Flood risk assessment of cultural heritage sites near lakes via advanced hydrodynamic modeling and digital technologies

Research paper Flood risk assessment of cultural heritage sites near lakes via advanced hydrodynamic modeling and digital technologies M.J. Alexopoulos a,* , T. Iliopoulou a , P. Mod´ e b , D. Istrati a,c , D. Koutsoyiannis a , S. Kr´ olewicz d , R. Graf d , L. Kaczmarek e , W. Rączkowski f a Department of Water Resources and Environmental Engineering, School of Civil Engineering, National Technical University of Athens, Athens, Greece b Institute of Structural Analysis and Antiseismic Research, School of Civil Engineering, National Technical University of Athens, Athens, Greece c Leichtweiß Institute for Hydraulic Engineering, Faculty of Architecture, Civil Engineering and Environmental Sciences, Technical University of Braunschweig, Germany d Institute of Physical Geography and Environmental Studies, Faculty of Geographic and Geological Sciences, Adam Mickiewicz University, Pozna´ n, Poland e Biological Spatial Information Laboratory, Faculty of Biology, Adam Mickiewicz University, Pozna´ n, Poland f Faculty of Archaeology, Adam Mickiewicz University, Pozna´ n, Poland ARTICLE INFO Keywords: Cultural heritage Flood risk HEC-RAS Computational hydraulics Rain-on-Grid ABSTRACT Cultural heritage flood risk assessments demand high-resolution, physics-based modeling frameworks capable of capturing complex terrain in diverse hydrological environments, and the subtle dynamics that threaten the relics. This study develops and applies such a framework at the lakeside archaeological site of Smuszewo, Poland, where flood hazards could arise from a combination of overland runoff and dynamic lake-stage fluctuations. To this end, we applied a high-resolution modeling approach combining overland terrain obtained via LiDAR and drone photogrammetry, and a digitized lake bathymetric model, within a HEC-RAS 2D Rain-on-Grid framework that enables detailed simulation of runoff–lake interactions and site-specific flood scenarios across design storms. We evaluate three scenarios: (i) model calibration using a five-day rainfall-stage event, (ii) the hydraulic impact of including versus omitting explicit lake bathymetry, across five design storms (1–50-year return periods); and (iii) a margin-to-failure analysis simulating lake-level rise from 0.00 to +1.50 m. Results show that omitting bathymetry underestimates peak flows for frequent storms due to artificial ponding on flat, dry-initiated surfaces (vs. realistic depths and wet cells enabling accurate volume propagation), whereas impact diminishes for the higher return periods. However, even under the worst-case scenario, an extreme storm on top of a +1.5 m lakestage rise, more than 1 m of freeboard remains. Our findings demonstrate the critical role of multiscale, highresolution terrain and physics-driven methods—and lay the groundwork for future digital-twin implementations, predictive maintenance strategies, and cloud-based simulations in cultural heritage flood-risk management. Secondary keywords Lake Bathymetry; Hydrodynamic Modeling 1. Introduction Flooding, identified as a costly natural peril globally [1], is not dominated by fluvial extremes alone; closed-basin and regulated lakes supply the hydraulic head that controls floodplain inundation [2]. Europe has not been spared: in 2024 the Copernicus European State of the Climate report registered the extensive inundation since 2013, with river discharges above the high-flow threshold along 30 % of the monitored network [3], while routed reconstructions of the last seven decades revealed an 11 % expansion in event footprints [4]. Trends of similar complexity demand flood-risk methodologies capable of resolving both overland runoff and dynamic stage fluctuations within lake basins. Lake-heritage interactions present a particularly underexplored dimension of this challenge. Scholarly attention to cultural heritage (CH) under hydro-climatic stress has evolved from a nascent thematic niche in 2003 to a multidisciplinary corpus spanning architecture, geoarchaeology, climatology, and disaster governance by the mid-2010s [5]. UNESCO’s early warning * Corresponding author. E-mail address: [email protected] (M.J. Alexopoulos). Contents lists available at ScienceDirect Results in Engineering journal homepage: www.sciencedirect.com/journal/results-in-engineering https://doi.org/10.1016/j.rineng.2025.107977 Received 28 July 2025; Received in revised form 22 October 2025; Accepted 29 October 2025 Results in Engineering 28 (2025) 107977 Available online 30 October 2025 2590-1230/© 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/ ). that water is “Europe’s most pervasive weathering agent” [6] presaged an era in which gradual forcings such as sea-level rise, permafrost decay, or desertification compound the frequency of acute floods [7]. Empirical studies verify such mechanisms: shifting hydrothermal regimes intensify winter decay of wooden ecclesiastical ensembles in southeastern Poland [8], and promote salt crystallization, corrosion, and structural fatigue in masonry [7]. Building on these insights, recent studies propose management frameworks [9], hazard mapping techniques [10], and loss quantification methods [11]. Yet, the lack of harmonized cross-national datasets [12] constraints quantitative indices [13,14]. The deficit is emphasized by ex-post analyses of 2022 floods in central Italy, which revealed systematic exposure misclassification for 14 damaged assets [15]. Despite this progress, existing CH flood-risk studies focus predominantly on riverine [16] or coastal hazards [17]. CH assets situated within or adjacent to closed lake basins, where both pluvial inputs and antecedent water levels govern inundation dynamics, remain largely unaddressed in the literature. The Late Bronze/Early Iron Age fortified settlement at Smuszewo, Poland, perched on an isthmus between two post-glacial lakes (Fig. 1), survives today because anaerobic lacustrine sediments inhibited decay. Yet this hydrological buffer is destabilized by both anthropogenic drainage and climatic perturbations. Hundreds of Polish and Central European heritage sites lie in similar environments—once waterlogged landscapes now at risk from climate change and human intervention—making Smuszewo an instructive case for assessing mid-19thcentury to present-day impacts and for modeling processes that affect still-preserved relics. The need to expand arable land (including meadows and pastures) encouraged farmers in the 19th and 20th centuries to drain the valley, which lowered the lake levels, exposing the wood and subjecting it to aerobic decomposition. Conversely, convective downpours drive ephemeral surges that raise the water level, while fertilizer-laden runoff from neighboring arable fields fuels eutrophication and biochemical attack. Recurrent stage oscillations thus may execute a dual attritional mechanism: mechanical fatigue of exposed wood and chemical alteration of buried assemblages. The archaeological oak structures preserved within the Smuszewo basin face deterioration when exposed to fluctuating moisture regimes. To understand this vulnerability, long-term observations [18] demonstrate that even temporarily drained waterlogged oak can suffer rapid biochemical decay and microbial colonization. The study documents that transition from anaerobic to aerobic conditions triggers oxidative degradation, loss of cellulose, and accelerated microbial attack, leading to structural disintegration of wooden relics [18]. Bj¨ ordal [19] further identifies soft rot fungi as key decomposers capable of thriving even at low oxygen concentrations, particularly under dynamic water tables where transient aeration allows colonization without complete drying. This implies that in the context of Smuszewo, stage oscillations of even ±0.25 m may not overtly inundate or expose the entire site, but still intermittently disturb the redox and microbial equilibrium. Wood that was once stabilized by permanent submersion may become vulnerable not from catastrophic flood but from a slow oscillatory regime that allows fungal establishment. Such conditions can be damaging for oak, whose apparent external preservation may mask internal structural weakening due to lignin-targeting degradation, as observed in multiple European lake-dwelling sites [19]. Effective conservation must therefore account not only for overt flood or drought risk, but also for seasonal or sub-seasonal exposure cycles that subtly breach the anaerobic