The maintenance of buildings, underpinned by the digital twin technique, becomes integral to heritage conservation efforts. To achieve efficient modelling with minimal manual intervention, automated component recognition based on semantic segmentation of point clouds is imperative. Confronted by the challenges of the paucity of requisite datasets and the inherent geometric diversity of historical buildings, a two-step strategy including feature extraction and classification is proposed. First, an improved SHOT descriptor is proposed to extract discriminative features by defining a specific local reference system and concatenating support fields at different scales. The extracted features are then classified with a learning-based network, avoiding a feature learning process that relies on sufficient data. Experiments on real-world heritage point clouds yield 93.7% accuracy and an 80.0% mean-intersection-over-union (mIoU) when descriptors with radii of 0.3 m and 0.9 m are combined, surpassing computationally expensive deep learning networks and data-intensive unsupervised learning. A slight decrease in segmentation performance with random removal of points indicates the high robustness of the proposed method against data missing and sampling density changes. Additionally, a geometric modelling process with an error of less than 10% is introduced to achieve a direct transition from point cloud to model, contributing to the establishment of digital twins for heritage structures.

The middle Eocene was a key period of global climate change from a warm “greenhouse” to a cooling “doubthouse”. At the middle Eocene, an extensive arid climate in China was recorded in the mudstones and shales interbedded with salinastone which formed good hydrocarbon source rocks, especially in the Bohai Bay Basin. The sedimentary paleoenvironment and its variations affected the source rock quality and distribution. Based on successive and dense sampling of the middle Eocene Shahejie Formation shale (MES shale) in Nanpu Sag of Bohai Bay Basin, petro-mineralogical tests, geochemical methods and spectrum analysis were performed to analyse the palaeoenvironmental fluctuations and the responses to Milankovitch cycles. Element analysis on MES shale shows that the middle Eocene climate was arid and the sediments were deposited under dysoxic/suboxic conditions with relatively high salinity. Based on the Δlog R method, the organic matter (OM) abundance was evaluated and ranged from 0.18 to 3.55 wt % for MES shale. The mean primary productivity of 545 g C m⁻² yr⁻¹ shows a eutrophic environment in middle Eocene Nanpu paleolake. Wavelet analysis performed on gamma logging and palaeoenvironmental proxies revealed clear Milankovitch cycles for MES shale, and the cyclostratigraphic and palaeoenvironmental fluctuations were effected by astronomical oscillation periods of the short eccentricity cycle (100 k.y.), obliquity cycle (39 k.y.) and precession cycle (19 k.y.). The depositional rate of MES shale was corrected to 0.43 m/k.y. The sedimental process of MES shale was associated with palaeoenvironmental variation, which have been divided into four stages affecting OM accumulation. A hypothesis that the vast carbon of “greenhouse” gases in the Palaeocene and early Eocene was fixed in organic matters during middle Eocene and enriched in the hydrocarbon source rocks of the sedimentary basins formed at early Eocene was proposed to explain the “doubthouse” world in the Eocene.