Long term changes in the quality of the aquatic environment of thermally polluted Lake Liche´nskie, Central Poland

Long term changes in the quality of the aquatic environment of thermally polluted Lake Liche´ nskie, Central Poland Michał Woszczyk a,b,* , Michael Brechbühler c a Biogeochemistry Research Group, Adam Mickiewicz University, Bogumiła Krygowskiego 10, Pozna´ n 61-680, Poland b Department of Life Safety and Environmental Protection, School of Earth Sciences, D. Serikbayev East Kazakhstan Technical University, Serikbayeva 19, ¨ Oskemen 070000, Kazakhstan c Department Surface Waters Research & Management, EAWAG, Überlandstrasse 133, Dübendorf 8600, Switzerland ARTICLE INFO Keywords: Lake Thermal pollution Salinization Alkalinity Water quality ABSTRACT Study region: Central Europe, Gniezno Lake District, Poland Study focus: The purpose of this study is to depict and explain long-term trends in surface water temperatures (LSWT) and chemical composition in Lake Liche´ nskie (LLi), which since 1960s has been involved in a cooling system of two electric power plants (PP) and thus has been prone to thermal pollution (TP). For the analyses we used 24-year long stationary monitoring and satellite data. New hydrological insights for the region: We estimated that LLi was 3.81◦C warmer than natural lakes in the region and demonstrated that the TP displayed spatial and seasonal variability. The data shows that owing to a reduction in the PP activity the LSWT has constantly been decreasing at a rate of 0.09◦C⋅y −1 . Because, the lake has also been supplied with saltwater and highly alkaline effluents from nearby brown coal mine, LLi has been prone to salinization and alkalinization. The former process is still ongoing but alkalinization is declining, which is interpreted as a self-recovery of the lake triggered by a reduction in brown coal mining in the region. The knowledge of environmental conditions in the lake as well as its long-term changes is crucial for developing lake management strategies in the face of planned incorporation of the lake in a cooling system of the new nuclear power plant in the vicinity of the lake. 1. Introduction Thermal pollution (TP) involves an increase (or decrease) in temperature of aquatic systems to non-natural levels and thus leads to a distortion in thermal equilibrium between the systems and the ambient atmosphere. Because pollution with cold water is rather uncommon (Kennedy, 2004), the TP is usually tantamount to the excess warming of lakes and/or rivers. The warming is primarily driven by a point-source discharges of heated industrial effluents, from energy-producing industry (both coal-based and nuclear) in particular (Rosen et al., 2015; Raptis et al., 2016, Miara et al., 2018; Yavari and Qaderi, 2020; Li et al.l, 2025), with some, usually minor, contribution from other processes (e.g. deforestation, urban runoff; Dodds and Whiles, 2010 and references therein). The highest discharges of heated waters are released by the power plants working in an once-through mode, in which the water collected from a river/lake is passed through the heat exchangers and finally rejected back into its source albeit at temperature 8 – 12◦C higher * Corresponding author at: Biogeochemistry Research Group, Adam Mickiewicz University, Bogumiła Krygowskiego 10, Pozna´ n 61-680, Poland. E-mail address: [email protected] (M. Woszczyk). Contents lists available at ScienceDirect Journal of Hydrology: Regional Studies journal homepage: www.elsevier.com/locate/ejrh https://doi.org/10.1016/j.ejrh.2025.102917 Received 12 June 2025; Received in revised form 19 October 2025; Accepted 4 November 2025 Journal of Hydrology: Regional Studies 62 (2025) 102917 Available online 11 November 2025 2214-5818/© 2025 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). (Madden et al., 2013; Raptis et al., 2016). As of the US for example it is estimated that nearly 50 % of surface waters thereof is exposed to thermal pollution (Echols et al., 2009). Rivers seem to be more jeopardized to TP than the other aquatic systems because they often receive direct inputs of heated waters. Raptis et al. (2016) showed that owing to this problem some rivers (e.g. Mississippi, Rhine, Po, Weser, Danube, Schelde) are permanently or temporarily heated over very long stretches of their valleys with the estimated warming of up to 12◦C. In lakes the thermal pollution is much less recognized but the available data indicate that the surface water temperature increase due to discharges of