preservation envelope. Hence, modeling must reproduce both short-term pluvial inputs and antecedent lake geometry to capture redox swings. To meet the dual requirements, we employ the Hydrologic Engineering Center’s River Analysis System (HEC-RAS) through its twodimensional Rain-on-Grid (RoG) solver, which imposes precipitation directly on computational cells and solves hydrologic and hydraulic equations in a unified framework [20–22]. RoG is adept at routing flow across complex micro-topographies and infrastructure [23]; yet its fidelity is highly sensitive to digital elevation model (DEM) resolution, with peak-stage attenuation and hydrograph flattening becoming conspicuous as grid spacing coarsens from 1 m to 5 m [24–26]. Recent benchmarking against TELEMAC-2D corroborates the method’s strengths—and highlights parameter sensitivities—when simulating short-duration mountain floods [27]. No study has yet examined in a structured way how the way lakes are modeled in RoG frameworks, especially whether bathymetry and initial water levels are included, affects predicted water stages for CH sites near lakes. The gap matters because lakes behave very differently from rivers: their large storage capacity slows changes in water level, but if their depth or initial water level is not set correctly, model results can yield misleading estimates of flooding potential. To address this gap, the present study delivers the first systematic evaluation of lake representation in a RoG framework for CH flood risk. It couples empirical calibration with scenario-based sensitivity analysis. First, we calibrate a high-resolution RoG model for the Czeszewo Lake Fig. 1. The Czeszewo Lake and its vicinity as located in Poland (A) and the setting of archaeological site between two lakes on the lower part of postglacial valley (B - oblique aerial photographs, 5.07.2014). M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 2 basin using a five-day event constrained by South-station rainfall and continuous lake-stage observations. Second, we examine lake representation by contrasting two configurations: a bathymetry-integrated DEM with stage-defined wet cells, and Airborne Laser Scanning (ALS) derived bare-earth DEM initiated dry except for measured water surfaces. Third, we investigate the potential for overtopping by deriving the lake-stage increment at which freeboard is exhausted. Section 2 details the geomorphic, hydro-meteorological, and land-cover datasets, including drone-based DEM acquisition and hyetograph synthesis. Section 3 outlines mesh construction, infiltration parameterization, calibration metrics, and the staged sensitivity workflow. Section 4 presents calibration performance, lake-scheme contrasts, and overtopping thresholds; Section 5 interprets uncertainties, methodological novelty, and generalizability to lake-fringed heritage sites. Concluding remarks distil actionable guidance for heritage flood-risk assessments and propose research extensions toward longer-term morphodynamic coupling. The ensuing analysis therefore delivers the first systematic evaluation of lake representation in a RoG framework for CH risk, coupling empirical calibration with scenario-based sensitivity to inform both engineering practice and conservation policy. 2. Study area and data 2.1. Site description The study site is a partly waterlogged fortified settlement dated to the Late Bronze Age/Early Iron Age. It was first exposed in 1863 during land-reclamation at the base of a post-glacial valley and the lake’s western shore. Lowered water levels revealed wooden breakwater structures in the peat–mineral valley fill [28,29]. Excavations from 1959 to 1966 uncovered remnants of wooden buildings and the base of a timber–earth rampart, analogous to larger finds at Biskupin (~25 km South-East; [30,31]). A 2004 geophysical survey mapped the settlement’s regular plan and detected charcoal from burned rampart sections [32] (Fig. 2). Aerial reconnaissance produced no new features, reflecting the subtle nature of the buried remains (Fig. 1B). Based on analysis of ceramics and metal artifacts, archaeologists initially dated the site to 500–400 BCE [33], but dendrochronology of four oak samples from the breakwaters indicates a mid-8th century BCE construction [34]. Wooden remains lie beneath ~1 m of sediment in the settlement’s interior, with preservation controlled by moisture and anaerobic conditions [18]. Digital Terrain Model (DTM) analysis on the western side shows a missing rampart section, likely eroded by higher lake waves before reclamation. 2.2. Rainfall and stage data The research used data on precipitation recorded at three nearby meteorological stations of the Institute of Meteorology and Water Management - National Research Institute (IMGW-PIB): Goła´ ncz (hereinafter referred to as Goła´ ncz), Chrząstowo (Chrzast) and Janowiec Wielkopolski (Janwlkp). In Poland, precipitation measurements are conducted using the standard IMGW-PIB measurement and observation network. Rainfall measurement data are directly available in databases on the https://hydro.imgw.pl/#/list/meteo and https://danepubliczne. imgw.plwebsites and on the MONITOR IMGW-PIB platform as operational data. For the sake of both clarity and brevity, we demonstrate the comparison of daily rainfall records for the years 2013, 2016, and 2018 (Fig. 3). For the purposes of the work, daily precipitation data from the period 1954–2024 were used for the Goła´ ncz station, 10-minute precipitation data for 2008–2024 for the Janwlkp station, and 10-minute precipitation and direction and additionally speed of wind more than 5, 10, 15 and 20 m/s for the Chrzast station from the years 2012–2024. These stations were considered crucial for conducting a hydrological analysis and assessing the potential impact of precipitation on the archaeological site in Smuszewo. They provide precipitation data and others meteorological data with different temporal and spatial ranges, enabling a comprehensive assessment of local and regional climatic conditions. The stations were selected due to their location in relation to the Smuszewo settlement. Goła´ ncz is located northwest of Smuszewo (approx. 9 km), hence it was considered a reference meteorological station, important for local analyses. It serves as the main source of base data. The Janowiec Wielkopolski station is located approx. 16 km to the south, while the furthest from it, to the north, is Chrząstowo (approx. 32 km). Both Fig. 2. Simplified plan of the settlement and preserved relics based on the interpretation of geophysical survey data [32]. Legend: 1 – long wooden houses divided into smaller ‘compartments’ (c. 6.5/7 ×7/7.5 m each) in which circular, stone hearths (2) were; 3 – main wooden paved street; 4 – side wooden streets; 5 – scatter of the burnt upper part of the rampart; 6 – breakwaters (?); 7 – gate (?); 8 – excavated area as visible at the HEXAGON image (16.02.1975). M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 3 stations can be a source of verification data, useful for validating regional data. The layout of stations around Smuszewo – north-west (Goła´ ncz), north (Chrząstowo), south (Janowiec Wlkp.) can enable the identification of local weather anomalies. The preliminary analysis allows us to assume that the precipitation data in the two stations Goła´ ncz and Janowiec Wlkp. are well correlated spatially. On the other hand, the Chrząstowo station shows differences, which is probably related to its considerable distance from Smuszewo and local anomalies in the environmentally specific Note´ c river valley, in the range of which the station is located. Janwlkp and Goła´ ncz correlate strongly and consistently (r≈0.84–0.86). By contrast, Chrzast (northeast) shows volatile correlations with both stations (Janwlkp-Chrzast: 0.43→0.88→0.58; ChrzastGoła´ ncz: 0.44→0.82→0.69). Consequently, Janwlkp serves as the subdaily rainfall proxy; Chrzast was excluded from calibration. The research also used data on the water surface elevation in Czeszewo Lake (Fig. 4–5), derived from direct field measurements. The lake is not covered by systematic water level measurements within the IMGW-PIB network. Local hydrological monitoring of water levels in Czeszewo Lake has been carried out since May 2024 as part of the TRIQUETRA project [35]. Regular measurements consist of automatic recording of water level changes in a continuous course using a ventilated water level recorder DL/N64 (STS, Switzerland), providing measurement of water column pressure with full compensation of atmospheric