A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Deep learning text classification of borehole logs for regional scale modeling of hydrofacies (Po Plain, N Italy)
Study region: The Po River alluvial plain in northern Italy, spanning 45,700 km², stands as one of Europe's most extensive groundwater reservoirs, serving the needs of approximately 15 million residents. It is underlain by a substantial sequence of quaternary fluvio-glacial and alluvial plain deposits originating from the Alpine and Apennine ranges. These deposits form aquifer systems at a regional scale, exhibiting grain size variations that correlate with lithology, proximity to source areas, and depositional age. Study focus: In this study, we have transformed a considerable volume of qualitative lithotextural data from borehole logs - comprising 39,265 boreholes and 387,297 descriptive intervals - into a semi-quantitative hydrogeological framework. This was achieved through an automated deep learning process that classified geological descriptions into hydrofacies based on the grain size. We employed a long short-term memory (LSTM) recurrent neural network algorithm, which was trained and validated using 86,611 pre-labelled entries encompassing all sediment types within the study region. The word embedding technique enhanced the model accuracy and learning efficiency by quantifying the semantic distances among geological terms. The primary objectives of this research are twofold: (i) to develop a robust deep learning classification model that leverages geological descriptions alongside grain size data, and (ii) to standardize a vast array of sparse and heterogeneous stratigraphic log data for integration into hydrofacies models tailored for hydrogeological applications. New hydrological insights for the region: The outcome of this work is a novel dataset of semi-quantitative hydrogeological information, boasting a classification model accuracy of 97.4 %. This dataset was incorporated into expansive modeling frameworks, enabling the assignment of hydrogeological parameters based on grain size data, integrating the uncertainty stemming from misclassification. This has markedly increased the spatial density of available information from 0.34 data points/km² to 8.7 data points/km². The study findings align closely with existing literature maps, offering a robust spatial reconstruction of hydrofacies at different scales. This has significant implications for groundwater research, particularly in the realm of quantitative modeling at a regional scale.
Deep learning text classification of borehole logs for regional scale modeling of hydrofacies (Po Plain, N Italy)
Study region: The Po River alluvial plain in northern Italy, spanning 45,700 km², stands as one of Europe's most extensive groundwater reservoirs, serving the needs of approximately 15 million residents. It is underlain by a substantial sequence of quaternary fluvio-glacial and alluvial plain deposits originating from the Alpine and Apennine ranges. These deposits form aquifer systems at a regional scale, exhibiting grain size variations that correlate with lithology, proximity to source areas, and depositional age. Study focus: In this study, we have transformed a considerable volume of qualitative lithotextural data from borehole logs - comprising 39,265 boreholes and 387,297 descriptive intervals - into a semi-quantitative hydrogeological framework. This was achieved through an automated deep learning process that classified geological descriptions into hydrofacies based on the grain size. We employed a long short-term memory (LSTM) recurrent neural network algorithm, which was trained and validated using 86,611 pre-labelled entries encompassing all sediment types within the study region. The word embedding technique enhanced the model accuracy and learning efficiency by quantifying the semantic distances among geological terms. The primary objectives of this research are twofold: (i) to develop a robust deep learning classification model that leverages geological descriptions alongside grain size data, and (ii) to standardize a vast array of sparse and heterogeneous stratigraphic log data for integration into hydrofacies models tailored for hydrogeological applications. New hydrological insights for the region: The outcome of this work is a novel dataset of semi-quantitative hydrogeological information, boasting a classification model accuracy of 97.4 %. This dataset was incorporated into expansive modeling frameworks, enabling the assignment of hydrogeological parameters based on grain size data, integrating the uncertainty stemming from misclassification. This has markedly increased the spatial density of available information from 0.34 data points/km² to 8.7 data points/km². The study findings align closely with existing literature maps, offering a robust spatial reconstruction of hydrofacies at different scales. This has significant implications for groundwater research, particularly in the realm of quantitative modeling at a regional scale.
Deep learning text classification of borehole logs for regional scale modeling of hydrofacies (Po Plain, N Italy)
Alberto Previati (author) / Valerio Silvestri (author) / Giovanni Crosta (author)
2025
Article (Journal)
Electronic Resource
Unknown
Metadata by DOAJ is licensed under CC BY-SA 1.0
| Elsevier | 2025
UK Chalk Group stratigraphy (Cenomanian - Santonian) determined from borehole geophysical logs
| Online Contents | 2006
Hydrogeologic structure of the Newark Basin, Pennsylvania, from borehole geophysical logs
| British Library Conference Proceedings | 1999
UK Chalk Group stratigraphy (Cenomanian - Santonian) determined from borehole geophysical logs
| Online Contents | 2006