More Like this

1. Federated Graph Learning with Structure Proxy Alignment

   https://doi.org/10.1145/3637528.3671717  

   Fu, Xingbo; Chen, Zihan; Zhang, Binchi; Chen, Chen; Li, Jundong (August 2024, Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining)

   Federated Graph Learning (FGL) aims to learn graph learning models over graph data distributed in multiple data owners, which has been applied in various applications such as social recommendation and financial fraud detection. Inherited from generic Federated Learning (FL), FGL similarly has the data heterogeneity issue where the label distribution may vary significantly for distributed graph data across clients. For instance, a client can have the majority of nodes from a class, while another client may have only a few nodes from the same class. This issue results in divergent local objectives and impairs FGL convergence for node-level tasks, especially for node classification. Moreover, FGL also encounters a unique challenge for the node classification task: the nodes from a minority class in a client are more likely to have biased neighboring information, which prevents FGL from learning expressive node embeddings with Graph Neural Networks (GNNs). To grapple with the challenge, we propose FedSpray, a novel FGL framework that learns local class-wise structure proxies in the latent space and aligns them to obtain global structure proxies in the server. Our goal is to obtain the aligned structure proxies that can serve as reliable, unbiased neighboring information for node classification. To achieve this, FedSpray trains a global feature-structure encoder and generates unbiased soft targets with structure proxies to regularize local training of GNN models in a personalized way. We conduct extensive experiments over four datasets, and experiment results validate the superiority of FedSpray compared with other baselines. Our code is available at https://github.com/xbfu/FedSpray. 

   more » « less
2. Quantitative critique of leaf‐based paleo‐CO ₂ proxies: Consequences for their reliability and applicability

   https://doi.org/10.1002/gj.3807  

   Konrad, Wilfried; Royer, Dana L.; Franks, Peter J.; Roth‐Nebelsick, Anita (March 2020, Geological Journal)

   A variety of proxies have been developed to reconstruct paleo‐CO₂from fossil leaves. These proxies rely on some combination of stomatal morphology, leafδ¹³C, and leaf gas exchange. A common conceptual framework for evaluating these proxies is lacking, which has hampered efforts for inter‐comparison. Here we develop such a framework, based on the underlying physics and biochemistry. From this conceptual framework, we find that the more extensively parameterised proxies, such as the optimisation model, are likely to be the most robust. The simpler proxies, such as the stomatal ratio model, tend to under‐predict CO₂, especially in warm (>15^°C) and moist (>50%humidity) environments. This identification of a structural under‐prediction may help to explain the common observation that the simpler proxies often produce estimates of paleo‐CO₂that are lower than those from the more complex proxies and other, non‐leaf‐based CO₂proxies. The use of extensively parameterised models is not always possible, depending on the preservation state of the fossils and the state of knowledge about the fossil's nearest living relative. With this caveat in mind, our analysis highlights the value of using the most complex leaf‐based model as possible. 

   more » « less
3. Reconstructing Annual Δ ¹⁴ C During Miyake Events Using Deciduous and Evergreen Trees

   https://doi.org/10.1029/2024GB008423  

   Walker, M_R; Shobe, C_M; Andreu‐Hayles, L.; Dey, L.; Suran, B.; Nachin, B.; Hessl, A_E (June 2025, Global Biogeochemical Cycles)

   Abstract Cosmic rays and solar energetic particles pose significant risks to satellites, space stations, and human space exploration. They also produce atmospheric radiocarbon (¹⁴C), which enters the carbon cycle and is recorded by paleoenvironmental proxies. Miyake events, rapid increases in atmospheric¹⁴C, first identified in annual tree rings and later confirmed through ice core¹⁰Be and³⁶Cl isotopes, are thought to result from extreme solar activity, are seven events identified over the last 14,300 years. However, uncertainty in annual¹⁴C measurements limits precise inferences about their timing and magnitude. This study examines uncertainties in¹⁴C during two Miyake events (774 CE and 993 CE) across trees with differing uptake, storage, and allocation of carbon. We hypothesize that tree species physiology affects tree‐ring Δ¹⁴C, with deciduous species recording lagged, attenuated tree‐ring Δ¹⁴C relative to evergreen species. Using Δ¹⁴C data from pine and larch in Mongolia and a larger multi‐species Northern Hemisphere data set, we employed a Bayesian framework to estimate the timing, duration, and magnitude of these two events. Our AMS results showed no differences in Δ¹⁴C between evergreen and deciduous species growing at similar sites during the 774 CE event. The 993 CE event was variable, but parameter estimates were consistent between species. Northern Hemisphere comparisons indicated that annual series of Δ¹⁴C from evergreen and deciduous conifers yielded relatively more precise modeled estimates of start date and duration relative to deciduous broadleaf species. Future studies should consider the role of species‐specific carbon allocation strategies and storage dynamics in determining the radiocarbon response to Miyake events. 