pressure thanks to the built-in capillary. The recorder is located in the eastern part of the lake basin (Fig. 4). The condition for correct recording of parameters was the appropriate installation of the device, ensuring a constant level of attachment of the above-water part of the recorder. This was achieved by installing the recorder in a casing pipe driven into the bottom of the water reservoir (difficult to access, reed zone). The device used allows for recording the pressure of the water column with an accuracy of 0.1 % (in practice, less than 3 mm). To control the stability of the assembly and the compliance of the obtained results with the actual water level, an underground control benchmark was installed at the lake shore, which is leveled to the national geodetic network of detailed elevation networks in the Amsterdam reference system using an optical level. The verification of the reference level of the benchmark was performed on the state benchmark of the detailed height control no. 61,881,625,035. H catalog benchmark is 99.86 m (Smuszewo, building no. 18, northern wall), H average of five Real Time Kinematics (RTK) measurements =99.87 m (PLEVRF2007-NH system - "Amsterdam"). The recorded data is read in the field by connecting the dataloggers Fig. 3. Daily precipitation hyetographs for Janwlkp, Chrzast, and Goła´ ncz stations in 2013, 2016, and 2018. M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 4 to a laptop and transferring the results. Lake stage starts at 88.41 m and declines quasi-linearly over the summer dry spell to 88.06 m on 2 November, therefore a 0.35 m drop in 165 days (Fig. 5). A late-autumn precipitation surge then raises stage by ≈0.21 m, which ends the year at 88.27 m. To assess long-term rainfall behavior, we analyzed extreme-precipitation indices at the Goła´ ncz gauge over 1956–2024. The precipitation time series analysis reveals characteristic patterns of temporal stationarity across multiple extreme precipitation indices. The distributional characteristics and temporal trends for all precipitation metrics are summarized in Table 1. Fig. 6 demonstrates the annual 95th-percentile daily totals, which evince an ordinary least squares slope of +0.011 mm yr⁻¹ (p =0.463), flanked by pronounced inter-annual volatility with a coefficient of variation of 21.9 % (Table 1). The temporal structure Fig. 4. Location of Water Surface Elevation (WSE) gauge and lake outlet. Fig. 5. Water surface elevation of Czeszewo Lake from May to December 2024, showing a gradual decline through late autumn followed by a rise in December. M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 5 exhibits a range of 12.2 mm, spanning from 5.1 to 17.1 mm over the 69year observational epoch. Fig. 7 shows a trend of –0.088 mm yr⁻¹ (p =0.320), skewness γ = 1.524, and excess kurtosis κ =3.076. The 82.3 mm range reflects extreme convective events in 1977 and 2016. Fig. 8 demonstrates effectively trend-less behavior (+0.003 d yr⁻¹, p =0.734). The temporal distribution exhibits a mean frequency of 2.4 events per year with a dispersion ratio (variance/mean) of 0.993, approximating unity and confirming consistency with Poissondistributed occurrences. The discrete range spans 0 to 6 events annually, with apparent clustering during specific decadal periods. Collectively, the station record is consistent with the null hypothesis of stationarity, in agreement with Pi´ nskwar et al. [36]. 2.3. Synthetic hyetographs Synthetic Hyetographs of 10 min temporal resolution were constructed based on local Intensity-Duration-Frequency (IDF) curves for the Smuszewo location (52◦53 ′ 34.48 ″ N/ 17◦23 ′ 33.74 ″ E) (Fig. 9) using the Alternating Blocks method [37] for 50-, 10-, 5-, 2and 1-year return periods (Fig. 10). The work uses data from the Polish Atlas of Rain Intensity PANDa (https://atlaspanda.pl) regarding the intensity of reliable rainfall with a specified duration and exceedance probability p=2 % and 10 % (data acquisition certificate 0325/1099,929/3), 20 % and 50 % (data acquisition certificate 0325/2795,467/3) and 99 % and 100 % (data acquisition certificate 0325/7953,534/3). The range of rainfall durations was t=5–4320 min. The calculations used the average rainfall intensity dm 3 /(s•ha) within the provided confidence intervals. Table 1 Summary statistics and distributional parameters for annual precipitation extremes at Goła´ ncz station (1956–2024). Metric Mean Range CV ( %) Skewness Kurtosis (excess) Dispersion Ratio (Variance/Mean) Trend (mm/yr or d/yr) p-value 95th percentile (mm) 11.4 12.2 21.9 — — — +0.011 0.463 Annual maximum (mm) —82.3 —1.524 3.076 —-0.088 0.320 Heavy days (≥20 mm) 2.4 6 — — — 0.993 +0.003 0.734 Fig. 6. Annual 95th percentile daily precipitation at Goła´ ncz station (1956–2024). Fig. 7. Annual maximum daily precipitation at Goła´ ncz station (1956–2024). M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 6 Atlas (PANDa) is an online platform containing information on reliable rainfall intensities for all cities in Poland. The atlas in digital format contains a catalog of nearly 13,000 local precipitation models developed for each mesh (with a resolution of 5 km) of the regular grid dividing the area of Poland [38]. 2.4. Terrain and bathymetry We downloaded 1 m DTM and Digital Surface Model (DSM) tiles (ASC, EPSG:2180) from Poland’s Geoportal using QGIS’s Data Downloader, covering the Czeszewo Lake and its catchment. Source data are 2014 airborne LiDAR [39,40]. Tiles were reprojected to UTM 33 N GeoTIFF and median-filtered (11 ×11) to remove interpolation artifacts. As part of the task to obtain a highly accurate DTM and DSM, a photogrammetric mission was carried out using the Yuneec H520 RTK drone equipped with a 20-megapixel E90X camera. Flights were conducted at an altitude of 120 m above ground level, which allowed for a ground sampling distance of approximately 3.5 cm. The mission was planned in a grid flight pattern, using an 85 % forward overlap and 75 % side overlap between images. Data acquisition was performed under favorable conditions with clear skies, no wind, and stable lighting, in order to minimize shadows and image distortions. The drone’s positioning relied on the RTK system, with real-time corrections delivered via a cellular network. By doing so, we achieved centimeter-level geolocation accuracy without the need for ground control points. The collected images were processed using photogrammetric software Pix4Dmapper. The workflow included importing the images, aligning them, generating a dense point cloud, and creating the DSM. To derive the DTM, the point cloud was filtered to remove surface elements such as vegetation. A high-resolution orthophoto was also generated based on the processed data. The final DTM and DSM models underwent accuracy assessment. Position verification was based on RTK data, and the resulting root mean Fig. 8. Annual count of heavy precipitation days (≥20 mm) at Goła´ ncz station (1956–2024). Fig. 9. IDF curves for Smuszewo (52◦53 ′ 34.48 ″ N, 17◦23 ′ 33.74 ″ E) showing rainfall intensity (dm 3 /(s•ha)) versus event duration (5–4 320 min) at exceedance probabilities of 2 %, 10 %, 20 %, 50 %, 99 % and 100 %. M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 7 square errors (RMS) ranged between 2–3 cm in both horizontal and vertical dimensions, confirming the high quality and precision of the deliverables. The data were exported in GeoTIFF and LAS formats, which allowed further processing in a Geographic Information System (GIS) environment. Originally developed by the Inland Fisheries Institute (IR´ S) (Fig. 11), First, the scanned bathymetric plan (Fig. 11a) was georeferenced using distinctive landscape features such as the shoreline and the mouths of inflowing streams. Next, the isobaths were digitized and overlaid on a contemporary orthophoto map, allowing for spatial verification of their accuracy (Fig. 11b). A raster bathymetric model of the lakebed was generated through morphology interpolation techniques (Fig. 11c). Fig. 10. Synthetic hyetographs of 10 min temporal resolution constructed using the Alternating Blocks method for 50-, 10-, 5-, 2and 1-year return periods for a 12 h event. Fig. 11. Conversion of the analog bathymetric plan by the IR´ S into digital format: a) georeferenced scan of the bathymetric plan based on distinctive features, b) digitized isobaths overlaid on a contemporary orthophoto map, c) lakebed model of Czeszewo Lake obtained through interpolation and profile lines, d) depth profile. M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 8 Additionally, a depth profile was created to illustrate the variation in lake depth along a selected transect (Fig. 11d). The DTM and DSM were integrated with a historical bathymetric map from 1961 [41]. The final continuous elevation model is created by merging the bathymetric data with the DTM/DSM using a custom script that harmonized raster resolution and spatial extent. The isobaths were vectorized and assigned elevation values relative to the lake’s mean water level (88.2 m a.s.l.) in the Polish vertical reference system KRON86. A continuous raster bathymetric model was then generated using surface-fitting techniques based on the contour lines, by applying morphological functions at a resolution consistent with the elevation model (Fig. 12). The original vertical and planimetric accuracy of source lidar data is estimated at 0.15 m. 3. Methodology This study employed a multi-tiered approach to systematically assess flood vulnerability at lakeside CH sites, with the overall methodological workflow illustrated in Fig. 13. 