Exploring the functional connectivity between the Kis-Balaton Water Protection System and Lake Balaton using satellite data

Lake Balaton, a shallow polymictic freshwater lake in Central Europe, became eutrophic in the 1970s. To retain the inorganic nutrients from the main tributary River Zala, a semi-artificial system called the Kis-Balaton Water Protection System (KBWPS) was constructed in the early 1980s. In 2015, the system was reconstructed and modernised, thus offering the opportunity to evaluate the effectiveness of the functional connection between the KBWPS and Lake Balaton over the past 20 years and to compare its impact before and after the reconstruction. To this end, time series data of algal biomass in Lake Balaton between 1999 and 2019 based on Landsat 7 satellite data were analysed. Over the last 20 years, the algal biomass in Lake Balaton showed an increasing trend (0.009 ± 0.011% increase per year), with territorial specificities also observed. No change was noted in the western part, while an increase was recorded in the eastern part of the lake. A significant difference in the rate of algal biomass accumulation was noticed before (annual increase of 0.008 ± 0.019%) and after (0.240 ± 0.306% per year) the KBWPS reconstruction. Given that the largest increase in algal biomass after reconstruction was observed in the outermost KBWPS basin of Lake Balaton, it appears that mesoscale environmental, water balance, or other factors affecting the lake are playing a role in this increase, rather than the KBWPS reconstruction. This research highlights the potential to study aquatic ecosystems using Earth observation techniques, and how mesoscale factors such as changes in the local climate regime or shifts in lake management can greatly impact the trophic state of a large shallow lake. Effectively identifying these factors is crucial in maintaining the proper status of aquatic ecosystems.


