KDKatherine Dagon

Katherine Dagon

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Katherine Dagon has a total of 226 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Investigating western North America atmospheric rivers with machine learning-based detection

    by Katherine Dagon, Teagan King, Christine A Shields, John Truesdale, Sol Kim, Kirsten J Mayer, Andre Graubner, Annette Greiner, Lukas Kapp-Schwoerer, Karthik Kashinath

  • Open Access

    Investigating Western North America Atmospheric Rivers with Machine Learning-Based Detection

    by Katherine Dagon, Teagan King, Christine A. Shields, John Truesdale, Sol Kim, Kirsten J. Mayer, Andre Graubner, Annette Greiner, Lukas Kapp-Schwoerer, Karthik Kashinath

    Atmospheric rivers (ARs) are long, narrow conduits transporting water vapor poleward. ARs are often associated with extreme precipitation and can have significant impacts on humans and the environment when they make landfall. Here we use a previously developed deep learning model, CGNet, to detect ARs in high resolution Community Earth System Model (CESM) historical and future simulations and evaluate the results using multiple reanalysis products. CGNet is a convolutional neural network trained on expert-labeled AR and tropical cyclone masks from the ClimateNet project. We probe the sensitivities of CGNet by testing how different characteristics of the input (variables, normalization values, spatial and temporal resolution) impact the resulting AR detection. Global and regional AR frequency magnitudes are strongly dependent on source data. To understand future changes, we compare features of the detected ARs across CESM simulations using historical and Representative Concentration Pathway (RCP) 8.5 forcing. Focusing on western North America, we find that ARs largely increase over California and decrease over the Pacific Northwest and western Canada in response to projected warming. We leverage built-in spatiotemporal tracking with CGNet to connect regional changes in ARs with shifts in lifecycle characteristics. We further investigate precipitation associated with ARs and find that AR precipitation increases in California are largely driven by changes in AR precipitation intensity, whereas AR frequency provides an important contribution to AR precipitation changes at more northerly latitudes. AR-associated extreme precipitation is projected to increase over much of the region, with important implications for impacts.

  • Open Access

    Efficient Emulation, Uncertainty Quantification, and Sensitivity Analysis for a Land Surface Model Using Evidential Deep Learning

    by Kachinga Silwimba, Alejandro N. Flores, Linnia R. Hawkins, Charles Becker, Katherine Dagon, David John Gagne, Jodi Mead, Daniel Kennedy, Irene Cionni, Stanley Akor, et al.

    Land surface models (LSMs), such as the Community Land Model version 5 (CLM5), represent complex vegetation processes; however, systematic biases persist between modeled and observed leaf area index (LAI) because of parameter uncertainties and knowledge gaps. The high computational cost of CLM5 is a barrier to extensive sensitivity analyses and ensemble simulations at the global scale. This study addresses these limitations by training an evidential deep neural network (EDNN) emulator on a 500‐member CLM5 perturbed parameter ensemble generated via Latin hypercube sampling of 32 key plant physiological parameters. The EDNN employs cyclic temporal encoding to preserve seasonal periodicity and to predict LAI anomalies, thereby emphasizing variability. It also quantifies predictive uncertainty by jointly learning aleatoric and epistemic components in a single forward pass, yielding well‐calibrated probabilistic outputs without requiring computationally intensive ensembles. Across the contiguous United States, the EDNN reproduces CLM5‐simulated LAI with a median on held‐out members and years while requiring substantially less computation. The EDNN emulator captures seasonal cycles, interannual variability in LAI, and uncertainties (aleatoric and epistemic) in a single pass. The sensitivity analysis highlights photosynthetic capacity and the leaf carbon‐to‐nitrogen ratio as dominant controls on LAI variability, with seasonal shifts in their influence reflecting phenological dynamics. These capabilities enable comprehensive parameter‐sensitivity studies and more efficient calibration and tuning of land surface models by supporting rapid probabilistic forecasting and adaptive model refinement, thereby paving the way for scalable Earth‐system modeling, robust parameter exploration, and uncertainty‐informed projections of land‐surface processes.