3.1. Rain-On-Grid model setup 3.1.1. Hydrodynamic solver and stability The Hydrologic Engineering Center’s River Analysis System (HECRAS) has matured into the hydraulics community’s open-source workhorse, courtesy of the U.S. Army Corps of Engineers’ steady iteration and transparent numerical kernels [42]. Its most consequential and somewhat recent upgrade is the two-dimensional RoG capability, by which observed or remotely sensed precipitation is applied directly to computational cells, thereby collapsing the traditional hydrologic–hydraulic partition and permitting runoff generation, flow routing, and flood-wave propagation to be resolved within a single engine [20,21]. Within the 2D module, users may invoke either the parabolic Diffusion Wave scheme or one of two manifestations of the depthaveraged Shallow Water Equations (SWE): the hybrid Eulerian–Lagrangian formulation (SWE-ELM) and the fully Eulerian finitedifference solver (SWE-EM). The latter tackles the conservative form of the SWE—valid where horizontal length scales dwarf the flow depth—through an upwind discretisation that tracks the temporal evolution of water depth, h, and the depth-averaged velocity vector, V: ∂ h ∂ t+ ∇(Vh) = q(1) ∂ V ∂ t+ (V∇)V= − g∇(h+z) + vt∇2V−cfV(2) where q represents source–sink terms such as rainfall and infiltration, g is gravitational acceleration, z is bed elevation, ν t is the horizontal eddyviscosity coefficient, and c f embodies bed-friction losses. Neglecting the advective and viscous terms simplifies (2) to the classic diffusion-wave momentum balance, ∂ V ∂ t+g∇(h+z) = gSf(3) which is computationally cheaper but less faithful when supercritical flows or rapid transients are present. All schemes are executed over a structured mesh with boundary conditions prescribed along basin limits. To forestall numerical oscillations, the user-selected time step, Δt, must satisfy the Courant–Friedrichs–Lewy (CFL) criterion, Cr =uΔt Δs(4) where u is the characteristic wave speed and Δs the representative cell dimension. Although SWE-EM is an explicit solution scheme that requires Courant numbers to remain at or below unity for stability [43], practitioners generally iterate toward 1—which adapts both grid resolution and time step to local topographic gradients and flow regimes—to secure a stable, non-diffusive solution. 3.1.2. Basin area The basin area of Czeszewo Lake was delineated through manual vectorization of basin boundaries based on publicly available data provided via a Web Map Service. The MPHP10k dataset defines the boundaries of administrative units used in Polish water management, including river basin districts (Pa´ nstwowe Gospodarstwo Wodne Wody Polskie), and is distributed under the CC BY 4.0 license, which permits data processing. The MPHP10k data were loaded into a GIS environment using the WMS service available at: https://mapy.isok.gov.pl/wms. Based on the “Basic Basins” layer, the boundary of the Czeszewo Lake basin was manually vectorized for topological accuracy using QGIS editing tools. The delineated boundary was validated against reference layers, including a DTM and the hydrographic network. Where necessary, adjustments were made to the boundary, particularly in areas with complex terrain morphology. The final basin boundary was exported to the shapefile format and reprojected to the UTM Zone 33 N coordinate system (EPSG:32,633), from the original PUWG 1992 system (EPSG:2180), in accordance with national GIS standards. 3.1.3. Mesh and boundary conditions We used an unstructured 10 ×10 m mesh to balance computational speed with sufficient resolution of flow dynamics and water-surface Fig. 12. DTM (left) and DSM (right) integrated with the bathymetry of Czeszewo Lake; basin area is presented with red solid line. M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 9 precipitation to overland flow in one coherent 2D model. This holistic strategy has been advocated in recent years as computing power and data availability have improved. By applying rainfall uniformly across the mesh and letting the shallow-water equations determine runoff paths, we inherently capture micro-topographic effects that traditional lumped runoff models or 1D river models would miss. Our calibration, achieving an RMSE on lake stage (~0.06 m) on par with the gauge accuracy, attests to the viability of RoG even in a data-sparse context. Under all simulated scenarios, the relics remain unflooded. Neither the calibrated July 2024 event nor any of the 1–50-year design storms produce water levels high enough to overtop the fortification’s crest. In fact, even a hypothetically extreme lake level rise of +1.5 m that far exceeds observed inter-annual fluctuations, leaves roughly one meter of freeboard intact (Fig. 19). Pluvial flooding alone is very unlikely to endanger the cultural asset in its current state. The site’s topographic setting (a raised isthmus and earthwork) affords substantial natural protection, which decouples it from all but the most extraordinary flood conditions. It is worth contrasting this with many cultural heritage sites documented in literature that face recurrent flood threats. For example, coastal UNESCO sites in the Mediterranean are already seeing frequent inundation; Reimann et al. [71] mention that 37 of 49 low-lying coastal heritage sites are at risk from a present-day 100-year flood. By comparison, our inland case study appears secure against floods up to the 50-year level. This shifts concern from acute inundation to subtler stresses—moisture cycling, eutrophication, and decay. Practically, stakeholders could shift from costly flood defenses to monitoring long-term lake levels and water quality, which pose greater risks. It is also instructive to examine why no hazardous flooding occurs in our scenarios. The combination of gentle basin slopes (0.2–0.8 %) and large storage volume in the lake yields a highly attenuating system. Rainfall of even 50-year intensity is mostly absorbed as lake level rise (~0.3–0.4 m in the worst case) and mild outflow, rather than rapid overland flow. Flow velocities remain very low over the archaeological ridge (≲0.05 m s⁻¹, Fig. 17), and depth–velocity products never exceeded the lowest hazard class (H1) even at the inundation fringes. Essentially, the lake functions as a detention basin by temporarily storing inflowing stormwater within its expansive volume and gentle topography. It is gradually releasing it through the narrow outlet at controlled rates, and minimizes overtopping risks at the adjacent heritage site. The finding aligns with the broader understanding that “bathtub” models (simple filling of depressions) can seriously mislead if used in place of physicsbased simulation [72]. In reality, the lake did not behave like a static bathtub with instant uniform filling; instead, the dynamic model captures the gradual expansion of a wetting front and the timing of overflow. Sanders et al. [73] emphasize this distinction, noting that simplistic static inundation approaches consistently underperform relative to full hydrodynamic models. Our results exemplify that point: a static approach might assume a uniform lake level rise translates immediately to flood depth against the earthwork, whereas the 2D simulation shows that any overflow is delayed and diminished by the lake’s own capacity and outlet flow. This outcome is encouraging for site preservation – it indicates a considerable inherent resilience. Regional trends suggest mean lake stages may rise by a few cm per decade. To