arXiv subjects

Ellen M. Considine

Publications and source records attributed to Ellen M. Considine.

4 recordsLinked to original sources

arXiv2026

A Spatiotemporal, Quasi-experimental Causal Inference Approach to Characterize the Effects of Global Plastic Waste Export and Burning on Air Quality Using Remotely Sensed Data

Open burning of plastic waste may pose a significant threat to global health by degrading air quality, but quantitative research on this problem -- crucial for policy making -- has been stunted by lack of data. Many low- and middle-income countries, where open burning is most concerning, have little to no air quality monitoring. Here, we leverage remotely sensed data products combined with spatiotemporal causal analytic techniques to evaluate the impact of large-scale plastic waste policies on air quality. Throughout, we study Indonesia before and after 2018, when China halted its import of plastic waste, resulting in diversion of this massive waste stream to other countries. We tailor cutting-edge statistical methods to this setting, estimating effects of increased plastic waste imports on fine particulate matter (PM$_{2.5}$) near waste dump sites in Indonesia as a function of proximity to ports, an induced continuous exposure. We observe strong evidence that monthly PM$_{2.5}$increased after China's ban (2018-2019) relative to expected business-as-usual (2012-2017), with increases up to 1.68 $μ$g/m$^3$ (95% CI = [0.72, 2.48]) when exposed to medium-high port proximity. Effects were more modest for very high port proximity exposure, possibly reflecting smaller increases in dumping/burning where government oversight is greater.

stat.AP↗

arXiv2024

Optimizing Heat Alert Issuance with Reinforcement Learning

A key strategy in societal adaptation to climate change is using alert systems to prompt preventative action and reduce the adverse health impacts of extreme heat events. This paper implements and evaluates reinforcement learning (RL) as a tool to optimize the effectiveness of such systems. Our contributions are threefold. First, we introduce a new publicly available RL environment enabling the evaluation of the effectiveness of heat alert policies to reduce heat-related hospitalizations. The rewards model is trained from a comprehensive dataset of historical weather, Medicare health records, and socioeconomic/geographic features. We use scalable Bayesian techniques tailored to the low-signal effects and spatial heterogeneity present in the data. The transition model uses real historical weather patterns enriched by a data augmentation mechanism based on climate region similarity. Second, we use this environment to evaluate standard RL algorithms in the context of heat alert issuance. Our analysis shows that policy constraints are needed to improve RL's initially poor performance. Third, a post-hoc contrastive analysis provides insight into scenarios where our modified heat alert-RL policies yield significant gains/losses over the current National Weather Service alert policy in the United States.

cs.LG↗

arXiv2023

Evaluation of Model-Based PM$_{2.5}$ Estimates for Exposure Assessment During Wildfire Smoke Episodes in the Western U.S

Investigating the health impacts of wildfire smoke requires data on people's exposure to fine particulate matter (PM$_{2.5}$) across space and time. In recent years, it has become common to use machine learning models to fill gaps in monitoring data. However, it remains unclear how well these models are able to capture spikes in PM$_{2.5}$ during and across wildfire events. Here, we evaluate the accuracy of two sets of high-coverage and high-resolution machine learning-derived PM$_{2.5}$ data sets created by Di et al. (2021) and Reid et al. (2021). In general, the Reid estimates are more accurate than the Di estimates when compared to independent validation data from mobile smoke monitors deployed by the US Forest Service. However, both models tend to severely under-predict PM$_{2.5}$ on high-pollution days. Our findings complement other recent studies calling for increased air pollution monitoring in the western US and support the inclusion of wildfire-specific monitoring observations and predictor variables in model-based estimates of PM$_{2.5}$. Lastly, we call for more rigorous error quantification of machine-learning derived exposure data sets, with special attention to extreme events.

stat.AP↗

arXiv2023

Investigating Use of Low-Cost Sensors to Increase Accuracy and Equity of Real-Time Air Quality Information

Environmental Protection Agency (EPA) air quality (AQ) monitors, the gold standard for measuring air pollutants, are sparsely positioned across the US due to their costliness. Low-cost sensors (LCS) are increasingly being used by the public to fill in the gaps in AQ monitoring; however, LCS are not as accurate as EPA monitors. In this work, we investigate factors impacting the differences between an individual's true (unobserved) exposure to fine particulate matter (PM2.5) and the exposure reported by their nearest AQ instrument, which could be either an EPA monitor or an LCS. Three factors contributing to these differences are (1) distance to the nearest AQ instrument, (2) local variability in AQ, and (3) device measurement error. We examine the contributions of each component to the overall error in reported AQ using simulations based on California data. The simulations explore different combinations of hypothetical LCS placement strategies (at schools, near major roads, and in environmentally and socioeconomically marginalized census tracts) for different numbers of LCS, with varying plausible amounts of LCS device measurement error. For each scenario, we evaluate the accuracy of daily AQ information available from individuals' nearest AQ instrument with respect to absolute errors and misclassifications of the Air Quality Index, stratified by socioeconomic and demographic characteristics. We illustrate how real-time AQ reporting could be improved (or, in some cases, worsened) by using LCS, both for the population overall and for marginalized communities specifically. This work has implications for the integration of LCS into real-time AQ reporting platforms.

stat.AP↗