Introduction
Aquatic ecosystems are facing significant threats globally (Revenga et al. 2000;Wang et al. 2018), with anthropogenic eutrophication being a significant contributor.While eutrophic lakes are a natural occurrence, human activities resulted in an increased rate of lake enrichment in the latter half of the twentieth century, causing severe problems in many countries (Heisler et al. 2008;O'Neil et al. 2012).This has led to the rapid accumulation of algae in lakes, disrupting the natural processes and phytoplankton communities of the aquatic ecosystem (Vollenweider 1968;Padisák and Reynolds 1998).Efforts have been made globally to reverse eutrophication (Moss et al. 1986;Jeppesen et al. 2005;Fastner et al. 2016), and Lake Balaton serves as a positive example.
Lake Balaton is a large (596 km 2 ), shallow (approx.3.7 m) recreational freshwater lake.In the early 1970s, a surge in nutrient levels caused a shift from mesotrophic to eutrophic in the lake's water (Herodek 1984).The primary source of nutrients was the Zala River, which flowed into the westernmost basin of Lake Balaton and increased primary production by five-fold within a decade (Herodek 1984).To combat eutrophication, the Kis-Balaton Water Protection System (KBWPS) was initiated in the early 1980s.The KBWPS includes an 18-km 2 shallow (less than 1.3 m) artificial lake system (Lake Hídvégi) and a 16-km 2 wetland, which were constructed at the site of the former Kis-Balaton wetland by the mid-1980s.Lake Hídvégi was designed to retain nutrients and provide ideal conditions for algae growth, while the wetland component of the KBWPS prevented these algae from entering Lake Balaton (Pomogyi 1993;Padisák and Istvánovics 1997;Hatvani et al. 2011).To ensure optimal 106 Page 2 of 11 functioning of the KBWPS, a hydrological retention time of 30 days in Lake Hídvégi and 90 days in the wetland is maintained (Pomogyi 1993;Tátrai et al. 2000).
Since its opening in 1984, the KBWPS has consistently been retaining increasing levels of phosphorus (P) each year (Pomogyi 1993;Tátrai et al. 2000;Istvánovics et al. 2007), eventually reversing the eutrophication process in Lake Balaton by 1995 (Padisák and Istvánovics 1997;Istvánovics et al. 2007).Following this, a steady oligotrophication of the lake took place, leading to a decrease in algal biomass and the return of the algal species present in the 1950s (Pálffy and Vörös 2019).In 2013, the KBWPS underwent modernisation and redesign, with a focus on increasing its flexibility during different water regimes.New water pathways and regulatory structures were added, along with new elements in the monitoring system.Additionally, the wetland area was expanded by 35 km 2 to ensure maximum efficiency.The redesigned KBWPS became fully operational in the spring of 2015.
Since 1995, Lake Balaton has been meso-oligotrophic, with occasional, spatially and temporally limited spikes of eutrophication (Palmer et al. 2015a).Between 2014 and 2019, excluding some localised and short-lived algal blooms (Pálmai et al. 2016), the eastern half of the lake was oligotrophic throughout the year.However, starting in June 2019, an unprecedented hypertrophic algal bloom in the western half of the lake happened, peaking in early September of that year (Fig. 1).The effectiveness of the KBWPS to protect Lake Balaton was questioned following this bloom, although an alternative hypothesis on the background of this hypertrophic event is supported by most experts (Istvánovics et al. 2022).Nevertheless, for the sake of clarity, the performance of the KBWPS was analysed by comparing the accumulation of algal biomass in Lake Balaton before 2013 and after 2015 using satellite data.
The use of remote sensing to study large areas is advantageous, since long-term and large-scale monitoring of open waters would be time-and resource-consuming (Palmer 2015a).Synoptic and regular monitoring of key water quality indicators in lakes can enhance our understanding of the processes involved and improve management of these bodies of water.
Lake Balaton is nearly an ideal water body for Earth observation and remote sensing studies of aquatic Fig. 1 Change of algal biomass (chl-a, µg l −1 ) at the end of summer, beginning of autumn in Lake Balaton.Chlorophyll-a maps were created using Sentinel-2 MSI satellite data using the Case-2 Regional ecosystems.It offers a variety of optical water types, including eutrophic westernmost and oligotrophic easternmost basins, high dissolved organic matter at the inflows, and differences in turbidity between the northern and southern shores.Moreover, the size of the lake allows the study of mesoscale processes, while the infrastructure around the lake makes it accessible from all 240 km of the shore.In addition, the relatively large macrophyte cover (~ 12 km 2 ) makes it an excellent target not only for water quality studies, but also for remote sensing of macrophytes.Consequently, numerous articles have focused on remote sensing of water quality in Lake Balaton, dating back to the early days of remote sensing (Büttner et al. 1987;Gitelson 1993), with some recent results (Tyler et al. 2006;Hunter et al. 2008;Palmer et al. 2015b).Furthermore, several studies have focused on remote sensing of its macrovegetation (Stratoulias et al. 2015;Stratoulias and Tóth 2020).
One of the crucial management concerns for inland lakes is the amount and biomass of algae (trophic state).This study evaluates the efficacy of the KBWPS by analysing the spatial and temporal distribution of algal biomass in Lake Balaton from 1999 to 2019.The main hypothesis of the study is that the reconstruction of the KBWPS had no impact on the accumulation of algal biomass before (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013) and after (2015-2019) its renovation.

Study areas
Lake Balaton and the Kis-Balaton Water Protection System Lake Balaton (Fig. 2a) is a freshwater lake located in Central Europe (46.83°N, 17.71° E).It is the 38th largest lake in Europe, covering an area of 596 km 2 .The lake is 78 km long, with a width ranging from 1.4 to 15.3 km.The average depth of the lake is 3.7 m, with its deepest point reaching 11.8 m.Lake Balaton is divided into four basins, with the Keszthely basin (basin 1 in this work) located on the western side and the Siófok basin (basin 4) on the eastern side.The catchment of Lake Balaton covers an area of 5765 km 2 and includes many small inflows and the main tributary, the Zala River, which flows into basin 1.The concentration of industry in the catchment is along the Zala River that supplies its waters to the Kis-Balaton Water Protection System (KBWPS-46.63°N, 17.15° E), consisting of an artificial lake in its western part and a wetland in its eastern part (Fig. 2b).