stress-test the system, we imposed incremental lake level rises in Block C simulations. The roughly linear freeboard decline (~40 cm lost per +25 cm stage) we found implies that even under a compounded scenario of a 50-year storm following a +0.5 m climatic lake rise, the fortification would still not flood. This margin-to-failure gives confidence that the potential of moderate future precipitation increases alone is unlikely to tip the balance. Only under highly exaggerated conditions (+1.5 m lake rise, beyond credible mid-century predictions) did the moat around the mound reach ~0.3 m depth – still below any structural threshold of concern. Such a conservative analysis suggests that the primary climate-related threat to the site is not outright inundation but rather the more gradual effects of moisture regime shifts (e.g. more frequent soil saturation could accelerate rot in wooden structures just above the water table, even if those structures are never submerged in a flood). In short, our flood hazard model signals “all clear” for overtopping risk in the coming decades; however, it should be integrated with broader vulnerability assessments that include material weathering under increased humidity or human factors like land-use change in the basin. While the model confirms that surface inundation thresholds are not breached under any tested scenario, the subsurface zone adjacent to the monument becomes increasingly saturated with each stage increment. Drawing on experimental decay studies at Biskupin, such oscillatory conditions at shallow burial depths, especially those hovering near the water table, may enable microbial colonization and trigger rapid deterioration of archaeological oak, even without direct exposure to surface flows. Thus, the absence of overt flooding does not equate to preservation safety. To quantify those risks, a 10-year burial experiment on archaeological oak [18], mass loss and biochemical degradation varied sharply with depth: shallow burial at 25 cm, particularly in mineral soils, corresponds to high-risk preservation zones. At 25 cm depth, waterlogged oak is saturated only 2–3 months each spring, which leaves extended aerobic exposure the rest of the year. The conditions lead to severe degradation: mass loss approaches 76 %, basic wood density decreases by nearly 67 % (leaving only 33.1 % residual density), and maximum water content increases by 328.3 % [18]. Volume loss reaches approximately 27 %, which further translates into a surface layer erosion of around 0.7 mm. Microscopic inspection confirms extensive deterioration of secondary cell wall layers, with hemicelluloses and cellulose largely depleted, leaving thinner cell walls. In contrast, oak samples buried at an intermediate depth of 50 cm in peat display moderate degradation. These specimens are subject to seasonal water table fluctuations, remaining submerged for roughly half the year. Although peat harbors more microbes than mineral soil, oak at 50 cm depth shows markedly better preservation than in shallow layers. Water content increases by 39.4 %, and porosity rises by 12.6 % [18], but basic density declines by only 19.8 %. The metrics suggest that intermediate burial layers form a transitional preservation regime that is sensitive to both groundwater dynamics and substrate chemistry. Oak buried at 100 cm, either in waterlogged peat or in water-filled trenches, shows optimal preservation. Conditions like these may mirror the anaerobic environments historically responsible for preserving sites like Biskupin and, potentially, Smuszewo. At this depth, mass loss is limited to just 8–10 % over the decade-long trial, with maximum water content increases ranging from 11.0–16.5 %, and density decreases constrained to 6.4–9.1 %. Volume loss is negligible (~2 %), and indicates a stable microenvironment conducive to long-term organic preservation. Redox potential measurements consistently remain below −100 mV, a threshold widely recognized as protective for buried wood. Chemical analyses confirm degradation varies with burial depth. In severely degraded shallow samples, holocellulose content falls from 356.0 kg/m³ in control samples to just 63.6 kg/m³, a more than five-fold reduction. Cellulose declines similarly—from 210.2 kg/m³ to 36.6 kg/ m³—indicating a near-total loss of structural polysaccharides [18]. By comparison, well-preserved samples from 100 cm retain 14–19 % of their original holocellulose and 14–16 % of cellulose. Lignin, while more resilient, still shows a notable decrease, with residual content dropping from ~90 % in deep burial conditions to as low as 45.4 % in shallow degraded wood. The holocellulose-to-lignin ratio fell at all depths, with the steepest drop at 25 cm. Complementary trials using contemporary, non-degraded oak buried at the same site emphasize the fragility of archaeological material. In both wet peat and submerged trench environments, fresh wood exhibits only 3.6–3.8 % mass loss over 10 years. Degradation is largely restricted to hemicellulose components, with minimal shifts in cellulose and lignin. Fourier Transform Infrared spectroscopy confirms the biochemical pathway: attenuation of the 1730 cm⁻¹ absorption band, associated with acetyl and carbonyl groups in xylan, identifies hemicellulose as the primary target of decay. In contrast, archaeological oak degrades 2.5 × M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 16 faster under the same conditions, likely due to loss of protective extractives. It is therefore emphasized that archaeological oak, even when reburied under seemingly favorable conditions, remains highly vulnerable if its biochemical resilience has already been compromised. Finally, this study illustrates a replicable framework for cultural heritage flood risk assessment that capitalizes on emerging technologies. The use of drone-derived topography allowed us to model the terrain with a high resolution, and allows us to capture the subtle basin morphology that coarser datasets would miss. Trepekli et al. [74] have shown that such ultra-high-resolution elevation models can improve flood predictions as their experiments in urban Ghana reveal that using a 0.3 m-resolution UAV LiDAR DTM reduces flow prediction errors by up to 62.5 % in flat areas compared to a 10 m DEM. Our work adds evidence from a different environment (a natural lake and archaeological mound) that local-scale surveying is well worth the effort for CH sites. Without the drone survey, precise hydraulic modeling would not have been possible; any analysis based on off-the-shelf terrain data (e.g. 5–10 m with water surfaces unpenetrated) would have defaulted to something like our bare DEM case. By representing actual bathymetry, we ensure the model’s physics—and its “no hazard” outcome—are credible. The framework also illustrates the importance of multi-scenario sensitivity testing. Rather than a single deterministic statement (“site will not flood for X-year event”), we have mapped a continuum of lake-level scenarios and storm return periods. Besides aligning with UNESCO guidelines for resilient heritage management [6], a similar approach could provide heritage managers a richer understanding: e.g., even under worst-case lake levels, a 50-year storm leaves ~1 m freeboard, and conversely, it would take an unprecedented combination of factors to cause flooding. Under compound climate stresses, nuanced, site-specific risk appraisals are essential. They allow for informed prioritization – sites that emerge as low risk for floods (like Smuszewo in pluvial terms) can focus on other issues, whereas truly flood-prone sites would warrant immediate adaptation measures. However, the model implementation in the present study contains three principal limitations that warrant consideration. Event-based calibration, adopted to reduce computational demands, introduces parametric uncertainty through event-dependent parameter optimization [75], leading to divergent parameter sets across calibration events that constrain transferability beyond the calibration domain [76]. Hereby, the RoG configuration assumes spatially uniform precipitation, which constitutes a simplification justified by the basin’s modest extent (~23 km²); this assumption would require reconsideration for larger catchments where distributed rainfall representation becomes necessary to capture spatial variability effects on runoff generation mechanisms [77]. Topographic data quality exerts governing influence on flood model accuracy, as systematic errors from DEM acquisition methods propagate