Spatial-temporal changes of the long-term average monthly runoff in Slovakia

This study aims to analyse the changes in the long-term average monthly runoff regime on the territory of Slovakia. For the analysis, the average monthly discharges from 57 gauging stations of the whole territory of Slovakia, which the Slovak Hydrometeorological Institute provided for the period 1961-2020, were used. The selected basin areas range from approximately 10 km2 to 1000 km2. The monthly discharge data available were divided into two periods: the old one, 1961-2000, and the new reference period, according to the World Meteorological Organization (WMO), from 1991 to 2020.  The appropriate number of clusters was determined according to the statistical analysis using the average Silhouette Width and the Elbow method. Subsequently, the PCA method and K-means clustering were performed to pool the catchments into groups. The results present the outputs of the particular runoff regime in the selected gauging stations divided into five clusters. South-central Slovakia and central Slovakia characterise Cluster No. 1; Cluster No. 2 by the northwest and northeast of the country; Cluster No. 3 for the centre of northern Slovakia; Cluster No. 4 for central Slovakia, and Cluster No. 5 for the east, south and west part of Slovakia. When comparing the changes in the regime of both periods, we can state that the best, 89% agreement was in Custer No.2, representing the High Tatras region, and 86 % agreement was in Cluster No.5 for the western part of Slovakia, with is 44% in Custer No.4, where the highest long-term average monthly flows remain in April, but the lowest normalized long-term average monthly flows shift from January to February. In Custer No.3, with 40 % agreement of catchment forming the clusters, we find a similar shift of peak long-term average monthly flows from April (1961-2000) to March (1991-2020). The most significant changes in the long-term average monthly runoff regime were found for the catchment in Cluster No.1.

Finally, the most important characteristic features of the individual clusters of gauging stations created were also analysed, which could help incorporate other catchments into appropriate regional types in the future. The methodological procedure developed could also be used in further studies to predict future flow regime changes on the territory of Slovakia.

Acknowledgements

This study was supported by the Slovak Research and Development Agency under Contract No. APVV-20-0374 and VEGA Grant Agency No 1/0782/21. The authors thank the agencies for their research support.