Satellite data acquisition and analysis
Freely available satellite remote sensing data were considered to investigate the effects of the reconstruction of the KBWPS on its nutrient retention and its role as a control of eutrophication in Lake Balaton.A comparative analysis of the effects before and after the reconstruction was considered essential, thus the study included data from 4 years after the reconstruction (2015-2019), which required a corresponding time frame of at least 4 years before the reconstruction (2009 to 2012).The Landsat 7 satellite was chosen for its higher resolution.
To assess the water chlorophyll-a content over a long period, Landsat 7 ETM+ data collected between June 1999 and July 2019 were analysed (Fig. 3).The data were obtained from Google Earth Engine (https:// earth engine.google.com/), and preprocessed to eliminate systematic geometric and radiometric distortions.The database was created by targeting clear Landsat images (with cloud cover < 5%) of the open water area of Lake Balaton.A total of 602 Landsat images were acquired and processed over the period, although owing to clouds and other errors, as few as 340 images were used for certain areas of the lake.A chlorophyll index was calculated for pixels with normalised difference water index (NDWI) values over a threshold of 0.6 using the band ratio between ETM+ 1 and ETM+ 3 (Gitelson et al. 1996) (Fig. 3).
owing to the rarity of Lake Balaton water with chlorophyll-a content below 3 µg l −1 , only the ratio ETM+ 3/ ETM+ 1 was used for the analysis.In the regression models established, the logarithmically transformed chlorophyll-a concentration was used as an independent variable, while the dependent variable was the logarithmically transformed ETM+ 1 and ETM+ 3 band ratio; from the best regressions (highest R 2 values) (R Development Core Team 2012), maps were generated.These data were downscaled by overlaying a 1000 m × 1000 m grid to reduce errors from the Landsat 7 ETM+ Scan Line Corrector failure (error data were omitted), and average water chlorophyll-a content was calculated for each cell (Fig. 3).
A trend analysis of the algal biomass accumulation in each 1000 m × 1000 m cell of Lake Balaton was performed to determine the direction of change during the study period.Trend analysis was performed on annual (all data) and summer (June-July-August) chlorophyll-a data separately.The average slope of the trend lines was also calculated for each 3-year rolling water chlorophyll value in each cell.

Water sampling
Water samples from Lake Balaton were collected on a nearmonthly basis from 1999 to 2019.Sampling was done at five locations in the middle of each basin of the lake (1: 46° 44.095′ N, 17° 16.583′ E; 2: 46° 45.112′ N, 17° 25.145′ Chlorophyll-a was determined on the same day as the sample was collected using a whole water column integrated water sample.Water (1000 ml) was filtered using Whatman GF-5 filter papers, and the pigments were analysed after extraction in 60 °C methanol, followed by clarification through centrifugation at 10,000 rpm for 10 min (Iwamura et al. 1970).
To validate satellite data, their consistency with in situ data was tested.The validity of all pixels in the 3 × 3 pixel box surrounding the in situ sampling location was checked, with a focus on low covariance between all nine pixels.A total of 33 same-day matches were found for the 1999-2019 timespan for Lake Balaton.The chlorophyll index obtained from satellite data was correlated with the in situ measured chlorophyll-a data (Fig. 4).The maps built from chlorophyll indices were then corrected using the following formula: chl-a = chl-index/1.7514.Using this correction, chlorophyll data were calculated for each available pixel (Fig. 3) and then averaged for each cell.
Total phosphorus data were obtained from West-Transdanubian Water Directorate.

Statistical analysis of the data
Empirical regression analyses were performed to assess the relationship between chlorophyll content measured in situ and the corresponding chlorophyll index estimated from remote sensing data.Ordinary least squares (OLS) regression was used for trend analysis.To avoid seasonal bias, additional trend analyses were performed using only summer (June, July, August) data sets.The strength of the association between the two variables was examined using Pearson product moment correlation analysis.Statistical analyses, including the calculation of averages, standard deviations, regressions and correlations, were performed using R (R Development Core Team 2012).

