The recent interest in using deep learning for seismic interpretation tasks, such as facies classification, has been facing a significant obstacle, namely the absence of large publicly available annotated datasets for training and testing models. As a result, researchers have often resorted to annotating their own training and testing data. However, different researchers may annotate different classes, or use different train and test splits. In addition, it is common for papers that apply machine learning for facies classification to not contain quantitative results, and rather rely solely on visual inspection of the results. All of these practices have lead to subjective results and have greatly hindered the ability to compare different machine learning models against each other and understand the advantages and disadvantages of each approach. To address these issues, we open-source a fullyannotated 3D geological model of the Netherlands F3 Block. This model is based on the study of the 3D seismic data in addition to 26 well logs, and is grounded on the careful study of the geology of the region. Furthermore, we propose two baseline models for facies classification based on a deconvolution network architecture and make their codes publicly available. Finally, we propose a scheme for evaluating different models on this dataset, and we share the results of our baseline models. In addition to making the dataset and the code publicly available, this work helps advance research in this area by creating an objective benchmark for comparing the results of different machine learning approaches for facies classification.

The analysis of the positive feedback between landslides and erosion requires determination of the precise temporal and spatial relations between events of colluvium delivery and fluvial erosion. In our study we use decennial datasets on the occurrence of landsliding and erosion achieved through dendrochronological methods. Four sites covering areas of landslide slopes and adjacent valley floors with stream channels were studied. Landsliding on slopes was dated from the tree-ring eccentricity developed in stems tilted due to bedrock instability. Erosion in channels was dated using the wood anatomy of roots exposed by erosion of the soil cover. Analysis of the temporal relations between dated landsliding, erosion and precipitation record has revealed that two types of repeating sequences can be observed: (1) rainfall → landsliding → erosion; (2) rainfall → erosion → landsliding. These sequences are an indication of the occurrence of slope-channel positive feedback in the sites studied. In the first type, landsliding triggered by rainfall delivers colluvia into the valley floor and causes its narrowing, which in turn causes increased erosion. In the second type erosion triggered by rainfall disturbs the slope equilibrium and causes landsliding. Landsliding and erosion, once triggered by precipitation, can occur alternately in years with average precipitation and reinforce one another. Bidirectional coupling between landsliding and channel erosion was shown notably through the effects of channel shifting and forced sinuosity and by increased erosion of the slopes opposite the active landslides. Observations also suggest that the repetition of sequences described over longer periods of time can lead to a general widening of the valley floor at the expense of slopes and to a gradual change of the valley crossprofile from narrow, V-shaped into a wide flat-bottomed. Thus landsliding-erosion coupling/positive feedback was recognized as an important factor shaping hillslope-valley topography of the mid-mountain areas studied.

No abstract

No abstract