Evaluation of individual and ensemble probabilistic forecasts of COVID-19 mortality in the United States

Short-term probabilistic forecasts of the trajectory of the COVID-19 pandemic in the United States have served as a visible and important communication channel between the scientific modeling community and both the general public and decision-makers. Forecasting models provide specific, quantitative, and evaluable predictions that inform short-term decisions such as healthcare staffing needs, school closures, and allocation of medical supplies. Starting in April 2020, the US COVID-19 Forecast Hub (https://covid19forecasthub.org/) collected, disseminated, and synthesized tens of millions of specific predictions from more than 90 different academic, industry, and independent research groups. A multimodel ensemble forecast that combined predictions from dozens of groups every week provided the most consistently accurate probabilistic forecasts of incident deaths due to COVID-19 at the state and national level from April 2020 through October 2021. The performance of 27 individual models that submitted complete forecasts of COVID-19 deaths consistently throughout this year showed high variability in forecast skill across time, geospatial units, and forecast horizons. Two-thirds of the models evaluated showed better accuracy than a naïve baseline model. Forecast accuracy degraded as models made predictions further into the future, with probabilistic error at a 20-wk horizon three to five times larger than when predicting at a 1-wk horizon. This project underscores the role that collaboration and active coordination between governmental public-health agencies, academic modeling teams, and industry partners can play in developing modern modeling capabilities to support local, state, and federal response to outbreaks.

Determination of acute exposure guideline levels in a dispersion model

The U.S. Environmental Protection Agency (EPA) is in the process of establishing acute exposure guideline levels (AEGLs) for a number of toxic chemicals. These guidelines represent predictable human consequences from measured exposures of selected toxic chemicals over time. They are intended for emergency planning and regulatory purposes. This paper presents a method that can be used by atmospheric dispersion models to compute AEGL values and graphically display the regions exposed to each guideline level on area maps. EPA currently defines three levels of AEGLs corresponding to increasingly severe symptoms, ranging from notable discomfort (AEGL-1) to serious adverse health effects (AEGL-2) to life-threatening effects or death (AEGL-3). For each chemical's three AEGL levels, guideline concentrations are defined for five exposure durations: 10 min, 30 min, 1 hr, 4 hr, and 8 hr. Dispersion models can compute a chemical dosage (time-integrated concentration) and a peak concentration throughout the area exposed to the plume. The AEGL then can be computed by finding the effective duration of the plume at each location, computing the AEGL-equivalent dosages for these durations, and comparing the dosage at each point with the AEGL-equivalent dosages. AEGL contours then can be plotted and readily interpreted in terms of expected toxicity levels for each level of health impact.

Atmospheric dispersion and deposition of 131I released from the Hanford Site

Approximately 2.6 x 10(4) TBq (700,000 Ci) of 131I were released to the air from reactor fuel processing plants on the Hanford Site in southcentral Washington State from December 1944 through December 1949. The Hanford Environmental Dose Reconstruction Project developed a suite of codes to estimate the doses that might have resulted from these releases. The Regional Atmospheric Transport Code for Hanford Emission Tracking (RATCHET) computer code is part of this suite. The RATCHET code implements a Lagrangian-trajectory, Gaussian-puff dispersion model that uses hourly meteorological and release rate data to estimate daily time-integrated air concentrations and surface contamination for use in dose estimates. In this model, iodine is treated as a mixture of three species (inorganic gases, organic gases, and particles). Model deposition parameters are functions of the mixture and meteorological conditions. A resistance model is used to calculate dry deposition velocities. Equilibrium between concentrations in the precipitation and the air near the ground is assumed in calculating wet deposition of gases, and irreversible washout of the particles is assumed. RATCHET explicitly treats the uncertainties in model parameters and meteorological conditions. Uncertainties in 131I release rates and partitioning among the nominal species are treated by varying model input. The results of 100 model runs for December 1944 through December 1949 indicate that monthly average air concentrations and deposition have uncertainties ranging from a factor of two near the center of the time-integrated plume to more than an order of magnitude near the edge. These results indicate that approximately 10% of the 131I released to the atmosphere decayed during transit in the study area, approximately 56% was deposited within the study area, and the remaining 34% was transported out of the study area while still in the air.