   more » « less
4. Lithological & geochemical analysis of the Middle to Upper Devonian Antrim Shale, Michigan Basin: insights into detrital input dynamics

   Giehler, M; Gugino, J; Voice, P; Zambito, J (April 2024, Geological Society of America Abstracts with Programs)

   The Middle to Upper Devonian is distinguished by global biotic events attributed to marine anoxia (black shale) and perturbations in the carbon cycle. While previous studies had documented these events in the North American Appalachian Basin, less attention has been directed toward discerning analogous global biotic occurrences within the neighboring Michigan Basin. This investigation aims to reconstruct the organic carbon isotopic composition of Middle to Upper Devonian sediments within the Antrim Shale, from the Krocker 1-17 and State Chester Welch 18 drill cores located in the north-central Michigan Basin. Since black shale organic matter (OM) is a combination of marine and terrestrial sources, the first step is to utilize elemental proxies for sedimentation to understand terrestrial OM input to the basin. Elemental analysis was performed on powdered core samples at approximately three-foot intervals across the (oldest to youngest) Norwood, Paxton, and Lachine members of the Antrim Shale. Initial pXRF analysis shows similar trends in sedimentation rate proxies between the two core locations, suggesting that basin-fill dynamics are spatially consistent. In the Norwood, Si%, Si/Al, and K% mostly correspond suggesting predominantly clay mineral deposition, though an anomalously high interval of Si%, Si/Al, and TOC (and low K%) is interpreted as high paleoproductivity and biogenic silica production (radiolarians). In the Paxton, all elemental proxies for sedimentation are relatively consistent through time, indicating minimal detrital flux changes. In the Lachine, the overall trend of increasing Si% and Si/Al and corresponding decreasing K and TOC suggests increased detrital input through time. These trends provide the necessary baseline for future work interpreting organic carbon isotopic data. A detailed understanding of local detrital fluxes into the Michigan Basin and potential terrestrial organic carbon input to marine settings is necessary to place the Michigan Basin carbon isotopic profile within the global framework. This study will provide valuable insights into deciphering global events within the stratigraphic succession of the Michigan Basin. The anticipated outcome is an enhanced understanding of the localized manifestations of Middle to Upper Devonian global events. 

   more » « less
5. Microphysical and Kinematic Characteristics of Anomalous Charge Structure Thunderstorms in Cordoba, Argentina

   https://doi.org/10.3390/atmos13081329  

   Medina, Bruno; Carey, Lawrence; Deierling, Wiebke; Lang, Timothy (August 2022, Atmosphere)

   Some thunderstorms in Cordoba, Argentina, present a charge structure with an enhanced low-level positive charge layer, and practically nonexistent upper-level positive charge. Storms with these characteristics are uncommon in the United States, even when considering regions with a high frequency of anomalous charge structure storms such as Colorado. In this study, we explored the microphysical and kinematic conditions inferred by radar that led to storms with this unique low-level anomalous charge structure in Argentina, and compared them to conditions conducive for anomalous and normal charge structures. As high liquid water contents in the mixed-phase layer lead to positive charging of graupel and anomalous storms through the non-inductive charging mechanism, we explored radar parameters hypothesized to be associated with large cloud supercooled liquid water contents in the mixed-phase layer and anomalous storms, such as mass and volume of hail and high-density graupel, large reflectivity associated with the growth of rimed precipitation to hail size, and parameters that are proxies for strong updrafts such as echo-top and Zdr column heights. We found that anomalous storms had higher values of mass and volume of hail in multiple sub-layers of the mixed-phase zone and higher frequency of high reflectivity values. Low-level anomalous events had higher hail mass in the lower portion of the mixed-phase zone when compared to normal events. Weaker updraft proxies were found for low-level anomalous events due to the shallow nature of these events while there was no distinction between the updraft proxies of normal and anomalous storms. 

   more » « less