  • Open Access

    Investigating Western North America Atmospheric Rivers with Machine Learning-Based Detection

    by Katherine Dagon, Teagan King, Christine A. Shields, John Truesdale, Sol Kim, Kirsten Mayer, Andre Graubner, Annette Greiner, Lukas Kapp-Schwoerer, Karthik Kashinath

    Atmospheric rivers (ARs) are long, narrow conduits transporting water vapor poleward. ARs are often associated with extreme precipitation and can have significant impacts on humans and the environment when they make landfall. Here we use the deep learning model CGNet to detect ARs in high resolution Community Earth System Model (CESM) simulations and evaluate the results using multiple reanalysis products. We probe the sensitivities of CGNet by testing how different characteristics of the input (variables, normalization values, spatial and temporal resolution) impact the resulting AR detection. Global and regional AR frequency magnitudes are strongly dependent on source data. To understand future changes, we compare features of the detected ARs across CESM simulations using historical and Representative Concentration Pathway (RCP) 8.5 forcing. Focusing on western North America, we find that ARs largely increase over California and decrease over the Pacific Northwest and western Canada in response to projected warming. We leverage built-in spatiotemporal tracking with CGNet to connect regional changes in ARs with shifts in lifecycle characteristics. We further investigate precipitation associated with ARs and find that AR precipitation increases in California are largely driven by changes in AR precipitation intensity, whereas AR frequency provides an important contribution to AR precipitation changes at more northerly latitudes. AR-associated extreme precipitation is projected to increase over much of the region, with important implications for impacts.

  • Open Access

    Evidential Deep Learning Emulation of Monthly Streamflow Anomalies From Land Surface Model Perturbed-Parameter Ensembles in Snow-Dominated Basins

    by Kachinga Silwimba, Alejandro N. Flores, Linnia R Hawkins, Katherine Dagon, Charles Becker, David John Gagne, Irene Cionni, Jodi Mead, Andrew W. Wood, Guoqiang Tang, et al.

  • Open Access

    Contrasting Parametric Sensitivities in Two Global Vegetation Models Using Parameter Perturbation Ensembles

    by A. C. Foster, L. R. Hawkins, D. Kennedy, G. B. Bonan, R. A. Fisher, J. F. Needham, R. G. Knox, C. D. Koven, W. R. Wieder, K. Dagon, et al.

    Uncertainty in land model projections remains high and the roles of parametric and structural uncertainty are difficult to disentangle. To compare parametric sensitivity across model structures we present two parameter perturbation ensembles using the Community Land Model (CLM) operating in satellite phenology mode. The ensembles contrast two vegetation modules: (a) the default CLM vegetation module and (b) the Functionally Assembled Terrestrial Ecosystem Simulator (CLM‐FATES). We perturbed over 300 parameters and quantified their effects on biophysical fluxes globally and across biomes. Most parameters have minimal impact on biophysical fluxes, with only a few substantially influencing results. While both models exhibit similar parameter sensitivity for some fluxes, CLM‐FATES shows larger spread in gross primary productivity (GPP), driven by strong sensitivity to carboxylation rate. CLM‐FATES also shows a weaker GPP response to soil hydrology parameters and exhibits higher water use efficiency (WUE). Cross‐model comparisons reveal similar sensitivities for some parameters (e.g., leaf dimension) but divergent responses to others (e.g., stomatal intercept), highlighting underlying structural differences. Differences in WUE and sensitivity to hydrology and stomatal conductance parameters underscore how model structure fundamentally alters parametric sensitivity. The data sets generated from these ensembles can be used to identify influential parameters and guide future calibration efforts.


  • Open Access

    Constraining climate model projections with observations amplifies future runoff declines

    by Hanjun Kim, Flavio Lehner, Katherine Dagon, David M. Lawrence, Sean Swenson, Andrew W. Wood


  • Open Access

    Diagnosing Parametric Controls on CLM Ecohydrology Using Machine Learning Emulators

    by Sarah Ryu, Linnia Hawkins, Katherine Dagon, Sudhanshu Kumar, Isabella Lu

    Accurate representation of land-atmosphere feedbacks is essential for improving the predictive skill of Earth system models on policy relevant timescales. A key component of these feedbacks is the coupling between soil moisture (SM) and evapotranspiration (ET), which controls surface water and energy exchanges. In this study, we examine whether SM–ET coupling in the Community Land Model (CLM6) can be improved through parameter tuning. To explore parametric effects on coupling, we developed a machine learning emulator that reproduces CLM output at a fraction of the computational cost. We trained a multi-layer perceptron on simulations from a CLM perturbed parameter ensemble to emulate annual root-zone SM and ET across space, time, and the parameter space of 56 plant traits and soil properties. The emulator accurately identifies regions where CLM underestimates SM–ET coupling relative to accepted benchmarks. Coupling strength responds differently to parameters across climate regimes: plant traits (e.g., maximum hydraulic conductance) control coupling in warm, dry regions, while soil traits (e.g., saturated volumetric water content) dominate in cold, wet areas. Arid regions show potential for improving coupling direction through parameter changes. In contrast, tropical forests exhibit persistent coupling biases that are insensitive to parameters, suggesting structural errors linked to limited plant functional types in biodiverse ecoregions and poor representation of complex canopies. Findings from this work can inform targeted model development and demonstrate how perturbed parameter ensembles and spatiotemporal emulators can be used to disentangle parametric versus structural sources of model error.