Trophic trends in Lake Balaton
The water chlorophyll content exhibits characteristic spatial and temporal patterns in Lake Balaton (Fig. 5).The trophic gradient of Lake Balaton determines the specific algal biomass of its basins, thus the westernmost basin 1 of the lake Balaton had water chlorophyll-a levels ranging from 3.2 to 92.4 µg l −1 , with an average of 20.8 ± 17.8 µg l −1 , making it meso-eutrophic according to the satellite data, with short periods of local hypertrophicity (Fig. 5).Conversely, the easternmost basin 4 of Lake Balaton was oligo-mesotrophic, with average water chlorophyll-a content of 4.9 ± 2.8 µg l −1 (Fig. 5).Temporal specificities were also observed in Lake Balaton, including a regular, annual large phytoplankton bloom in late summer and early fall, and occasional smaller blooms in spring, which were more prominent in basins 1 and 2, i.e., the western half of Lake Balaton (Fig. 5).
Since the construction of the KBWPS, the efficiency of the system has decreased significantly as the total phosphorus retention decreased from 40.3 ± 27.4% in 1986-1999 to −1.4 ± 30.5% in the study period (t-test t = 8130, P = 1.7 × 10 -13 ).Nevertheless, the phosphorus retention capacity of the KBWPS between 1999 and 2019 showed no significant trend (Pearson product moment correlation r = −0.171,P = 0.093, Supplementary Fig. 1).
The analysis of algal biomass accumulation in Lake Balaton was conducted between 1999 and 2019.During this period, the average trend of algal biomass accumulation was positive, with an increase of 0.009% per year.The highest increase was recorded in basin 4 (0.016% per year), while basins 1 and 2 showed no signs of change or even a negative slope (−0.0045% in basin 2).The summer data sets showed similar trends (Supplementary Table 1, Supplementary Fig. 2).
The efficiency of the KBWPS was evaluated by comparing the algal biomass accumulation trends before (prior to 2013) and after (post 2014) the system reconstruction.Before the KBWPS reconstruction, the trend of algal biomass change in Lake Balaton was slight positive except for basin 2, which showed a negative trend.After the KBWPS reconstruction, the overall trend increased by 3273% from annual 0.007-0.24%per year biomass accumulation (Table 1).Very similar data were obtained if only summer data sources were analysed (Supplementary Fig. 3) (Figs. 6, 7).
Further temporal analysis showed that, in most cases (10 out of 18 years), short-term trends were close to zero, especially before 2014 (Fig. 8).Occasionally (e.g. in basins 1 and 2 in 2010 and 2011, and in basin 4 in 2008), trends exceeded the standard deviation for the 20-year dataset, but these observed changes were not long-term Table 1 Lake-wide and basinspecific averages of algal biomass change (% per year) in Lake Balaton during the study period

