The GRUVO web application: Bringing groundwater level predictions across Germany to the public

The provision of current and predicted groundwater levels across Germany has become increasingly important, particularly due to the increasing likelihood of consecutive dry years. To address this issue, we present the interactive web application GRUVO, which was developed as a first step to provide groundwater level forecasts and relevant information in a targeted manner for different user groups. We also provide an overview of the features and operation of the application in its current version.

In addition to the visualisation of current groundwater levels, this mainly includes the presentation of monthly updated groundwater level forecasts and projections for short-term (up to 3 months), medium-term (up to 10 years) and long-term (up to 2100) forecast horizons at over 100 so-called reference monitoring sites (RM) distributed throughout Germany. Each of these RMs represents the groundwater levels or dynamics of a few thousand so-called cluster monitoring sites (CMs). This mapping of RMs to CMs was previously determined using a clustering approach. The RM prediction is based on 1-D convolutional neural networks (CNN), which are trained using time series of measured groundwater level data from the responsible state offices as target variables and measured meteorological forcing data from the German Weather Service (DWD) as predictors. Forecasted or projected meteorological information from the DWD is then used to predict future groundwater levels.

Apart from the available features of the current version, this contribution highlights operational challenges and nuances. It also outlines possible extensions for future development.