cooling waters is in the order of 1 – 2 ◦C in Lake Stechlin (Germany; Kirillin et al., 2013), 0.3 ◦C in Lake Bienne (Switzerland; Vinnå et al. 2017), 2 – 6 ◦C in lakes Clinton, Newton and Coffeen (US; Mulhollem et al., 2016), 1.2 ◦C in Urias Lagoon (Mexico; Cardoso-Mohedano et al., 2015) and 0.5 – 6.3 ◦C in Lake Macquarie (Australia; Ingleton and Minn 2012). Thermal pollution possesses considerable environmental threats to the affected rivers/lakes because of its multifaceted negative impacts on hydrodynamics, nutrient cycles, water quality and biota (Davidson and Bradshow, 1967; Ingleton and McMinn, 2012; Kirillin et al., 2013; Madden et al., 2013; Cardoso-Mohedano et al., 2015; Raptis et al., 2017; Vinnå et al. 2017; Dziuba et al., 2020; Gashkina and Moiseenko, 2020). In lakes TP is thought to stabilize vertical stratification and consequently leads to deterioration of red-ox conditions in the near bottom water. Given that the energy demand is increasing globally and conventional energy sources are Fig. 1. Location and bathymetry of Lake Liche´ nskie. A) Location of L. Liche´ nskie and other lakes mentioned in the text. B) The lakes involved in the cooling system of the KPPP. C) Bathymetric map of L. Liche´ nskie and location of the data collection sites. LCz - L. Czarne, LD – L. Dębno, LGop - L. Gopło, LJ - L. Jasne, LKa - L. Kamionkowskie, LKi - L. Kierskie, LKo - L. Kortowskie, LLaz - L. Łazduny, LLD – L. Ł´ odzko-Dymaczewskie, LS - L. Suminko, LSar - L. Sarbsko, LSz – L. Szurpiły, LT - L. Trze´ sniowskie, LZab - L. ˙ Zabi´ nskie. M. Woszczyk and M. Brechbühler Journal of Hydrology: Regional Studies 62 (2025) 102917 2 still gaining importance, the TP and its consequences in aquatic systems will likely remain a concern-rising issue in the following decades. Despite the fact that the thermally polluted lakes can act as test sites for the research on some potential consequences of climate warming in aquatic systems and resilience of the latter to the temperature changes, the lakes have not been extensively studied so far. The current study provides a comprehensive analysis of Lake Liche´ nskie, the most heavily thermally polluted lake in Poland. Lake Liche´ nskie (LLi, central Poland), together with a few adjacent lakes (Lake Gosławskie, Lake Pątnowskie, Lake WąsoskoMikorzy´ nskie and Lake ´ Slesi´ nskie), for over 50 years has been involved in the cooling system of the Konin – Pątn´ ow power plants (KPPP). The hydrographic network in the vicinity of the KPPP has been modified so that the power plants could use water from the lakes for cooling electricity generators. Naturally existing lakes were connected (or reconnected) by a system of canals to enforce constant water circulation over a distance of 32 km, sufficiently long to ensure undisturbed supply of cold water to the power plants. To further enhance the heat transfer to the atmosphere, the cooling system is equipped with a few pumping stations, spillways and weirs localized between the connected lakes. The KPPP collect water from Lake Gosławskie and Lake Pątnowskie (Fig. 1) and discharge it to the canals, which distribute the effluents over the whole system of connected lakes. The disposal of cooling effluents has considerably transformed the Konin lakes. In case of LLi the discharge led to the increase in surface water temperatures (LSWT), reduction of spatial extent and duration of winter ice as well as disturbances in seasonal vertical water mixing and changes in chemical composition of water. Some authors claimed that the lake had shifted from dimictic regime towards monomixis (Socha and Hutorowicz, 2009; Stawecki et al., 2013) while others postulated that it had become polymictic (Brzozowska et al., 2007). In addition, owing to continuous throughflow of cooling water, the lake has a very high horizontal water exchange rate (water renewal time) which is currently estimated to 2 – 9 days (Stawecki et al., 2013). In addition, the emissions of CH 4 and N 2 O from the lake to the atmosphere are considerably low (Woszczyk and Schubert, 2023). The latter observation acts as an interesting aspect of the transformation of the system because anthropogenic impact usually leads to enhancement of GHG production in lakes. Last but not least, the TP of LLi has also facilitated the spread of invasive plant and animal species (Zdanowski et al., 2020) and enforced water organisms to develop adjustment strategies to living in a heated habitat (Dziuba et al., 2020). Owing to the above changes, LLi is found to be the most