through the hydrologic-hydraulic modeling chain, affecting derived terrain parameters and subsequent hydraulic simulations [78]. The divergent risks identified in this study compel decision-makers to prioritize management strategies based on nuanced risk classification, requiring a critical trade-off assessment. Given the site’s considerable inherent resilience against pluvial overtopping, the marginal utility of investing in costly structural flood defenses is low. Conversely, the finding that subsurface saturation drives severe material degradation, exemplified by the rapid decay of archaeological oak near the water table (up to 76 % mass loss at 25 cm depth), indicates that the primary threat is hydrological stability, not hydrodynamic force. Therefore, managers must execute a strategic pivot: shifting resources away from traditional flood defenses toward adaptive soil moisture management [79]. This decision-making process inherently involves balancing competing objectives under uncertainty, particularly between the need for model precision, as high-resolution bathymetry and computational intensity guarantee higher accuracy, and the constraints of available budgetary and technical resources [80,81]. The optimal preservation strategy is thus defined not by eliminating surface risk, but by implementing multi-objective frameworks that minimize the long-term, irreversible material decay caused by moisture regime shifts [82,83]. 6. Conclusions In this work, we aimed to resolve lake-flood dynamics at high spatiotemporal resolution in order to assess the flood risk at one’s of Poland’s main cultural heritage sites. To this aim, we implemented a two‑dimensional RoG methodology in HEC‑RAS by combining drone‑photogrammetry‑derived overland terrain (via a Yuneec H520 RTK drone equipped with a 20-megapixel E90X camera) with a georeferenced, interpolated bathymetric model from the IR´ S plan. This novel integration leveraged ALS for basin-wide topography, drone-derived digital terrain models with 2–3 cm RMS accuracy, and a ventilated DL/N64 sensor for real-time lake-stage monitoring with 0.1 % precision. We applied this workflow to Czeszewo Lake in Smuszewo, Poland, assessing flood hazard across design storms (1–50‑year return periods) and systematic lake‑level increments (0–1.50 m), informed by continuous highaccuracy measurements (0.1 % precision, <3 mm in practice) from the water level recorder stabilized via a leveled underground benchmark. The RoG framework employed a structured mesh with CourantFriedrichs-Lewy stability criteria, boundary conditions along basin limits, and land-cover-based hydraulic parameterization (Manning’s n: 0.025–0.100; Curve Numbers: 71–100). The resulting margin‑to‑failure curve provides a concise, operational metric for monitoring potential overtopping at heritage sites. Despite recent advances in hydraulic modeling, the treatment of lake bathymetry and initial water surface elevation in RoG meshes has remained largely ad hoc, with limited understanding of their effects on overtopping thresholds. Accurate representation of lake bathymetry is the primary determinant of hydraulic fidelity in RoG simulations for lake-fringed floodplains. Introducing the 8 m basin depth increased peak discharges by an order of magnitude and advanced flood arrival times relative to a flat-bed surrogate, demonstrating that simplified terrain misrepresents both flow magnitude and timing for design-level storms due to artificial ponding on flat, dry-initiated surfaces (vs. realistic depths and wet cells enabling accurate volume propagation). Bathymetry omission biased flood extents by 15–20 % and underestimated peaks by up to 55x in moderate events, though differences diminished in 50-year extremes as basins saturated. Velocities stayed mild, with all areas rated safe (H1 class) even at peak stages. Despite this sensitivity, all scenarios—including a 50-year storm imposed on a +1.5 m antecedent stage—maintained at least 1 m of freeboard at the Smuszewo fortification, which confirms that available storage, rather than statistical storm rarity, may govern present overtopping risk. The linear decline of freeboard (~0.40 m per 0.25 m stage rise) provides a tractable metric for operational monitoring. Methodologically, the combination of digital technologies (e.g. drones), high-resolution digital elevation models and advanced RoG hydraulic modeling offers a replicable protocol; omitting submerged geometry leads to systematic underestimation of hazard for moderate-return events, the regime most relevant to routine conservation planning. This resilience decouples the site from acute pluvial threats, redirecting focus to subtler risks like moisture cycling and microbial decay in waterlogged oaks, as observed in longterm studies. Future heritage risk assessments should therefore treat bathymetric acquisition as a non-negotiable input especially when moderatefrequency flood scenarios are of interest in order to properly evaluate site vulnerability. Overall, our proposed work illustrates the potential of multiscale data and physics-based RoG modeling, laying groundwork for emerging applications like digital twins and cloud-based simulations in heritage flood management. CRediT authorship contribution statement M.J. Alexopoulos: Conceptualization, Formal analysis, M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 17 Investigation, Methodology, Visualization, Writing – original draft, Data curation. T. Iliopoulou: Conceptualization, Formal analysis, Investigation, Methodology, Writing – review & editing. P. Mod´ e: Formal analysis, Writing – original draft. D. Istrati: Conceptualization, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – review & editing. D. Koutsoyiannis: Supervision, Writing – review & editing. S. Kr´ olewicz: Data curation, Methodology, Methodology, Resources, Visualization, Writing – original draft. R. Graf: Data curation, Methodology, Resources, Visualization, Writing – original draft. L. Kaczmarek: Data curation, Methodology, Resources. W. Rączkowski: Data curation, Methodology, Resources, Visualization, Writing – original draft. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This work is based on procedures and tasks implemented within the project “Toolbox for assessing and mitigating Climate Change risks and natural hazards threatening cultural heritage – TRIQUETRA”, which is a Project funded by the EU HE research and innovation programme under GA No. 101094818. https://triquetra-project.eu/ Project made using data from the Polish Atlas of Rainfall Intensities PANDa. Data availability Data will be made available on request. References [1] Aon. (2024). Climate and catastrophe insight. https://ppl-ai-file-upload.s3.ama zonaws.com/web/direct-files/attachments/991733/21b68a16-0bc2-4346-a95 e-52dd6d5907d1/climate-and-catastrophe-insights-report.pdf. [2] P.E. Todhunter, B.C. Rundquist, Flood damage assessment and survey of mitigation efforts at Stump Lake, North Dakota, A Study Closed-basin Lake Flood (2003). https://hazards.colorado.edu/uploads/basicpage/qr164.pdf. [3] Copernicus Climate Change Service (C3S), & World Meteorological Organization (WMO), European State of the Climate 2024, 2025, https://doi.org/10.24381/ 14j9-s541. [4] B. Fang, E. Bevacqua, O. Rakovec, J. Zscheischler, An increase in the spatial extent of European floods over the last 70 years, Hydrol. Earth. Syst. Sci. 28 (16) (2024) 3755–3775, https://doi.org/10.5194/HESS-28-3755-2024. [5] S. Fatori´ c, E. Seekamp, Are cultural heritage and resources threatened by climate change? A systematic literature review, Clim. Change 142 (1–2) (2017) 227–254, https://doi.org/10.1007/S10584-017-1929-9. [6] UNESCO, Vulnerability of Cultural Heritage to Climate Change, Council of Europe, Strasbourg, 2009. [7] E. Sesana, A.S. Gagnon, C. Ciantelli, J.A. Cassar, J.J. Hughes, Climate change impacts on cultural heritage: A literature review. Wiley Interdisciplinary Reviews, Clim. Change 12 (4) (2021) e710, https://doi.org/10.1002/WCC.710. [8] H. Hu, Impact of climatic-meteorological conditions on Polish wooden cultural heritage: an example of world heritage sites featuring wooden churches near Krakow, Human. Soc. Sci. Commun. 11 (1) (2024) 1–13, https://doi.org/10.1057/ s41599-024-03854-0. [9] S.A. Orr, J. Richards, S. Fatori´ c, Climate Change and Cultural Heritage: A systematic Literature Review (2016–2020). The historic Environment, Policy & Practice 12 (3–4) (2021) 434–477, https://doi.org/10.1080/ 17567505.2021.1957264. [10] R. Figueiredo, X. Rom˜ ao, E. Paup´ erio, Flood risk assessment of cultural heritage at large spatial scales: framework and application to mainland Portugal, J. Cult. Herit. 