6974 research outputs found
Sort by
Refining Late Holocene explosive eruption histories of the Main Ethiopian Rift with lake sediment tephra records
Full text linkThe Main Ethiopian Rift (MER) hosts a rapidly growing population exposed to eruption hazards from ∼60 active Holocene volcanoes. The geological record preserves significant information about the nature and frequency of past explosive volcanism in the rift, but the Holocene eruption record remains incomplete and its implications for contemporary volcanic hazards are yet to be fully understood. Here we use lake sediments preserved at four sites: Babogaya (MER), Haro Kori and Wergoba (southeastern Ethiopian plateau), and Dendi (western Ethiopian plateau) to constrain the Late Holocene tephrostratigraphic record. We focus on Lake Babogaya in the Bishoftu Volcanic Field which preserves 5 visible and 2 cryptotephra layers dating between 0.4 and 4.5 ka. Distal ash in this record is chemically correlated by major element composition to at least three eruptions of the Boset Volcanic Complex and two eruptions of the Corbetti Volcanic System in the last 5 ka, providing the first absolute age control for recent explosive activity of Boset. Ashfall originating from the 1.3–1.9 ka VEI 5 Wendo Koshe Younger Pumice eruption of Corbetti is identified in three records allowing re-estimation of total erupted volume for this event with improved distal constraints. The total erupted volume is >1.3 times larger than previous estimates, dispersing ∼4.0 km3 (2.75–6.75 km3) of tephra over ∼90,000 km2, and predicts ashfall within range of Addis Ababa. An additional visible tephra preserved at Haro Kori, dating to 2.3 ka BP, indicates another large-magnitude eruption of Corbetti occurred within the last 2.5 ka. These findings demonstrate the value of distal lake records for improving the regional eruption record of the MER, providing crucial controls on the timing and magnitude of recent rift volcanism. Ash dispersal from central rift volcanism is capable of inundating populous areas across central Ethiopia; our work supports more informed preparation for future volcanic ash hazards in a region undergoing rapid development.</p
Testing stimulus generalisation as a mechanism for impression formation
Full text linkPeople readily judge trustworthiness based on others' facial appearance, but less is known about how our prior experiences shape who we find trustworthy. Stimulus generalisation is one mechanism which may explain how experience influences impressions of strangers. This fundamental learning principle argues that learning about one stimulus generalises to stimuli that resemble the original stimulus. Here, we asked whether stimulus generalisation, specifically, based on facial resemblance, can influence impressions of trustworthiness. We used a novel face resemblance paradigm to test whether naturally acquired attitudes held towards known individuals (celebrities) predict trustworthiness impressions of strangers' faces that were manipulated to resemble these identities. Across three studies (Total N = 283) and two countries (UK, Australia), we confirmed that pre-existing attitudes towards known individuals significantly predicted trustworthiness impressions of strangers' faces that merely resemble these individuals. Importantly, pre-existing attitudes remained significant after multi-level modelling accounted for variation in both facial appearance and participant differences. We found strong support for stimulus generalisation, demonstrating that social learning in the real world predicts individuals' impressions of trustworthiness. Therefore, impression formation involves integrating visual appearance with prior experiences to help us decide which people we trust. Our work demonstrates an important but neglected theoretical overlap between person perception, attitude formation, and learning principles.</p
DINOv3-Driven Semantic Segmentation for Landslide Mapping in Mountainous Regions
Full text linkLandslide hazard assessment increasingly demands the joint analysis of heterogeneous remote sensing data; however, automating this process remains difficult due to the pronounced resolution and texture discrepancies existing between satellite and aerial sensors. To address these limitations, this study proposes a robust segmentation framework capable of extracting sensor-robust representations. The framework leverages a DINOv3 transformer encoder and exploits representations from multiple transformer layers to capture complementary visual information, ranging from fine-grained surface textures to global semantic contexts, overcoming the receptive field constraints of conventional CNNs. Experiments on the Longxi satellite dataset achieve a Dice coefficient of 0.96 and an IoU of 0.938, and experiments on the Longxi UAV dataset achieve a Dice coefficient of 0.965 and an IoU of 0.941. These results show consistent segmentation performance on both the Longxi satellite and UAV datasets, despite differences in spatial resolution and surface appearance between acquisition platforms