heavily human-altered lake in the system of the Konin lakes. Despite that LLi has been intensely used for economical purposes for decades, in the following years the human pressure on the lake is expected to increase because of the plans for setting up a 2.8 GW nuclear power plant (NPP) in Pątn´ ow. The NPPs are known to be more environmentally impactful than the coal-fired plants (Raptis et al., 2017). This new NPP will succeed the present day brown coal-fired Pątn´ ow PP (PPP) and will also use the system of Konin lakes as a source of cooling water. Therefore, before any management decisions regarding the lake are made, it is reasonable to assess the degree of transformation of this system and to identify possible environmental threats to the lake (if there are any) that may arise during further exploitation. In the current study we combined existing historical monitoring records of lake water parameters (collected between 1995 and 2023), spaceborne thermal images (from 2000 to 2025) and our own hydrochemical data from LLi to investigate long-term changes to the lake. Based on this we found that, unlike in other Polish lakes, during the last three decades the lake surface water temperatures in LLi have shown a decreasing trend. There was also a considerable increase in lake water salinity and a decline in alkalinity. From the data it thus follows that, to large degree, the lake processes have followed a different trajectory than in natural lakes prone to ongoing environmental changes related to global warming. At the same time, the changes to the lake can act as a manifestation of processes of adaptation of the system to declining anthropogenic pressure. Our results demonstrate that despite being heavily disrupted, lakes display considerable capacity to re-equilibrate with the ambient environment. We believe, that this study will contribute to better understanding of the functioning of aquatic systems under a strong human pressure but also will be helpful in developing strategies for sustainable management and protection of lakes. 2. Study area Lake Liche´ nskie (LLi) is a small (1.48 km 2 surface area; 7.47 ×10 6 m 3 water volume) and shallow (mean water depth 4.5 m; max depth 12.5 m) post-glacial lake located in the SE part of the Gniezno Lake District, central Poland (Fig. 1A). The lake is classified as moderately eutrophic (Pyka et al., 2007) and is known to be home for some invasive and thermophilus plants and fish such as Vallisneria spiralis and Pseudorasbora parva, respectively. LLi acts as the smallest in the system of the Konin lakes encompassing also Lake Gosławskie (LGo; 4.54 km 2 ; 13.8 ×10 6 m 3 ; 5 m max depth), Lake Pątnowskie (LPa; 2.82 km 2 ; 7.26 ×10 6 m 3 ; 5.5 m max depth), Lake Wąsosko-Mikorzy´ nskie (LWM; 2.53 km 2 ; 29.3 ×10 6 m 3 ; 36.5 m max depth) and Lake ´ Slesi´ nskie (LSl; 1.52 km 2 ; 11.6 ×10 6 m 3 ) (Fig. 1B). The lake catchment is primarily occupied by agriculture (60 %) and forests (22 %). Urban areas constitute c.a. 9 % of the catchment. Since 1960s the lakes have been involved in the cooling system of the Konin and Pątn´ ow electric power plants (KPPP). The cooling waters from the Pątn´ ow PP (PPP) are discharged to LGo and then, via the system of canals, flow to LLi and LWM. The temperatures of these waters vary between 13 and 15◦C in winter and 30◦C in summer (Fig. S1) and are on average 8.2 ±2.8◦C warmer than at the entrance to the cooling system. On its way to LLi, the PPP effluents meet the Konin PP (KPP) cooling waters. The inflow of cooling waters to LLi is from the south and the mouth of the inflow canal is in the middle part of the lake (Fig. 1C). Over the long term the discharge of the KPPP effluents received by LLi has declined and currently it is estimated to c.a. 22 m 3 ⋅s −1 . From LLi, the cooling waters, after 2 – 9 day-long retention, outflow to LPa (outflow 1; Fig. 1C) and/or to LSl (outflow 2; Fig. 1C). The former outflow is active throughout the whole year, while the latter is only open during summers. The water level in the Konin lakes is artificially kept constant at 84.90 – 85.10 m asl in LGo, 83.70 – 84.10 m asl in LLi and between 83.17 and 83.87 m asl in LPa, LWM and LSl. During dry periods the system of Konin lakes is supplied with water from the Warta river. M. Woszczyk and M. Brechbühler Journal of Hydrology: Regional Studies 62 (2025) 102917 3 However, there is also whole-year-round water input from the natural, albeit, anthropogenically transformed tributaries, the Struga Kleczewska and the Struga Biskupia (SB), in particular. These streams receive discharges of post-mining waters from the nearby brown coal mines and thus are important for shaping chemical composition of the Konin lakes. 