Groundwater Level Fluctuation Analysis in a Semi-Urban Area Using Statistical Methods and Data Mining Techniques—A Case Study in Wrocław, Poland

Abstract

Long-term groundwater level analysis, which is usually based on traditionally defined hydrological years is essential in an era of global warming and other climatic and environmental changes, especially in urban areas. A complex interplay of multiple factors influencing the groundwater level makes the investigation of their interdependencies a challenge. Based on multiple data sets and a long time series available as well as specific geological and hydrological conditions, a semi-urban district of Wrocław/Poland was selected as a case study for investigating these dependencies. This paper presents an interdisciplinary approach to the analysis of groundwater level fluctuations by combining mathematics, signal processing, hydrogeology, and meteorology. Applying well-known methods from disciplines other than hydrogeology, the authors investigated seasonal behavior and similarity of groundwater level fluctuations during 15 hydrological years. Based on segmentation and agglomerative clustering (AHP), five classes of groundwater levels fluctuations for predefined hydrologic years and the corresponding seasons were identified and compared to the classification scheme by Pleczyński. Additionally, the relationship between precipitation and groundwater level was investigated using Pearson, Kendall and Spearman correlations. This led to the identification of “typical” and “untypical” seasons for the correlation between the cumulative precipitation sum and groundwater levels. The results presented here will be used for further investigations of groundwater level fluctuations using additional factors and statistical methods. These aim to identify periods that describe similarities better than the commonly used hydrological year.