) cross-platform geoinformation system "QGIS", it was possible to accurately determine the area and boundaries of the water area with developed vegetation, to measure the intensity of its growth in dynamics.
At the same time, unlike existing studies, the NDVI index was used for the analysis of marine areas, information about which can be obtained from existing satellite images. Compared to studies where chlorophyll indices were measured, the NDVI assessment in this study allows you to immediately demonstrate which part of the water area has a high level of vegetation development (regardless of the characteristics of the object under study).
Thus, the possibility of estimating the level of eutrophication without field studies with water sampling or using lidars has been shown. At the same time, a minimum level of costs and significant measurement accuracy are achieved. Although in some cases field studies are necessary for a more complete assessment of the situation in small water bodies or sections of large water bodies, the methodology presented in the study is well suited for the purposes of global or national monitoring of water bodies.
The research was conducted on the equipment of the Research Equipment Sharing Center of Kemerovo State University, agreement No. 075-15-2021-694 dated August 5, 2021, between the Ministry of Science and Higher Education of the Russian Federation (Minobrnauka) and Kemerovo State University (KemSU).