3. Methods The study is based on self-collected hydrochemical observations, archival limnological data and satellite-based thermal data. 3.1. LLi water sampling and in situ measurements LLi water samples were taken monthly in three sites, L1, L2 and L3 (Fig. 1B), between March 2022 and February 2023. Locations of these sites were chosen so that they covered spatial variability of thermal pollution in the lake. L1 is located in the deepest point (12.5 m) in the SW part of the lake, L2 is located just in front of the mouth of the discharge canal and L3 is located in 8-m deep depression in the north part of the lake. The L1 site was previously monitored in 2014/15 (Woszczyk and Schubert, 2023). In all these sites we measured lake water temperature, conductivity, pH, and dissolved O 2 . The measurements were performed from the surface to the bottom with 1-m resolution using YSI Professional Plus probe, calibrated and checked with certified reference material (Harbour water, NWHAMIL-20.2) beforehand. The lake surface water was sampled for total P (P tot ) and alkalinity (A t ) and the samples were taken to PTFE containers from just below air-water interface. 3.2. Acquisition of archival limnological data For sake of comparison with the LLi data we also used multi-parameter records from several lakes in Poland (Fig. 1A). The data was taken from both openand closed-access databases. Albeit the data was collected in different periods, there was considerable overlapping of the datasets with monitoring of LLi. Historical limnological data from the Konin Lakes for 1995 – 2023 was obtained at the Department of Environmental Protection of the Zesp´ oł Elektrowni Pątn´ ow-Adam´ ow-Konin, Poland (ZEPAK; https://www.zepak.com.pl/en). The ZEPAK has monitored the physical-chemical parameters of the lakes since the opening of the KPPP in late 1960s. The observations (e.g. surface water temperatures, LSWT [◦C]; conductivity, κ [ μ S⋅cm −1 ]; pH; alkalinity, A T [mmol⋅L −1 ]; calcium, Ca 2+ [mg⋅L −1 ]; chlorides, Cl - [mg⋅L −1 ]; chemical oxygen demand, COD [mgO 2 ⋅L −1 ]; total P, P tot [mg⋅L −1 ]) were collected at L0 station (Fig. 1B) from surface waters with monthly resolution and using standard ISO methods (pH: PN-EN ISO 10523:2012; κ: PN:EN 27888:1999; A T : PN-EN ISO 9963–1:2001 +Ap1:2004; Ca 2+ : PN-ISO 6058:1999; COD: PN-EN ISO 8457:2001; P tot : PN-EN ISO 6878:2006 +Ap1:2010 +Ap2:2010 pt 7; Cl - : PN-EN ISO 10304–1:2009 +AC:2012). The data from L. Suminko (LS), L. Szurpiły (LSz) and L. Łazduny (LLaz) encompassing monthly records of SWT and A T for the period between October 2007 and May 2010 was obtained on request from prof. Wojciech Tylmann (University of Gda´ nsk, PL). A T data from L. Kierskie (LK) was obtained from prof. Karina Apolinarska (Adam Mickiewicz University, Pozna´ n, PL) (Apolinarska et al., 2020). The data on Cl - in L. ˙ Zabi´ nskie (LZab) for 2012 – 2022 is available at https://doi.org/10.34808/bsk4-eg58 (Tylmann et al., 2023). Monitoring data on A T and Cl - from L. Czarne (LCz), L. Kamionkowskie (LKa) and L. Kortowskie (LKo) was obtained on request at the Polish Chief Inspectorate of Environmental Protection (GIO´ S). The former two lakes are involved in the Integrated Environmental Monitoring (ZM´ SP) programme of the GIO´ S. Hydrochemical data from L. Ł´ odzko-Dymaczewskie (LLD), L. Dębno (LD) and L. Trze´ sniowskie are available at https://doi.pangaea.de/10.1594/PANGAEA.977483. 3.3. Collection of satellite-based thermal data Satellite-derived thermal data from LLi and adjacent lakes were obtained from the Landsat Level 2 Collection 2 (L2C2) archive (U.S. Geological Survey, 2020), covering the period from January 2000 to January 2025. Surface temperature products, derived through standard atmospheric correction procedures, were used directly. Initial scene selection was based on cloud coverage, retaining only images with less than 75 % total cloud cover. Subsequently, a minimum threshold of 10 % clear-sky water coverage over the study area was applied, resulting in a dataset of 693 thermal images acquired by the Landsat 7 Enhanced Thematic Mapper Plus (ETM+) (Earth Resources Observation and Science EROS Center, 2020a), Landsat 8 Thermal Infrared Sensor (TIRS), and Landsat 9 Thermal Infrared Sensor 2 (TIRS-2) (Earth Resources Observation and Science EROS