More Like this

1. The grain size of sediments delivered to steep debris‐flow prone channels prior to and following wildfire

   https://doi.org/10.1002/esp.5819  

   Neely, Alexander_B; Moon, Seulgi; DiBiase, Roman_A; Sklar, Leonard_S; Argueta, Marina_O (April 2024, Earth Surface Processes and Landforms)

   Abstract Debris flows are powered by sediment supplied from steep hillslopes where soils are often patchy and interrupted by bare‐bedrock cliffs. The role of patchy soils and cliffs in supplying sediment to channels remains unclear, particularly surrounding wildfire disturbances that heighten debris‐flow hazards by increasing sediment supply to channels. Here, we examine how variation in soil cover on hillslopes affects sediment sizes in channels surrounding the 2020 El Dorado wildfire, which burned debris‐flow prone slopes in the San Bernardino Mountains, California. We focus on six headwater catchments (<0.1 km²) where hillslope sources ranged from a continuous soil mantle to 95% bare‐bedrock cliffs. At each site, we measured sediment grain size distributions at the same channel locations before and immediately following the wildfire. We compared results to a mixing model that accounts for three distinct hillslope sediment sources distinguished by local slope thresholds. We find that channel sediment in fully soil‐mantled catchments reflects hillslope soils (D₅₀ = 0.1–0.2 cm) both before and after the wildfire. In steeper catchments with cliffs, channel sediment is consistently coarse prior to fire (D₅₀ = 6–32 cm) and reflects bedrock fracture spacing, despite cliffs representing anywhere from 5% to 95% of the sediment source area. Following the fire, channel sediment size reduces most (5‐ to 20‐fold) in catchments where hillslope sources are predominantly soil covered but with patches of cliffs. The abrupt fining of channel sediment is thought to facilitate postfire debris‐flow initiation, and our results imply that this effect is greatest where bare‐bedrock cliffs are present but not dominant. A patchwork of bare‐bedrock cliffs is common in steeplands where hillslopes respond to channel incision by landsliding. We show how local slope thresholds applied to such terrain aid in estimating sediment supply conditions before two destructive debris flows that eventually nucleated in these study catchments in 2022. 

   more » « less
2. Univariate classification of phosphine ligation state and reactivity in cross-coupling catalysis

   https://doi.org/10.1126/science.abj4213  

   Newman-Stonebraker, Samuel H.; Smith, Sleight R.; Borowski, Julia E.; Peters, Ellyn; Gensch, Tobias; Johnson, Heather C.; Sigman, Matthew S.; Doyle, Abigail G. (October 2021, Science)

   Chemists often use statistical analysis of reaction data with molecular descriptors to identify structure-reactivity relationships, which can enable prediction and mechanistic understanding. In this study, we developed a broadly applicable and quantitative classification workflow that identifies reactivity cliffs in 11 Ni- and Pd-catalyzed cross-coupling datasets using monodentate phosphine ligands. A distinctive ligand steric descriptor, minimum percent buried volume [% V bur (min)], is found to divide these datasets into active and inactive regions at a similar threshold value. Organometallic studies demonstrate that this threshold corresponds to the binary outcome of bisligated versus monoligated metal and that % V bur (min) is a physically meaningful and predictive representation of ligand structure in catalysis. 

   more » « less
3. Stream hydrology controls on ice cliff evolution and survival on debris-covered glaciers

   Petersen, E; Hock, R; Loso, M (May 2024, Earth Surface Dynamics)

   Ice cliffs are melt hot spots that contribute disproportionately to melt on debris-covered glaciers. In this study, we investigate the impact of supraglacial stream hydrology on ice cliffs using in situ and remote sens- ing observations, streamflow measurements, and a conceptual geomorphic model of ice cliff backwasting applied to ice cliffs on Kennicott Glacier, Alaska. We found that 33 % of ice cliffs (accounting for 69 % of the ice cliff area) are actively influenced by streams, while half are nearer than 10 m from the nearest stream. Supraglacial streams contribute to ice cliff formation and maintenance by horizontal meandering, vertical incision, and de- bris transport. These processes produce an undercut lip at the ice cliff base and transport clasts up to tens of centimeters in diameter, preventing reburial of ice cliffs by debris. Stream meander morphology reminiscent of sedimentary river channel meanders and oxbow lakes produces sinuous and crescent ice cliff shapes. Stream avulsions result in rapid ice cliff collapse and local channel abandonment. Ice cliffs abandoned by streams are observed to be reburied by supraglacial debris, indicating a strong role played by streams in ice cliff persistence. We also report on a localized surge-like event at the glacier’s western margin which drove the formation of ice cliffs from crevassing; these cliffs occur in sets with parallel linear morphologies contrasting with the crescent planform shape of stream-driven cliffs. The development of landscape evolution models may assist in quanti- fying the total net effect of these processes on steady-state ice cliff coverage and mass balance, contextualizing them with other drivers including supraglacial ponds, differential melt, ice dynamics, and collapse of englacial voids. 

   more » « less
4. Less is More: Culling the Training Set to Improve Robustness of Deep Neural Networks

   https://doi.org/http://doi.org/10.1007/978-3-030-01554-1_6  

   Liu, Yongshuai; Chen, Jiyu; Chen, Hao (September 2018, International Conference on Decision and Game Theory for Security)

   Deep neural networks are vulnerable to adversarial examples. Prior defenses attempted to make deep networks more robust by either changing the network architecture or augmenting the training set with adversarial examples, but both have inherent limitations. Motivated by recent research that shows outliers in the training set have a high negative influence on the trained model, we studied the relationship between model robustness and the quality of the training set. We first show that outliers give the model better generalization ability but weaker robustness. Next, we propose an adversarial example detection framework, in which we design two methods for removing outliers from training set to obtain the sanitized model and then detect adversarial example by calculating the difference of outputs between the original and the sanitized model. We evaluated the framework on both MNIST and SVHN. Based on the difference measured by Kullback-Leibler divergence, we could detect adversarial examples with accuracy between 94.67% to 99.89%. 

   more » « less
5. Robust Differential Abundance Analysis of Microbiome Sequencing Data

   https://doi.org/10.3390/genes14112000  

   Li, Guanxun; Yang, Lu; Chen, Jun; Zhang, Xianyang (November 2023, Genes)

   It is well known that the microbiome data are ridden with outliers and have heavy distribution tails, but the impact of outliers and heavy-tailedness has yet to be examined systematically. This paper investigates the impact of outliers and heavy-tailedness on differential abundance analysis (DAA) using the linear models for the differential abundance analysis (LinDA) method and proposes effective strategies to mitigate their influence. The presence of outliers and heavy-tailedness can significantly decrease the power of LinDA. We investigate various techniques to address outliers and heavy-tailedness, including generalizing LinDA into a more flexible framework that allows for the use of robust regression and winsorizing the data before applying LinDA. Our extensive numerical experiments and real-data analyses demonstrate that robust Huber regression has overall the best performance in addressing outliers and heavy-tailedness. 

   more » « less