Beteiligte:
Ma, Yueling;
Montzka, Carsten;
Naz, Bibi S;
Kollet, Stefan
Erschienen:
IOP Publishing, 2022
Erschienen in:
Environmental Research Letters, 17 (2022) 11, Seite 114037
Sprache:
Nicht zu entscheiden
DOI:
10.1088/1748-9326/ac9c1e
ISSN:
1748-9326
Entstehung:
Anmerkungen:
Beschreibung:
<jats:title>Abstract</jats:title>
<jats:p>The main challenge of pan-European groundwater (GW) monitoring is the sparsity of collated water table depth (<jats:italic>wtd</jats:italic>) observations. The <jats:italic>wtd</jats:italic> anomaly (<jats:italic>wtd<jats:sub>a</jats:sub>
</jats:italic>) is a measure of the increased <jats:italic>wtd</jats:italic> due to droughts. Combining long short-term memory (LSTM) networks and transfer learning (TL), we propose an AI-based methodology LSTM-TL to produce reliable <jats:italic>wtd<jats:sub>a</jats:sub>
</jats:italic> estimates at the European scale in the absence of consistent <jats:italic>wtd</jats:italic> observational data sets. The core idea of LSTM-TL is to transfer the modeled relationship between <jats:italic>wtd<jats:sub>a</jats:sub>
</jats:italic> and input hydrometeorological forcings to the observation-based estimation, in order to provide reliable <jats:italic>wtd<jats:sub>a</jats:sub>
</jats:italic> estimates for regions with no or sparse <jats:italic>wtd</jats:italic> observations. With substantially reduced computational cost compared to physically-based numerical models, LSTM-TL obtained <jats:italic>wtd<jats:sub>a</jats:sub>
</jats:italic> estimates in good agreement with <jats:italic>in-situ wtd<jats:sub>a</jats:sub>
</jats:italic> measurements from 2569 European GW monitoring wells, showing <jats:italic>r</jats:italic> ⩾ 0.5, root-mean-square error ⩽1.0 and Kling-Gupta efficiency ⩾0.3 at about or more than half of the pixels. Based on the reconstructed long-term European monthly <jats:italic>wtd<jats:sub>a</jats:sub>
</jats:italic> data from the early 1980s to the near present, we provide the first estimate of seasonal <jats:italic>wtd<jats:sub>a</jats:sub>
</jats:italic> trends in different European regions, that is, significant drying trends in central and eastern Europe, which facilitates the understanding of historical GW dynamics in Europe. The success of LSTM-TL in estimating <jats:italic>wtd<jats:sub>a</jats:sub>
</jats:italic> also highlights the advantage of combining AI techniques with knowledge contained in physically-based numerical models in hydrological studies.</jats:p>