Center, 2020b) processed data at 30 m spatial resolution, as provided in the L2C2 products. This resolution results from resampling coarser native thermal data (60 m for ETM+; 100 m for TIRS and TIRS-2) and is appropriate given the spatial scale of the study lakes, allowing meaningful analysis of intra-lake thermal variability. Satellite overpasses occurred at a mean acquisition time of 09:30 UTC ±20 min (1 σ ), corresponding to approximately 10:30 or 11:30 local time depending on the application of daylight saving time. Cloud masking was performed using CFMask (Foga et al., 2017) quality assessment flags, retaining only clear-sky water pixels. Pixels flagged as ice or exhibiting surface temperatures below 0 ◦C were assigned a value of 0 ◦C to account for frozen surface conditions. Where multiple acquisitions were available for the same day, mosaics were generated by prioritizing images with greater clear-sky water pixel coverage. The agreement between satellite-based estimates of LSWT and ground observation was assessed using same-day match-ups with available in situ measurements collected by local authorities and corresponding satellite observations. For each sampling site, satellitebased temperatures were extracted as the mean of a 3 ×3 pixel window (90 ×90 m) centered on the measurement location. M. Woszczyk and M. Brechbühler Journal of Hydrology: Regional Studies 62 (2025) 102917 4 To address data gaps and irregular temporal sampling, we employed a gap-filling approach combining lake-specific seasonal-trend decomposition and Empirical Orthogonal Function (EOF) analysis. Prior to decomposition, outlier filtering was applied using a Z-score threshold (|Z| >2.5) to remove residual land and cloud-contaminated pixels. For each lake, the corresponding time series was then decomposed into seasonal, trend, and residual components using Seasonal-Trend decomposition based on LOESS (STL; Cleveland et al., 1990), which extracts smoothed estimates of each component using a locally estimated scatterplot smoothing (LOESS) method based on local polynomial regressions. Spatial gaps in the residual component were addressed using the Data Interpolating Empirical Orthogonal Functions (DINEOF) method (Beckers and Rixen, 2003; Alvera-Azc´ arate et al., 2005), followed by linear temporal interpolation to address temporal gaps. The gap-free residuals were then recombined with the seasonal and trend components to reconstruct the complete, continuous daily lake surface water temperature (LSWT) time series for each lake. Reconstruction uncertainty was assessed using internal cross-validation within the DINEOF algorithm, applied to the residual component of the STL-decomposed temperature series. For each lake, 1 % of valid water pixels were randomly withheld, reconstructed, and compared with their original values. 3.4. Analytical methods Chemical composition of water was analyzed on the day of collection or the following day the latest. P tot was analyzed with cuvette photometric tests LCK349 by Hach-Lange. Unfiltered samples were first mineralized using LT200 thermostat (Hach-Lange) and then analyzed with DR 1900 spectrophotometer (Hach-Lange). The average recovery of P measurements, assessed on the basis of certified reference materials, was 90 ±3 % (m ±SD). Total alkalinity A T was determined by titration with 0.1 M HCl with regard to methyl orange. The average recovery of A T analyses was 105 ±2 % (m ±SD). 3.5. Calculations and statistical data treatment Stratification stability index (W S ; J⋅m −2 ) (Idso, 1973) acts as a measure of stability of vertical stratification of lake water column and is calculated as Ws=g Al∫ Zm 0 (z−z∗)( ρ z− ρ mean)Azdz (1) where g represents acceleration due to gravity (m⋅s −2 ), A l – lake surface area (m 2 ), A z – surface area at depth z (m 2 ), z – depth (m), z* - depth at ρ (m), ρ z – water density at z (kg⋅m −3 ) and ρ mean – mean lake water density in the water column (kg⋅m −3 ). Lake water density ρ z was calculated as a function of temperature, salinity and hydrostatic pressure according to Sun et al. (2008). The higher the W S , the stronger is the vertical thermal (and density) stratification of lake waters. Rates of removal of O 2 from the near bottom waters (mmol⋅m −2 ⋅d −1 ) were calculated from the slope of the linear regression line (mmol⋅d −1 ) relating the changes of the near-bottom water O 2 over time to the area of the near bottom water (m −2 ) (Matthews et al., 2005). Saturation index for calcite SI calc was calculated from