43 (2020) 163–174, https://doi.org/10.1016/J.CULHER.2019.11.007. [11] C. Arrighi, M. Brugioni, F. Castelli, S. Franceschini, B. Mazzanti, Flood risk assessment in art cities: the exemplary case of Florence (Italy), J. Flood. Risk. Manage 11 (2018) S616–S631, https://doi.org/10.1111/JFR3.12226. [12] E.G.L. Nobile, R. Figueiredo, C. Arrighi, X. Rom˜ ao, M.L.V. Martina, Flood risk assessment of cultural heritage across countries and spatial scales, International J. Disast. Risk Reduct. 118 (2025) 105236, https://doi.org/10.1016/J. IJDRR.2025.105236. [13] Turkington, T., Breinl, K., Ettema, J., Alkema, D., & Jetten, V. (2016). A new flood type classification method for use in climate change impact studies. Weather and Climate extremes, 14, 1–16. https://doi.org/10.1016/J.WACE.2016.10.001. [14] W.joon Wang, D. Kim, H. Han, K. Tak Kim, S. Kim, H. Soo Kim, Flood risk assessment using an indicator based approach combined with flood risk maps and grid data, J. Hydrol. 627 (2023) 130396, https://doi.org/10.1016/J. JHYDROL.2023.130396. [15] C. De Lucia, M. Amaddii, C. Arrighi, Tangible and intangible ex post assessment of flood-induced damage to cultural heritage, Nat. Hazards Earth Syst. Sci. 24 (12) (2024) 4317–4339, https://doi.org/10.5194/NHESS-24-4317-2024. [16] B. Carmichael, C. Daly, S. Fatori´ c, M. Macklin, S. McIntyre-Tamwoy, W. Pittungnapoo, Global riverine archaeology and Cultural heritage: flood-risk management and adaptation for the anthropogenic climate change crisis, Climate, 11 (10) (2023) 197, https://doi.org/10.3390/cli11100197. [17] I.C. Nicu, S. Fatori´ c, Climate change impacts on immovable cultural heritage in polar regions: A systematic bibliometric review. Wiley Interdisciplinary Reviews, Clim. Change 14 (3) (2023) e822, https://doi.org/10.1002/wcc.822. [18] L. Babi´ nski, M. Zborowska, E. Fabisiak, W. Prądzy´ nski, Are the wooden remains of the Lusatian culture settlement at Biskupin safe? Decomposition of archaeological oak wood samples during a 10-year experiment, Archaeol. Anthropol. Sci. 11 (2019) 6583–6594, https://doi.org/10.1007/s12520-019-00926-0. [19] C.G. Bj¨ ordal, Decay of waterlogged archaeological wood: the importance of keeping it wet, Herit. Sci. 8 (2020) 3, https://doi.org/10.1186/s40494-020-003903. [20] S. Hariri, S. Weill, J. Gustedt, I. Charpentier, A balanced watershed decomposition method for rain-on-grid simulations in HEC-RAS, J. Hydroinform. 24 (2) (2022) 315–332, https://doi.org/10.2166/hydro.2022.078. [21] R. Hinsberger, A. Biehler, A. Y¨ orük, Influence of water depth and slope on roughness—Experiments and roughness approach for rain-on-grid modeling, Water (Switzerland) 14 (24) (2022), https://doi.org/10.3390/w14244017. [22] T. Iliopoulou, P. Dimitriadis, D. Koutsoyiannis, Pluvial flood risk assessment in urban areas: A case study for the archaeological site of the roman Agora, Athens, Heritage 6 (11) (2023) 7230–7243, https://doi.org/10.3390/HERITAGE6110379, 2023VolPages67230–7243. [23] W. Ennouini, A. Fenocchi, G. Petaccia, E. Persi, S. Sibilla, A complete methodology to assess hydraulic risk in small ungauged catchments based on HEC-RAS 2D Rainon-grid simulations, Nat. Hazards 120 (8) (2024) 7381–7409, https://doi.org/ 10.1007/S11069-024-06515-2. [24] M.J. Alexopoulos, P. Dimitriadis, T. Iliopoulou, N. Bezak, M. Kobold, D. Koutsoyiannis, Effects of Digital Elevation Model resolution on rain-on-grid simulations: a case study in a Slovenian watershed, Hydrol. Sci. J. (2024), https:// doi.org/10.1080/02626667.2024.2378487. [25] J. Rocha, A. Duarte, M. Silva, S. Fabres, J. Vasques, B. Revilla-Romero, A. Quintela, The importance of high resolution digital elevation models for improved hydrological simulations of a Mediterranean forested catchment, Remote Sens. (Basel) 12 (20) (2020) 3287, https://doi.org/10.3390/RS12203287, 2020Vol. 12, Page 3287. [26] A. Sivasena Reddy, M. Janga Reddy, Evaluating the influence of spatial resolutions of DEM on watershed runoff and sediment yield using SWAT, J. Earth Syst. Sci. 124 (7) (2015) 1517–1529, https://doi.org/10.1007/S12040-015-0617-2. [27] N. Godara, O. Bruland, K. Alfredsen, Comparison of two hydrodynamic models for their rain-on-grid technique to simulate flash floods in steep catchment, Front. Water. 6 (2024) 1384205, https://doi.org/10.3389/FRWA.2024.1384205. [28] K. Libelt, Mieszkania nawodne przedhistoryczne, Tygodnik Wielkopolski, R (1870) 61–62. I(1-5), 1–3, 13–14, 25–26, 37–38, 49–50. [29] J. Łepkowski, Szczątki mieszka´ n nawodnych w Czeszewie, Tygodnik Ilustrowany VIII (199) (1871) 200–201. [30] K. Dzięgielewski, The rise and fall of Biskupin and its counterparts, in: U. Bugaj (Ed.), The Past societies: Polish lands from the First Evidence of Human Presence to the Early Middle Ages, The Past societies: Polish lands from the First Evidence of Human Presence to the Early Middle Ages, 3, Institute of Archaeology and Ethnology, Polish Academy of Sciences, Warszawa, 2017, pp. 341–366. [31] J. Kostrzewski, Biskupin. An early Iron age village in western Poland, Antiquity. 12 (1938) 311–317. [32] A. Harding, W. Rączkowski, Living on the lake in the Iron age: new results from aerial photographs, geophysical survey and dendrochronology on sites of Biskupin type, Antiquity. 84 (324) (2010) 386–404, https://doi.org/10.1017/ S0003598X00066655. [33] D. Durczewski, Prasłowia´ nski gr´ od w Smuszewie, Pozna´ n: Muzeum Archeologiczne w Poznaniu (1970). [34] W. Chudziak, R. Ka´ zmierczak, J. Niegowski, Podwodne Archeologiczne Dziedzictwo Polski. Katalog stanowisk (badania 2006–2009), Uniwersytet Mikołaja Kopernika, Toru´ n, 2011. [35] C. Ioannidis, S. Verykokou, S. Soile, D. Istrati, C. Spyrakos, A. Sarris, D. Akritidis, H. Feidas, A.K. Georgoulias, E. Tringa, P. Zanis, C. Georgiadis, S. Martino, F. Feliziani, G.M. Marmoni, D. Cerra, M. Ottinger, F. Bachofer, A. Anastasiou, G. C. Anyfantis, Safeguarding our heritage—The TRIQUETRA project approach, Heritage 7 (2) (2024) 758–793, https://doi.org/10.3390/heritage7020037. [36] I. Pi´ nskwar, A. Chory´ nski, D. Graczyk, Z.W. Kundzewicz, Observed changes in extreme precipitation in Poland: 1991–2015 versus 1961–1990, Theor. Appl. Climatol. 135 (1–2) (2019) 773–787, https://doi.org/10.1007/S00704-018-23721. [37] Sutcliffe, J.V. (1978). Methods of Flood estimation, A guide to Flood studies report. [38] P. Licznar, J. Zaleski (Eds.), Metodyka Opracowania Polskiego Atlasu Natę˙ ze´ n Deszcz´ ow (PANDa), IMGW-PIB, Warszawa, 2020. [39] GUGiK. (2014a). Gł´ owny Urząd Geodezji i Kartografii. (2014a). Numeryczny Model Terenu (NMT/DTM) – dane z lotniczego skanowania laserowego (LiDAR). Pa´ nstwowy Zas´ ob Geodezyjny i Kartograficzny. https://www.geoportal.gov.pl. M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 18 [40] GUGiK. (2014b). Gł´ owny Urząd Geodezji i Kartografii. (2014b). Numeryczny model Pokrycia Terenu (NMPT/DSM) – dane z lotniczego skanowania laserowego (LiDAR). Pa´ nstwowy Zas´ ob Geodezyjny i Kartograficzny. https://www.geoportal. gov.pl. [41] Instytut Rybactwa ´ Sr´ odlądowego im. Stanisława Sakowicza – PIB. (1961). Plan batymetryczny jeziora Czeszewskie go. Olsztyn, Polska. [42] US Army Corps of Engineers. (2024a). HEC-RAS River Analysis System Hydraulic Reference Manual Version 6.6. [43] B. Leonard, Stability of explicit advection schemes. The balance point location rule, Int. J. Numer. Methods Fluids. 38 (5) (2002) 471–514, https://doi.org/10.1002/ fld.189. [44] US Army Corps of Engineers, H. E. C. (2024b). HEC-RAS River Analysis System HEC-RAS 2D User’s Manual. [45] USDA. (1986). Urban hydrology for small watersheds. [46] Zanaga, D., Van De Kerchove, R., De Keersmaecker, W., Souverijns, N., Brockmann, C., Quast, R., Wevers, J., Grosu, A., Paccini, A., Vergnaud, S., Cartus, O., Santoro, M., Fritz, S., Georgieva, I., Lesiv, M., Carter, S., Herold, M., Li, L., Tsendbazar, N.- E., Arino, O. (2021). ESA WorldCover 10 m 2020 v100. https://doi.org/10. 5281/ZENODO.5571936. [47] US Army Corps of Engineers. (2020). HEC-RAS mapper User’s manual. [48] L. Cea, M. ´ Alvarez, J. Puertas, Using integrated hydrological-hydraulic modeling and global data sources to analyse the February 2023 floods in the Umbeluzi Catchment (Mozambique), Nat. Hazards Earth Syst. Sci. 24 (1) (2024) 225–243, https://doi.org/10.5194/nhess-24-225-2024. [49] E.A. Baltas, N.A. Dervos, M.A. Mimikou, Technical note: determination of the SCS initial 870 abstraction ratio in an experimental watershed in Greece, Hydrol. Earth. Syst. Sci. 871 (6) (2007) 1825–1829, https://doi.org/10.5194/HESS-11-18252007, 11. [50] B.R. Hodges, Conservative finite-volume forms of the Saint-Venant equations for hydrology and urban drainage, Hydrol. Earth. Syst. Sci. 23 (3) (2019) 1281–1304, https://doi.org/10.5194/HESS-23-1281-2019. [51] M. Saraceni, B. Brunone, L. Silvestri, S. Meniconi, P. Bongioannini Cerlini, A water mass balance–Based procedure using ERA5 land reanalysis and level observation to reconstruct the past level changes of closed lakes toward future management, J. Hydrometeorol. 