Forecasting groundwater table for the sustenance and conservation of water-dependent ecosystems in protected areas: the case of the Wielkopolski National Park in Poland

• Published in: Stochastic environmental research and risk assessment Vol. 39; no. 5; pp. 2159 - 2182
• Main Authors: Graf, Renata, Lech, Kaczmarek, Mariusz, Pełechaty, Rafał, Kurczewski
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.05.2025; Springer Nature B.V
• Subjects: Aquatic Pollution; Catchment areas; Chemistry and Earth Sciences; Computational Intelligence; Computer Science; Earth and Environmental Science; Earth Sciences; Ecosystems; Environment; Environmental monitoring; Environmental protection; Forecasting; Groundwater; Groundwater levels; Lake catchments; Land cover; Landscape protection; Math. Appl. in Environmental Science; Monitoring; National parks; Neural networks; Original Paper; Physics; Piezometers; Probability Theory and Stochastic Processes; Protected areas; River basins; Seasonal variations; Soil permeability; Statistics for Engineering; Trends; Typology; Waste Water Technology; Water conservation; Water Management; Water monitoring; Water Pollution Control; Water scarcity; Water table; Poland; National park; Water-dependent ecosystem; Hybrid model; Water deficiency hazard; Groundwater modeling; Forecast
• ISSN: 1436-3240; 1436-3259
• DOI: 10.1007/s00477-025-02962-w

Groundwater table (GWT) forecasting is central to groundwater monitoring, particularly in regions suffering from severe water scarcity that poses a threat to groundwater-dependent ecosystems (GDEs). The main aim of this study was to assess fluctuations (2016–2021) and to forecast GWT changes with a view to sustaining and protecting GDEs in a legally protected area that is subject to the strictest diverse nature protection regime (comprising strict, active and landscape protection) in the Wielkopolski National Park (WNP) in Western Poland. The study covers a river basin and two endorheic lake catchment areas. It relies on data from the meteorological and hydrological monitoring carried out in the park. To predict GWT changing patterns, a five-component hybrid model was applied: “Best ARIMA” model, “Exponential Smoothing with Drift”, “Neural Network Autoregression”, “Seasonal and Trend Decomposition using Loess” and furthermore “Trigonometric Seasonal, Box-Cox Transformation, ARMA Errors, Trend, and Seasonal Components”. The GDEs typology was carried out in a source data matrix with a mesh size of 10 × 10 m using GIS techniques that rely on soil permeability, land cover and terrain as key water reception indicators. The hybrid model was found to increase the information potential of future GWT change patterns. Each method in hybrid model addressed different aspects of GWT dynamics, including seasonality, long-term trends, and random components, and their combination allowed to capture nonlinear and heterogeneous dependencies that single models often fail to address. The forecasts identified the predominant trend of steady GWT lowering (in the range of 0.02–0.29 m) subject to regular seasonality. Only two piezometers displayed a slight tendency of GWT rise (up to 0.29 m). A growing GW deficit in the WNP has been suggested by a downward trend that has continued since 2019, and by the prolonged persistence of GWT below the median. The projected GW deficiency was confirmed to contribute to GDEs degradation, also in strictly protected areas, its specific impacts varying depending on hydrogenic habitat type.