activities of aCa 2+ [mol⋅kg −1 ] and aCO 3 2- [mol⋅kg −1 ] and solubility product K calc with the formula SI calc =(aCa 2+ ⋅aHCO 3 - )/K calc (2) The ion activities were obtained as described by Woszczyk et al. (2023) i.e. using ionic strength derived from measured conductivity κ. Statistical data treatment involved calculating Pearson’s correlation coefficients r (and corresponding probability values p) between variables, assessing statistical significance of the differences obtained with the Kruskall-Wallis test supported with Dunn’s post hoc test as well as determining statistical significance of temporal trends with Mann-Kendall test. The calculations were carried out with Past 4.09 (Hammer et al., 2001). 4. Results 4.1. LSWT in Lake Liche´ nskie and adjacent lakes To compare the temperature regime of the Konin lakes as well as to assess the differences between the latter group and the natural lakes in the area we used LSWT values extracted from the spaceborne thermal images from 2000 – 2025 instead of long-term in-situ instrumental records collected at L0 site (Fig. 1B) between 1995 and 2024. The major advantage of this approach was that it used spatially resolved temperature data for each lake which, given the considerable areal LSWT variability in the Konin lakes, enabled us to avoid poor representativeness of a single site-based monitoring for the whole lake systems. The validation based on same-day comparisons with in situ measurements yielded mean absolute errors (MAE) of 0.36 ◦C, 0.39 ◦C, and 0.44 ◦C for LGo, LLi, and LSl, respectively, with corresponding standard deviations of 0.41 ◦C to 0.56 ◦C (Fig. S2, Tab. S1). These errors are notably low given that the match-ups were uncoordinated and therefore subject to uncertainties arising from differences in observation time of day (i.e., diurnal cycle), skin effect, georeferencing accuracy, and measurement conditions such as sampling depth.The internal cross-validation M. Woszczyk and M. Brechbühler Journal of Hydrology: Regional Studies 62 (2025) 102917 5 indicates that the DINEOF-based LSWT gap reconstruction performs consistently across all lakes. Mean absolute error (MAE) values range from 0.35 ±0.41 ◦C to 0.44 ±0.56 ◦C (±1 σ ), reflecting low average gap reconstruction uncertainty in the LSWT fields (Tab. S2). This analysis showed that the mean daily LSWT in LLi during the last two and a half decade varied between 3.8 and 28.0◦C and that these values were statistically different from those in other Konin lakes, except LGo (Fig. S3; Tab. S2). At the same time, the LSWT differences between LGo and LPa as well as LWM were insignificant. Although thermally similar overall, LLi and LGo exhibited seasonal differences in LSWT patterns. During spring and summer (mid-April to mid-October) the LSWT values in LLi were up to 2.7◦C higher than in LGo, while from mid-October to mid-April the opposite was true (SWT in LGo was up to 2.1◦C warmer than in LLi) (Fig. 2). Except for LSl, the LSWTs in the Konin lakes for the period 2000 – 2025 were significantly higher than in L. Gopło, acting as a non-heated natural reference lake. The LSWT differences between LLi and LGop showed quite irregular annual pattern and ranged from 2.9 to 5.8◦C (on average 3.8◦C; Fig. 2). The spatial distribution of LSWT in LLi varied seasonally. As shown in Fig. 3, throughout the whole year the highest SWT values occurred adjacent to the mouth of the inflow canal (the L2 site). The differences in the mean daily LSWT between the L2 site and L1 and L3 stations, situated outside the plume, varied between 0.5 and 2.8◦C and were statistically significant. However, in winter (DJF) the spatial temperature gradients were stronger than in the summer (JJA) and the ΔSWT between L2 and L1/L3 sites were 1.82 – 2.78◦C (x=2.37◦C) and 1.58 – 2.56◦C (x=2.16◦C), respectively. In summer (JJA), the lake was more thermally homogenous with respective ΔLSWT values from 0.79 – 1.74◦C (x=1.30◦C) and 0.53 – 1.39◦C (x=0.93◦C). Despite the fact that the LSWTs in L1 and L3 were not significantly different, for the major part of the year the north part of LLi (L3 site) was warmer than the west part of the lake (L1 site) and the average ΔLSWT was 0.26 ±0.29 ◦C. Considerable role in protecting the lake from overheating was played by the outflow to Lake ´ Slesi´ nskie (outflow 2; Fig. 1C). The weir in the north part of LLi, directing the water to LSl, opened whenever the LSWT reached 30◦C thus leading to a rapid drop in temperature of up to 8◦C within a month (Fig. S4). The