Machine learning approach for prediction of groundwater levels based on ERA5 reanalysis
- 1AGH University of Krakow, Faculty of Geology, Geophysics and Environmental Protection, Krakow, Poland (zurek@agh.edu.pl)
- 2AGH University of Krakow, Faculty of Physics and Applied Computer Science, Krakow, Poland (rszostak@agh.edu.pl)
We have examined the feasibility of ECMWF Reanalysis (ERA5) data for groundwater level prediction for 19 groundwater wells from two neighboring Groundwater Bodies (GWB) comprising around 4000 km2. Groundwater level data were retrieved from monitoring wells operated within the framework of the Polish Hydrogeological Survey. ERA5 reanalysis data were averaged for all grid points within the modelling area. Predictions were made using various machine learning regression algorithms incorporating autoregression and exogeneous variables derived from ERA5 reanalysis (precipitation amount, evapotranspiration, runoff, snowmelt). Training sets were extracted from time series of data representing period from November 2001 to November 2022. The applied approach allows for predicting groundwater levels based on current meteorological conditions.
This research was funded by National Science Centre, Poland, project WATERLINE (2020/02/Y/ST10/00065), under the CHISTERA IV programme of the EU Horizon 2020 (Grant no 857925).
How to cite: Żurek, A. J., Szostak, R., Wachniew, P., and Zimnoch, M.: Machine learning approach for prediction of groundwater levels based on ERA5 reanalysis, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-12660, https://doi.org/10.5194/egusphere-egu24-12660, 2024.
