Unsupervised clustering of deep image features is often proposed for unlabeled image archives, yet it is usually judged by internal criteria rather than by agreement with expert annotations. In a single domain of texture images, namely full-size rock-core scans, we evaluate unsupervised partitions against depth-registered expert descriptions of 20 wells (nine classes), holding out entire wells. Color statistics, texture descriptors, VGG16, two DINOv2 models and SimCLR encoders pre-trained on the images are combined with k-means, Gaussian mixtures and SCAN, several seeds, and pre-training runs. Across all combinations, the mean class accuracy plateaus at 0.35–0.40 (chance level 0.12), and six color statistics reach this plateau with a Gaussian mixture model (0.37). Apparent gains of in-domain self-supervised features depend on the clustering algorithm and the pre-training run. The plateau is robust to the evaluation protocol, a coarser class scheme and more diverse fitting wells, and persists on color-normalized and grayscale images. Linear probes on the same frozen features reach 0.47–0.54, which locates the bottleneck in the unsupervised objectives. Fitting the clustering transductively on all unlabeled images lifts in-domain features to 0.42–0.48, but this gain does not transfer to unseen wells. The conclusions are specific to this domain.

This study presents a comprehensive numerical simulation of reservoir dam failure based on the two-dimensional hydrodynamic model MIKE 21. To reproduce the real accident process, a detailed digital elevation model derived from LiDAR survey data was constructed, incorporating valley microtopography, river channel geometry, and hydraulic structure elements. The modeling was performed in a stepwise manner and included the simulation of breach formation using a time-varying digital elevation model, the drawdown of the reservoir, and the propagation of the dam-break flood wave in the downstream reach, as well as an assessment of the hydrodynamic impact of the flow on small dams located further downstream. The simulations produced spatiotemporal distributions of flow depths and velocities, quantified the temporal evolution of reservoir water volume, and determined overflow parameters at the small dams. Based on the analysis of bed shear stress distribution, zones of increased hydrodynamic loading were identified and compared with observed damage areas. The results confirm the applicability of the adopted modeling framework for detailed reconstruction of dam-break events. The proposed approach can be applied both to the analysis of past dam failures and for predictive purposes when assessing the potential consequences of possible accidents at other reservoirs. The methodology enables preliminary evaluation of inundation zones, erosion intensity, and impacts on downstream hydraulic structures, making it a valuable tool for safety assessment and the planning of protective measures in areas with complex terrain conditions.