Discussion
Assessment of temporal patterns and spatial zonation within lakes is crucial for a better understanding of the processes and functions of aquatic ecosystems, and for their effective management.In response to this need, an increasing number of large-scale process studies have been initiated to assess spatial and temporal aspects of abiotic and biotic elements of ecosystems (Blowes et al. 2019;Ho et al. 2019;Pilotto et al. 2020).However, these studies require consistent and adequate data collection, which is not always feasible with traditional sampling methods.Earth observation, despite its limitations, provides a good estimate with a synoptic view of water quality parameters (Giardino et al. 2001;Hestir et al. 2015;Palmer et al. 2015a) and has become essential tool in the study of Lake Balaton (Büttner et al. 1987;Tyler et al. 2006;Palmer et al. 2015b).This study used Landsat 7 ETM+ data to investigate the long-term effects of a water protection system on a lake.Several challenges were encountered.The reduced radiometric and spectral resolution of Landsat 7 limited its suitability for assessing water quality.In addition, the failure of the scan line corrector further hampered the usability of the data.A further complication arose from the use of an algorithm originally developed for dinoflagellate algae, whereas scumforming cyanobacteria were the dominant algae in Lake Balaton during the study period.Despite all these, the Earth observation data of this study, supported by in situ measurements, showed that the phytoplankton phenology of Lake Balaton was characterised by two main blooms during the study period.Chronologically, the first bloom, which occurs almost annually in spring, was a relatively small phytoplankton bloom (up to 16.1 ± 12.9 µg l −1 in basin 1) dominated by diatoms.The second, larger eutrophication event (up to 49.3 ± 21.5 µg l −1 in basin 1) occurred in late summer and early autumn and was dominated by Cyanobacteria.These phenological events fit perfectly into the data of previously discussed studies (Mózes et al. 2006;Hajnal and Padisák 2007;Présing et al. 2008;Palmer et al. 2015b).Both blooms, especially the latter, were more pronounced in the western, more eutrophic half of the lake.During the time between the blooms, clear water stages with an average chlorophyll-a content of 6.5 ± 4.5 µg l −1 were recorded.
The long-term trend of algal biomass change derived from Landsat 7 data showed no significant change in the trophic state of Lake Balaton between 1999 and 2019, with an annual accumulation of 0.009% of algal biomass throughout the lake.However, a clear distinction between the western and eastern parts of Lake Balaton was observed: a slight decrease in algal biomass (−0.005%) in the western part and a slight increase (0.016%) in phytoplankton abundance in the eastern basins.This may seem to contradict the established trophic gradient of the lake, but the increase in chl-a content in the eastern parts of the lake was not substantial (0.010% and 0.016% per year in basin 3 and 4, respectively), and until 2019 the chlorophyll-a content in basin 4 remained below 7 µg l −1 .The results of the current study go beyond the traditional spatial division of Lake Balaton into eutrophic western and mesotrophic eastern parts (Somlyódi and Van Straten 1986;Herodek et al. 1988) and show a slow but steady shift away from this internal zonation.
Phytoplankton is crucial to the survival of lake ecosystems, but excessive growth due to factors such as humancaused nutrient pollution can severely degrade water quality, reduce biodiversity, and alter the species composition and food chain of the system.In the 1970s, eutrophication of Lake Balaton was caused by increasing pollution from local agriculture and industry, combined with inadequate wastewater treatment.Intense algal growth, dominated by Raphidiopsis (Cylindrospermopsis) raciborskii, continued until 1994 (Padisák 1997(Padisák , 1998;;Sprőber et al. 2003).Proper management of nutrient sources has been shown to alleviate eutrophication problems and restore the ecological status of the lake (Jeppesen et al. 2005;Istvánovics et al. 2007;Hatvani et al. 2011), but positive changes can easily be reversed by local, regional or global processes.
The KBWPS was established with the aim of protecting Lake Balaton through phosphorus retention.During the first 13 years of its operation, the system did its job and proved to be successful, capturing about 40% of the total phosphorus from the Zala River.The KBWPS had a significant impact on the water quality of Lake Balaton in the past, especially in the 1980s (Tátrai et al. 2000;Hatvani et al. 2011).It significantly reduced the algal biomass in Lake Balaton within a decade of its construction (Padisák and Reynolds 1998;Istvánovics et al. 2007;Fastner et al. 2016).However, between 1999 and 2019, the KBWPS was not always able to achieve the projected phosphorus retention, and sometimes, especially during low water periods (2000-2004 and 2011-2012), it was even a source of phosphorus.This study also showed that, between 1999 and 2013, before the reconstruction of the KBWPS, there was only a slight change in the algal biomass in the lake (between −0.002% and 0.01% per year).However, this stability was relative and larger interannual variations were observed (−0.342% to 0.233% per year), probably due to interannual variability of microand mesoclimatic parameters or the water balance of the lake.This led to a small (0.055% per year) increase in phytoplankton abundance in the lake as a whole.
From the point of view of phosphorus retention, the reconstruction of the KBWPS had no effect, as the amount of total phosphorus entering Lake Balaton remained practically unchanged between 1999 and 2019.Nevertheless, after the reconstruction of the KBWPS in 2015 and later, a noticeable increase in algal biomass was recorded, with an average of 0.240% per year for the total lake level.However, the trends in the basin levels varied between −0.015% and 0.378% per year, and the 3-year rolling trend varied between −0.056% and 0.151% per year.The outermost basin of Lake Balaton from the KBWPS (basin 4) showed the highest increase in algal biomass, but owing to the 3-year retention time of the lake and the large distance, the direct impact of the KBWPS reconstruction is considered negligible.Without further analysis, it is possible only to speculate on the causes behind this phenomenon.The long-term changes in phytoplankton abundance or shifts in any community type could be the result of global or regional changes (Blowes et al. 2019;Ho et al. 2019;Pilotto et al. 2020) such as the slow warming of the lake due to climate change (Pálffy and Vörös 2019), or the combination of increased warming temperatures (Joehnk et al. 2008;Paerl and Huisman 2008;Kosten et al. 2012;Taranu et al. 2012), intensified run-offs (Paerl and Huisman 2009;Michalak et al. 2013), and increased atmospheric CO 2 (Hutchins et al. 2013;Sandrini et al. 2016;Visser et al. 2016) on algal growth and biodiversity.
Knowledge of the spatial patterns and temporal dynamics of lakes is crucial for their effective management, especially in the case of large, heterogeneous water bodies.This study showed that the functionality of the Kis-Balaton Water Protection System remained largely unchanged over time.Its reconstruction coincided with autochthonous changes in the lake resulting from persistent regional climate forcing.Furthermore, this long-term data set is of even greater importance as it provides an opportunity to analyse biotic processes within the lake, to reveal previously undetected trends and to help identify evolving characteristics of the Lake Balaton ecosystem.This study has demonstrated the potential of using Earth observation data for more than just extrapolation of existing in situ data sets, by attempting to establish additional key limnological relationships within a large shallow freshwater lake.
The 2015 reconstruction did not improve the operational efficiency of the KBWPS, as neither the long-term nor the short-term trends in the western basins of Lake Balaton showed any significant improvement.Nevertheless, the KBWPS should remain an important system for the protection of Lake Balaton, but its performance needs to be continuously evaluated and improved, while new sciencebased strategies should also be explored.In general, it can be said that management of the KBWPS will not be possible without assessing and understanding its actual functioning in this changing environment.The large areas of the KBWPS and Lake Balaton make it impossible to carry out effective monitoring without automated high-frequency monitoring systems or remote sensing techniques, as the detection of nutrient surges outside and inside the KBWPS and the early warning of phytoplankton blooms could only be successfully implemented in this way.This proactive, data-driven approach will allow water authority managers to implement timely, automated, and targeted interventions to mitigate the effects of nutrient surges or lack thereof.In addition to early warning systems that can provide real-time alerts, the development of forecasts is necessary to support strategic decision-making processes for both the KBWPS and Lake Balaton management and water resource protection.This low-maintenance, high-throughput, multi-platform approach will provide researchers, lake managers and policy-makers with a comprehensive view of aquatic ecosystem functioning, enabling informed decision-making and timely responses to environmental challenges.

