The paper develops an ensemble deep learning framework to forecast groundwater storage (GWS) under hydrological variability in the Middle Mahanadi Basin, Odisha, India.

The paper develops an ensemble deep learning framework to forecast groundwater storage (GWS) under hydrological variability in theMiddle Mahanadi BasinOdisha, India.

The objectives were to

• estimate monthly streamflow and potential evapotranspiration (PET) from 1981–2022
• derive long term GWS trends using a water budget approach
• compare several deep learning models—LSTM, stacked LSTM, BiLSTM, GRU, and their ensemble—for predicting GWS up to 2028.

The study area covers the Middle Mahanadi sub basin (drainage 9,421 km²) with eight rain gauge stations across Odisha districts such as Padampur, Phulbani, Deogan, Kantamal and Kesinga, characterized by monsoonal rainfall (1,170–1,490 mm) and contrasting upper water stress and lower water logging zones.

The methodology used IMD gridded rainfall/temperature, SRTM DEM, Landsat LULC, and soil grids to compute PET via the Hargreaves–Samani method and surface runoff via SCS CN, then estimated GWS from a basin water balance S=P-ET-Q_s. Time series of

Digital marketplace for Ui/Ux designers.

consectetur adipiscing elit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat mattis vulputate cupidatat.

• Consectetur adipiscing elit in voluptate velit.
• Mattis vulputate cupidatat.
• Vulputate enim nulla aliquet porttitor odio pellentesque
• Ligula ullamcorper malesuada proin

consectetur adipiscing elit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat mattis vulputate cupidatat.