Search for: All records

In the event of a nuclear accident, or the detonation of a radiological dispersal device, quickly locating the source of the accident or blast is important for emergency response and environmental decontamination. At a specified time after a simulated instantaneous release of an aerosolized radioactive contaminant, measurements are recorded downwind from an array of radiation sensors. Neural networks are employed to infer the source release parameters in an accurate and rapid manner using sensor and mean wind speed data. We consider two neural network constructions that quantify the uncertainty of the predicted values; a categorical classification neural network and a Bayesian neural network. With the categorical classification neural network, we partition the spatial domain and treat each partition as a separate class for which we estimate the probability that it contains the true source location. In a Bayesian neural network, the weights and biases have a distribution rather than a single optimal value. With each evaluation, these distributions are sampled, yielding a different prediction with each evaluation. The trained Bayesian neural network is thus evaluated to construct posterior densities for the release parameters. Results are compared to Markov chain Monte Carlo (MCMC) results found using the Delayed Rejection Adaptive Metropolis Algorithm. The Bayesian neural network approach is generally much cheaper computationally than the MCMC approach as it relies on the computational cost of the neural network evaluation to generate posterior densities as opposed to the MCMC approach which depends on the computational expense of the transport and radiation detection models. 

Abstract Measuring, reporting, and verification (MRV) of ocean-based carbon dioxide removal (CDR) presents challenges due to the dynamic nature of the ocean and the complex processes influencing marine carbonate chemistry. Given these challenges, finding the optimal sampling strategies and suite of parameters to be measured is a timely research question. While traditional carbonate parameters such as total alkalinity (TA), dissolved inorganic carbon (DIC), pH, and seawater pCO2 are commonly considered, exploring the potential of carbon isotopes for quantifying additional CO2 uptake remains a relatively unexplored research avenue. In this study, we use a coupled physical-biogeochemical model of the California Current System (CCS) to run a suite of Ocean Alkalinity Enhancement (OAE) simulations. The physical circulation for the CCS is generated using a nested implementation of the Regional Ocean Modeling System (ROMS) with an outer domain of 1/10 ̊ (~10 km) and an inner domain of 1/30 ̊ (~3 km) resolution. The biogeochemical model, NEMUCSC, is a customized version of the North Pacific Ecosystem Model for Understanding Regional Oceanography (NEMURO) that includes carbon cycling and carbon isotopes. The CCS is one of four global eastern boundary upwelling systems characterized by high biological activity and CO2 concentrations. Consequently, the CCS represents an essential test case for investigating the efficacy and potential side effects of OAE deployments. The study aims to address two key questions: (1) the relative merit of OAE to counter ocean acidification versus the additional sequestration of CO2 from the atmosphere, and (2) the footprint of potentially harmful seawater chemistry adjacent to OAE deployments. We plan to leverage these high-resolution model results to competitively evaluate different MRV strategies, with a specific focus on analyzing the spatiotemporal distribution of carbon isotopic signatures following OAE. In this talk, we will showcase our initial results and discuss challenges in integrating high-resolution regional modeling into models of the global carbon cycle. More broadly, this work aims to provide insights into the plausibility of OAE as a climate solution that maintains ocean health and to inform accurate quantification of carbon uptake for MRV purposes. https://agu.confex.com/agu/OSM24/prelim.cgi/Paper/1491096 

Hyperspectral unmixing is an important remote sensing task with applications including material identification and analysis. Characteristic spectral features make many pure materials identifiable from their visible-to-infrared spectra, but quantifying their presence within a mixture is a challenging task due to nonlinearities and factors of variation. In this paper, spectral variation is considered from a physics-based approach and incorporated into an end-to-end spectral unmixing algorithm via differentiable programming. The dispersion model is introduced to simulate realistic spectral variation, and an efficient method to fit the parameters is presented. Then, this dispersion model is utilized as a generative model within an analysis-by-synthesis spectral unmixing algorithm. Further, a technique for inverse rendering using a convolutional neural network to predict parameters of the generative model is introduced to enhance performance and speed when training data is available. Results achieve state-of-the-art on both infrared and visible-to-near-infrared (VNIR) datasets, and show promise for the synergy between physicsbased models and deep learning in hyperspectral unmixing in the future.