distribution of temperature in the lake water column showed that LLi was prone to a density stratification. Stratification buildup was between April and May and the break-up occurred between August and September, however in 2023 a weak stratification in the near bottom waters persisted until November (Fig. S5). The stratification stability index (W s ) values for LLi during the maximum stability reached 86.7 and 96.9 J⋅m −2 in 2014 and 2023 respectively. In L1 and L3 sites the stratification built up synchronously and a thermocline (Δ temperature/Δ depth of up to 7◦C/1 m) occurred between 1 and 3 m depth. In L2 the stratification developed more irregularly. There was no continuous period of summer stratification. Instead, stratification was observed occasionally in March and November, when the water column in L1 and L3 was homothermic. The thermocline was weak (up to 2.5◦C/1 m) and often occurred close to the lake surface (between 0 and 1 m depth). Over the long term, the LSWT in LLi displayed a statistically significant declining trend and the rate of change was −0.09◦C⋅y −1 (p=0.003) (Fig. 4). Significant tendencies were also obtained in LGo (-0.10◦C⋅y −1 ; p=0.008) and LPa (-0.07◦C⋅y −1 ; p=0.007), while the SWT in LWM and LSl remained unchanged (Fig. S6). At same time in LGop there occurred an increasing temperature trend at 0.07◦C⋅y −1 (p=0.008) (Fig. 4). 4.2. Hydrochemistry of Lake Liche´ nskie and adjacent lakes: present day conditions and long-term changes The mean annual A T in LLi between 1995 and 2023 ranged from 4.1 to 5.2 mmol⋅L −1 (Fig. S7). These values were significantly Fig. 2. Annual SWT cycle in L. Liche´ nskie (LLi; black), L. Gosławskie (LGo; red) and L. Gopło (LGop; grey) as well as SWT differences (ΔSWT) between LLi and LGop calculated on the basis of surface averaged temperature values for the period between 2000 and 2025. LLi – L. Liche´ nskie, LGo – L. Gosławskie, LSl – L. ´ Slesi´ nskie, LWM – L. Wąsosko-Mikorzy´ nskie, LPat – L. Pątnowskie. M. Woszczyk and M. Brechbühler Journal of Hydrology: Regional Studies 62 (2025) 102917 6 different from LGo and LSl but not from LWM and LPa (Tab. S3). The measurements conducted in 2022–23 in three sites in LLi showed only minor differences in A T throughout the lake. A T in LLi varied in an annual cycle between maximum in winter (January – March) and the lowest values in August and September (Fig. S8). Fig. 3. Long-term spatial SWT distribution in L. Liche´ nskie and adjacent lakes in winter (DJF) and summer (JJA). Fig. 4. Long-term SWT trends in L. Liche´ nskie (grey circles) and L. Gopło (black circles) between 2000 and 2025 as well as annual energy production in the KPPP between 1995 and 2023. The circles represent annual mean SWT for the lakes. M. Woszczyk and M. Brechbühler Journal of Hydrology: Regional Studies 62 (2025) 102917 7 Over the long-term, A T concentrations in all Konin lakes have displayed marked statistically significant declining trends (Fig. 5; Fig. S9). In LLi the A T has been decreasing at a rate of 26 µmol⋅L −1 ⋅y −1 . In LMW, LSl and LPa the rates were similar, while in LGo the decrease in A T pool was as high as 42 µmol⋅L −1 ⋅y −1 . The mean annual pH of the LLi waters was between 7.07 and 8.46 and these values were not significantly different from other Konin lakes (Fig. S10). In an annual cycle there was a maximum in spring while the lowest values occurred in October (Fig. S11). Since the mid-1990s the mean annual pH values in all Konin lakes have remained unchanged. The long-term median pH was 8.40. Throughout the last 25 years LLi surface waters have constantly showed highly positive SI calc values of 1.2 ±0.5 (x± σ ) implying persistent supersaturation with CaCO 3 in the epilimnion. The highest SI calc were in spring-early summer and the lowest in autumn (Fig. S12). The long term trend in SI calc was statistically insignificant. Mean annual values of chemical oxygen demand (COD), acting as a proxy for the concentration of O 2 -consuming chemical species (primarily organic matter) in lake water, varied between 5.4 and 8.9 mg O 2 ⋅L −1 (Fig. S13) and the differences between LLi and other Konin lakes were minor and statistically insignificant. The COD displayed a slight long-term increase, however the trend was not statistically significant. Mean annual Cl - concentrations in LLi waters for the period from 1995 to 2023 varying between 16.9 and 39.4 mg⋅L −1 were similar to the Cl - throughout the Konin lakes (15.9 – 40.3 mg⋅L −1 ) (Fig. S14). The Cl - showed only minor