  • Open Access

    Scalable, Uncertainty-Aware Streamflow Prediction in Snow-Dominated Catchments Using Evidential Deep Learning and CLM5 Ensembles

    by Kachinga Silwimba, Alejandro N. Flores, Linnia Hawkins, Charles Becker, Katherine Dagon, David John Gagne, Irene Cionni, Jodi Mead, Stanley Akor, Pamela L Sullivan, et al.

    Accurate streamflow prediction in snow-dominated, ungauged basins remains a persistent challenge in hydrology due to strong seasonal variability, sparse observations, and complex topography. Predicting streamflow in these regions is crucial for improving water resource management and climate adaptation. In this study, we employed a model emulation approach that combines physically-based hydrologic ensemble simulations with Evidential Deep Neural Networks (EDNNs) to deliver both accurate predictions and uncertainty estimates that are statistically consistent with observations. We trained EDNNs on simulated streamflow generated by ensemble simulations of the Community Land Model version 5 (CLM5), where each ensemble member (EM) represents a different parameter configuration designed to sample parametric uncertainty. We identified snow-dominated basins among the 627 catchments from the CAMELS (Catchment Attributes and Meteorology for Large-sample Studies) dataset in the contiguous United States using hydroclimatic thresholds, followed by unsupervised clustering based on seasonal streamflow signatures and physiographic descriptors. Initially, we trained the EDNN on simulated outputs from a single CLM5 EM in a basin cluster and tested its generalization to other clusters and parameter configurations. Subsequently, we incrementally increased the training dataset by progressively adding five additional EMs and approximately five catchments at each increment, without altering the EDNN architecture, to evaluate how increasing training data influenced predictive accuracy (e.g., Nash–Sutcliffe Efficiency) and uncertainty calibration. Our findings revealed that the EDNN successfully captured streamflow dynamics while decomposing predictive uncertainty into epistemic (model-driven) and aleatoric (data-driven) components. Both uncertainty calibration and predictive accuracy improved substantially as more EMs and catchments were included, underlining the benefits of physics-based ensemble training data in sparsely observed regions. This approach offers a scalable framework for streamflow prediction in snow-dominated basins, reduces the computational costs of ensemble-based calibration efforts, and enables parameter transfer to out-of-sample watersheds.

  • Open Access

    Observed and Modeled Amplification of the Frequency, Duration, and Extreme Heat Impacts of the Pacific Trough Regime

    by Jhayron S. Pérez‐Carrasquilla, Maria J. Molina, Kirsten J. Mayer, Katherine Dagon, John T. Fasullo, Isla R. Simpson

    The large‐scale atmospheric circulation is a key driver for regional climate extremes, yet its response to anthropogenic forcing remains uncertain. The Pacific trough (PT) regime is a persistent circulation pattern modulating temperature, precipitation, and fires over North America. We show that the observed boreal winter‐spring (December to May) PT frequency and duration have increased significantly over the past 76 years, contributing to amplified extreme anomalous heat over western and central Canada. These observed changes are not well represented in the climate simulations analyzed herein. However, our results indicate that rising greenhouse gas concentrations likely contribute to increased winter‐spring PT frequency, which is further modulated by sea surface temperatures (SSTs). While the recent La Niña‐like and negative Pacific Decadal Oscillation‐like SST trends have dampened this increase, our results suggest that if an eventual emergence of the modeled El Niño‐like response to elevated were to occur in reality, it would reinstate the increase in PT frequency, duration, and downstream amplification of regional extreme heat. However, the occurrence, timing, and magnitude of this shift remain uncertain, given the complex, interlaced role of external forcings and internal variability in modulating historical trends, as well as models' inability to reproduce them. Additionally, modeling decisions regarding future trajectories for anthropogenic emissions, including aerosols and greenhouse gases, play a critical role in projecting future changes in PT frequency. Our findings underscore the need for a better understanding and modeled representation of long‐term changes in the atmospheric circulation to inform climate adaptation and risk assessment.