26 (1) (2025) 49–67, https://doi.org/10.1175/JHM-D-240019.1. [52] N. Krvavica, J. Rubini´ c, Evaluation of design storms and critical rainfall durations for flood prediction in partially urbanized catchments, Water. (Basel) 12 (7) (2020) 2044, https://doi.org/10.3390/w12072044. [53] G.W. Brunner, S.S. Piper, M.R. Jensen, B. Chacon, Combined 1D and 2D hydraulic modeling within HEC-RAS. World Environmental and Water Resources Congress 2015: floods, droughts, and ecosystems, in: Proceedings of the 2015 World Environmental and Water Resources Congress, 2015, pp. 1432–1443, https://doi. org/10.1061/9780784479162.141. [54] T. Yang, X. Li, Y. Yang, L. Wen, Z. Cao, X. Wang, S. Zhang, Q. Liang, Evolution from shallow-water deltas to fluvial fans in lacustrine basins: A case study from the Middle Jurassic Shaximiao Formation in the central Sichuan Basin, China, Sedimentology 71 (3) (2024) 1023–1055, https://doi.org/10.1111/SED.13163. [55] B. Jacob, E.V. Stanev, Understanding the impact of bathymetric changes in the German bight on coastal hydrodynamics: one step toward realistic morphodynamic modeling, Front. Mar. Sci. 8 (2021) 640214, https://doi.org/10.3389/ FMARS.2021.640214. [56] J. Sun, L. Peng, X. Zhu, Z. Li, H. Shi, C. Zhan, Z. You, The effects of changes in the coastline and water depth on tidal prism and water exchange of the Laizhou Bay, China, Front. Mar. Sci. 11 (2024) 1459482, https://doi.org/10.3389/ FMARS.2024.1459482. [57] J. Struzewska, J.W. Kaminski, M. Jefimow, Changes in temperature and precipitation trends in selected Polish cities based on the results of regional EUROCORDEX climate models in the 2030–2050 horizon, Appl. Sci. 14 (1) (2023) 9, https://doi.org/10.3390/APP14010009, 2024, Vol. 14, Page 9. [58] B. Ghazi, R. Przybylak, A. Pospieszy´ nska, Projection of climate change impacts on extreme temperature and precipitation in Central Poland, Sci. Rep. 13 (1) (2023) 1–13, https://doi.org/10.1038/S41598-023-46199-5. ; SUBJMETA=106,2739,2786,694,704;KWRD=CLIMATE-CHANGE+IMPACTS, PROJECTION+AND+PREDICTION. [59] I. Pi´ nskwar, A. Chory´ nski, D. Graczyk, Vulnerability to extreme weather events: mapping future hazards in Wielkopolska region, Poland, Region. Environ. Change 24 (4) (2024) 1–16, https://doi.org/10.1007/S10113-024-02305-5. [60] G. Zittis, A. Bruggeman, J. Lelieveld, Revisiting future extreme precipitation trends in the Mediterranean, Weather. Clim. Extrem. 34 (2021) 100380, https://doi.org/ 10.1016/J.WACE.2021.100380. [61] W. Ingram, Extreme precipitation: increases all round, Nat. Clim. Chang. 6 (5) (2016) 443–444, https://doi.org/10.1038/nclimate2966. [62] H. Tabari, Climate change impact on flood and extreme precipitation increases with water availability, Sci. Rep. 10 (1) (2020) 1–10, https://doi.org/10.1038/ S41598-020-70816-2. [63] M. Ptak, K. Szyga-Pluta, M. Sojka, Long-term trends in water level fluctuations in lowland lakes in central Europe (Northern Poland), Sci. Reports, 15 (1) (2025) 1–15, https://doi.org/10.1038/S41598-024-83286-7. [64] M.B. Kayastha, X. Ye, C. Huang, P. Xue, Future rise of the Great Lakes water levels under climate change, J. Hydrol. 612 (2022) 128205, https://doi.org/10.1016/J. JHYDROL.2022.128205. [65] M. Somogyv´ ari, D. Scherer, F. Bart, U. Fehrenbach, A. Okujeni, T. Krueger, A hybrid data-driven approach to analyze the drivers of lake level dynamics, Hydrol. Earth. Syst. Sci. 28 (18) (2024) 4331–4348, https://doi.org/10.5194/ HESS-28-4331-2024. [66] National Flood Risk Advisory Group, Technical Flood Risk Management guideline: Flood hazard, 2017. https://knowledge.aidr.org.au/media/3518/adr-guidelin e-7-3.pdf. [67] A. Quin, G. Destouni, Large-scale comparison of flow-variability dampening by lakes and wetlands in the landscape, Land. Degrad. Dev. 29 (10) (2018) 3617–3627, https://doi.org/10.1002/ldr.3101. [68] J. Garrote, M. Gonz´ alez-Jim´ enez, C. Guardiola-Albert, A. Díez-Herrero, The Manning’s roughness coefficient calibration method to improve flood hazard analysis in the absence of river bathymetric data: application to the urban historical Zamora City Centre in Spain, Appl. Sci. 11 (19) (2021) 9267, https://doi. org/10.3390/app11199267. [69] S.A.R. Tilano, M.-A. Boucher, J. Lacey, J. Parent, Quantifying changes in floods under different bathymetry conditions for a lake setting, Canad. J. Civil Eng. (2024), https://doi.org/10.1139/CJCE-2023-0237. [70] A. Cook, V. Merwade, Effect of topographic data, geometric configuration and modeling approach on flood inundation mapping, J. Hydrol. 377 (1–2) (2009) 131–142, https://doi.org/10.1016/J.JHYDROL.2009.08.015. [71] L. Reimann, A.T. Vafeidis, S. Brown, J. Hinkel, R.S.J. Tol, Mediterranean UNESCO World Heritage at risk from coastal flooding and erosion due to sea-level rise, Nat. Commun. 9 (1) (2018) 1–11, https://doi.org/10.1038/S41467-018-06645-9. [72] Y. Shen, N. Tahvildari, M.M. Morsy, C. Huxley, T.D. Chen, J.L. Goodall, Dynamic modeling of inland flooding and storm surge on coastal cities under climate change scenarios: transportation infrastructure impacts in Norfolk, Virginia USA as a case study, Geosciences. (Basel) 12 (6) (2022) 224, https://doi.org/10.3390/ geosciences12060224. [73] B.F. Sanders, O.E.J. Wing, P.D. Bates, Flooding is not like filling a bath, Earth’s Future 12 (12) (2024) e2024EF005164, https://doi.org/10.1029/2024EF005164. [74] K. Trepekli, T. Balstrøm, T. Friborg, B. Fog, A.N. Allotey, R.Y. Kofie, L. MøllerJensen, UAV-borne, LiDAR-based elevation modeling: a method for improving local-scale urban flood risk assessment, Nat. Hazards 113 (1) (2022) 423–451, https://doi.org/10.1007/S11069-022-05308-9. [75] M. Schütze, P. Willems, G. Vaes, Integrated simulation of urban wastewater systems - how many rainfall data do we need?. Global Solutions for Urban Drainage American Society of Civil Engineers, 2002, p. 244, https://doi.org/10.1061/40644 (2002)244. [76] F. Tscheikner-Gratl, P. Zeisl, C. Kinzel, J. Leimgruber, T. Ertl, W. Rauch, M. Kleidorfer, Lost in calibration: why people still do not calibrate their models, and why they still should – A case study from urban drainage modelling, Water Sci. Technol. 74 (10) (2016) 2337–2348, https://doi.org/10.2166/wst.2016.395. [77] R. Loritz, M. Hrachowitz, M. Neuper, E. Zehe, The role and value of distributed precipitation data in hydrological models, Hydrol. Earth. Syst. Sci. 25 (1) (2021) 147–167, https://doi.org/10.5194/hess-25-147-2021. [78] A. Annis, F. Nardi, A. Petroselli, C. Apollonio, E. Arcangeletti, F. Tauro, C. Belli, R. Bianconi, S. Grimaldi, UAV-DEMs for small-scale flood hazard mapping, Water. (Basel) 12 (6) (2020) 1717, https://doi.org/10.3390/w12061717. [79] J.L. Mercau, M.D. Nosetto, F. Bert, R. Gim´ enez, E.G. Jobb´ agy, Shallow groundwater dynamics in the Pampas: climate, landscape and crop choice effects, Agric. Water Manage. 163 (2016) 159–168, https://doi.org/10.1016/j. agwat.2015.09.013. [80] Y. Abebe, A. Ghorbani, I. Nikolic, Z. Vojinovic, A. Sanchez, Flood risk management in Sint Maarten – A coupled agent-based and flood modelling method, J. Environ. Manage (2019) 248, https://doi.org/10.1016/j.jenvman.2019.109317. [81] M.J. Alexopoulos, T. Iliopoulou, P. Mod´ e, D. Istrati, C.C. Spyrakos, S. Soile, S. Verykokou, C. Ioannidis, D. Koutsoyiannis, A novel and scalable flood risk assessment framework for cultural heritage based on unmanned aerial vehicle photogrammetry and multi-scale rain-on-grid hydraulic modeling, Sci. Total. Environ. (2025) 998, https://doi.org/10.1016/j.scitotenv.2025.180256. [82] X. Huang, B. Xu, P. Zhong, H. Yao, H. Yue, F. Zhu, Q. Lu, Y. Sun, R. Mo, Z. Li, W. Liu, Robust multiobjective reservoir operation and risk decision-making model for real-time flood control coping with forecast uncertainty, J. Hydrol. (2022) 605, https://doi.org/10.1016/j.jhydrol.2021.127334. [83] Q. Zhao, D. Li, X. Cai, Online generic diagnostic reservoir operation tools, Environ. Modell. Software (2021) 135, https://doi.org/10.1016/j.envsoft.2020.104918. M.J. Alexopoulos et al. Results in Engineering 28 (2025) 107977 19