and insignificant annual variability. In long-term perspective, the Cl - concentrations in LLi displayed a significant increasing trend at a rate of 0.53 mgCl - ⋅L −1 ⋅y −1 (Fig. 6). The trend was stepwise, however, and in a recent decade, the Cl - increase has clearly accelerated. Compared to the LLi water chlorinity in the late 1990s (1995 – 2000), the Cl - concentrations has doubled in a time span of 20 years. The same tendencies were identified in other Konin lakes (Fig. S15). Total phosphorous (P tot ) concentrations in LLi varied between 58 and 175 µg⋅L −1 which fitted well the range of 38 – 182 µg⋅L −1 obtained in other Konin lakes (Fig. S16). Based on data from 2022 to 23, there were no spatial differences in P tot throughout the lake. On an annual basis, considerably higher P tot values were during winters (DJF) while the lowest values typically occurred in June (Fig. S17). Between 1995 and 2023 the P tot in LLi has increased substantially at an overall rate of 2.7 µg⋅L −1 ⋅y −1 , albeit the increasing trend was non-linear (Fig. 7). The statistically significant increase occurred in March, May and October – November (Tab. S4). 5. Discussion 5.1. The problem of thermal pollution of Lake Liche´ nskie Because LLi is one of the smallest lakes (in terms of volume) in the KPPP cooling system and since it has received a major part of thermal effluents discharged from the KPPP (30 % from the PPP and 54 % from KPPP; Stawecki et al., 2007) for over the last 60 years, it is regarded the most strongly thermally polluted in a group of Konin lakes. Hilbricht-Ilkowska and Zdanowski (1988) estimated long-term SWT increase in LLi for c.a. 5.2 – 12.9◦C, which was considerably higher than 3 – 7◦C warming in L. Gosławskie and L. Pątnowskie, 3 – 4◦C in L. ´ Slesi´ nskie and 1 – 2◦C in L. Wąsosko-Mikorzy´ nskie. Our results are consistent with the above pattern, albeit with some differences. First, the thermal pollution of LLi and LGo was similar, and the best estimate for the lake area-averaged warming was 3.81◦C and 3.64◦C, respectively. These values appear high compared to the LSWT increase of 0.8◦C and 0.23–0.32◦C in L. Stechlin and L. Biel, respectively, both heated by the effluents discharging from the local NPPs (Kirillin et al., 2013; Vinnå et al. 2017). Second, LSl, which displayed only minor and insignificant LSWT difference to LGop (on average 1.30 ±0.41◦C), seemed to be close to its natural thermal regime. Three, LPa and LWM showed a moderate degree of thermal pollution and the LSWT difference between these lakes and non-heated waters was 2.23 ±0.64◦C and 2.33 ±0.50◦C, respectively. Fig. 5. Long term A T trends in Lake Liche´ nskie between 1995 and 2023. The electricity production in the KPPP (in red) shown for comparison. M. Woszczyk and M. Brechbühler Journal of Hydrology: Regional Studies 62 (2025) 102917 8 In LLi, there was considerable spatial variability of thermal pollution from 5.30 ±0.70◦C in the plume zone to 3.70 ±0.55◦C – 3.44 ±0.54◦C in the north and west parts of the basin, respectively, and the warming varied irregularly throughout the year between 2.93 and 5.84◦C. In LGo the seasonal changes were stronger than in LLi. The maximum thermal pollution (6.49◦C) was in winter and the minimum (1.11◦C) – in summer. In winter LGo was thus warmer than LLi while in summer the opposite situation occurred. The Fig. 6. Long term Cl - trend in Lake Liche´ nskie between 1995 and 2023. The electricity production in the KPPP (in red) shown for comparison. Fig. 7. Long term P tot trend in Lake Liche´ nskie between 1995 and 2023. The electricity production in the KPPP (in red) shown for comparison. Table 1 Estimation of cooling time of KPPP thermal effluents in L. Gosławskie. Month h c ρ T 0 T 1 T env t c W⋅m −2 ⋅K −1 J⋅K −1 ⋅kg −1 kg⋅m −3 [◦C] [h] January 4191 24.2 1000 13.0 5.2 0.6 20 July 4174 24.2 1000 31.1 25.87 25.80 86 The time of cooling t c was calculated from the Newton’s law of cooling expressed with the formula: tc= ln(T0−Tenv T1−Tenv) k, where T 0 stands for the initial temperature of cooling waters, T 1 is the final temperature of cooling waters (equal to the SWT in LGo) and T env is the temperature of the ambient air. The coefficient k was derived as k=hAlake c• ρ •Vwater, where A lake stands for lake surface, V water is volume of thermal effluents discharging during one day, c stands for specific heat capacity, h is heat transfer coefficient and ρ is water density. M. Woszczyk and M. Brechbühler Journal of Hydrology: Regional Studies 62 (2025) 102917 9