Fig. 2 A
Fig. 2 A Situation of Lake Balaton (a) and the Kis-Balaton Water Protection System (b) within Hungary.B The division of Lake Balaton into basins as used in this study

Fig. 3
Fig. 3 Flowchart of the data processing used in the study

Fig. 4
Fig.4Correlation of same-day matches between in situ chlorophylla measurements in the pelagic zone of Lake Balaton and the chlorophyll index (chl-index) calculated from Landsat 7 satellite imagery.The regression line (chl-a = chl-index/1.7514)is shown as a solid line, while the dotted red lines are 95% confidence intervals.Pearson product moment correlation coefficient r = 0.639, P ≤ 0.001(n = 33)

Fig. 6
Fig.6Change of algal biomass (% per year) in Lake Balaton between 1999 and 2019 based on annual data sets.The map shows the slope of the trend line from the linear regression fitted to the sequences of water chlorophyll-a content from each 1000 m × 1000 m cell in Lake Balaton.Calculated from Landsat 7 imagery acquired between June 1999 and July 2019

Fig. 7
Fig. 7 Change of algal biomass (% per year) in Lake Balaton between 1999 and 2013 (A) and between 2015 and 2019 (B) based on annual data sets.Map of trend line slopes from the linear regression line fit-

Fig. 8
Fig. 8 Change of algal biomass (% per year) in the basins of Lake Balaton between 2000 and 2018.Dotted line is the 20-year average, while dashed lines are standard error lines for the 20-year data set