Fluvial channel‐belt clustering has recently been documented using quantitative metrics for systems dominated by autogenic controls. It has long been recognized that allogenic forcing (tectonic and eustatic controls) can lead to confinement of fluvial systems, resulting in clustering of channel belts. To date, no study has quantitatively documented the differences in channel‐belt clustering, compensational stacking of channel belts and interchannel‐belt connectivity in unconfined and confined systems. This study quantitatively compares world‐class outcrops of an unconfined fluvial system (Palaeocene lower Wasatch Formation) with outcrops of a confined fluvial system (Cretaceous Dakota Sandstone). Two new methods have been developed to quantitatively document channel‐belt clustering and intrachannel‐belt connectivity. These new methods, and other previously developed methods, are used to document an increase in channel‐belt clustering and intrachannel‐belt connectivity downdip in both systems. Additionally, it was found that channel belts within the unconfined system stack more compensationally than those in the confined system. These new methods and empirical relationships can be used for predicting intrachannel‐belt connectivity, and accurately modelling unconfined and confined fluvial systems in the subsurface.

Presented here is a transfer‐learning model for classifying basin‐scale stratigraphic geometries from subsurface formation tops. Support vector, decision trees, random forests, AdaBoost and K‐nearest neighbour classification models are evaluated to support this challenge. Each model is trained on labelled synthetic stratigraphic geometry data generated in Python using observable geological principles and concepts. Accuracy is measured using a weighted Jaccard similarity coefficient score, and certainty of each prediction is quantified using margin sampling. The random forest classifier has the highest initial accuracy, and the optimal hyperparameters for the model that yield 88.4% accuracy and 72.8% mean certainty via five‐fold cross‐validation and active learning are documented on a real‐world subsurface dataset. The random forest classifier with optimised hyperparameters is then used to make predictions on the real‐world subsurface formation tops dataset. The dataset consists of formation tops for the Upper Cretaceous and Palaeocene strata of the Eastern Greater Green River Basin in south‐central Wyoming. Results from model predictions include an area of truncation in the Lance Formation across the basin, and an area of onlap and truncation on the nose of the Rock Springs Uplift that previous studies in the region corroborate. It is believed that this model is most useful for guided interpretation, and identifying regions that warrant further inquiry by domain experts.

No abstract

No abstract