Machine learning-based extreme event attribution.

The observed increase in extreme weather has prompted recent methodological advances in extreme event attribution. We propose a machine learning-based approach that uses convolutional neural networks to create dynamically consistent counterfactual versions of historical extreme events under different levels of global mean temperature (GMT). We apply this technique to one recent extreme heat event (southcentral North America 2023) and several historical events that have been previously analyzed using established attribution methods. We estimate that temperatures during the southcentral North America event were 1.18° to 1.42°C warmer because of global warming and that similar events will occur 0.14 to 0.60 times per year at 2.0°C above preindustrial levels of GMT. Additionally, we find that the learned relationships between daily temperature and GMT are influenced by the seasonality of the forced temperature response and the daily meteorological conditions. Our results broadly agree with other attribution techniques, suggesting that machine learning can be used to perform rapid, low-cost attribution of extreme events.

Tree species explain only half of explained spatial variability in plant water sensitivity.

Spatiotemporal patterns of plant water uptake, loss, and storage exert a first-order control on photosynthesis and evapotranspiration. Many studies of plant responses to water stress have focused on differences between species because of their different stomatal closure, xylem conductance, and root traits. However, several other ecohydrological factors are also relevant, including soil hydraulics, topographically driven redistribution of water, plant adaptation to local climatic variations, and changes in vegetation density. Here, we seek to understand the relative importance of the dominant species for regional-scale variations in woody plant responses to water stress. We map plant water sensitivity (PWS) based on the response of remotely sensed live fuel moisture content to variations in hydrometeorology using an auto-regressive model. Live fuel moisture content dynamics are informative of PWS because they directly reflect vegetation water content and therefore patterns of plant water uptake and evapotranspiration. The PWS is studied using 21,455 wooded locations containing U.S. Forest Service Forest Inventory and Analysis plots across the western United States, where species cover is known and where a single species is locally dominant. Using a species-specific mean PWS value explains 23% of observed PWS variability. By contrast, a random forest driven by mean vegetation density, mean climate, soil properties, and topographic descriptors explains 43% of observed PWS variability. Thus, the dominant species explains only 53% (23% compared to 43%) of explainable variations in PWS. Mean climate and mean NDVI also exert significant influence on PWS. Our results suggest that studies of differences between species should explicitly consider the environments (climate, soil, topography) in which observations for each species are made, and whether those environments are representative of the entire species range.

Drought impacts on the electricity system, emissions, and air quality in the western United States.

The western United States has experienced severe drought in recent decades, and climate models project increased drought risk in the future. This increased drying could have important implications for the region's interconnected, hydropower-dependent electricity systems. Using power-plant level generation and emissions data from 2001 to 2021, we quantify the impacts of drought on the operation of fossil fuel plants and the associated impacts on greenhouse gas (GHG) emissions, air quality, and human health. We find that under extreme drought, electricity generation from individual fossil fuel plants can increase up to 65% relative to average conditions, mainly due to the need to substitute for reduced hydropower. Over 54% of this drought-induced generation is transboundary, with drought in one electricity region leading to net imports of electricity and thus increased pollutant emissions from power plants in other regions. These drought-induced emission increases have detectable impacts on local air quality, as measured by proximate pollution monitors. We estimate that the monetized costs of excess mortality and GHG emissions from drought-induced fossil generation are 1.2 to 2.5x the reported direct economic costs from lost hydro production and increased demand. Combining climate model estimates of future drying with stylized energy-transition scenarios suggests that these drought-induced impacts are likely to remain large even under aggressive renewables expansion, suggesting that more ambitious and targeted measures are needed to mitigate the emissions and health burden from the electricity sector during drought.

Exploring the Influence of Summer Temperature on Human Mobility During the COVID-19 Pandemic in the San Francisco Bay Area.

Studies on the relationship between temperature and local, small scale mobility are limited, and sensitive to the region and time period of interest. We contribute to the growing mobility literature through a detailed characterization of the observed temperature-mobility relationship in the San Francisco Bay Area at fine spatial and temporal scale across two summers (2020-2021). We used anonymized cellphone data from SafeGraph's neighborhood patterns data set and gridded temperature data from gridMET, and analyzed the influence of incremental changes in temperature on mobility rate (i.e., visits per capita) using a panel regression with fixed effects. This strategy enabled us to control for spatial and temporal variability across the studied region. Our analysis suggested that all areas exhibited lower mobility rate in response to higher summer temperatures. We then explored how several additional variables altered these results. Extremely hot days resulted in faster mobility declines with increasing temperatures. Weekdays were often more resistant to temperature changes when compared to the weekend. In addition, the rate of decrease in mobility in response to high temperature was significantly greater among the wealthiest census block groups compared with the least wealthy. Further, the least mobile locations experienced significant differences in mobility response compared to the rest of the data set. Given the fundamental differences in the mobility response to temperature across most of our additive variables, our results are relevant for future mobility studies in the region.

Data-driven predictions of the time remaining until critical global warming thresholds are reached.

Leveraging artificial neural networks (ANNs) trained on climate model output, we use the spatial pattern of historical temperature observations to predict the time until critical global warming thresholds are reached. Although no observations are used during the training, validation, or testing, the ANNs accurately predict the timing of historical global warming from maps of historical annual temperature. The central estimate for the 1.5 °C global warming threshold is between 2033 and 2035, including a ±1σ range of 2028 to 2039 in the Intermediate (SSP2-4.5) climate forcing scenario, consistent with previous assessments. However, our data-driven approach also suggests a substantial probability of exceeding the 2 °C threshold even in the Low (SSP1-2.6) climate forcing scenario. While there are limitations to our approach, our results suggest a higher likelihood of reaching 2 °C in the Low scenario than indicated in some previous assessments-though the possibility that 2 °C could be avoided is not ruled out. Explainable AI methods reveal that the ANNs focus on particular geographic regions to predict the time until the global threshold is reached. Our framework provides a unique, data-driven approach for quantifying the signal of climate change in historical observations and for constraining the uncertainty in climate model projections. Given the substantial existing evidence of accelerating risks to natural and human systems at 1.5 °C and 2 °C, our results provide further evidence for high-impact climate change over the next three decades.

Potential for perceived failure of stratospheric aerosol injection deployment.

As anthropogenic activities warm the Earth, the fundamental solution of reducing greenhouse gas emissions remains elusive. Given this mitigation gap, global warming may lead to intolerable climate changes as adaptive capacity is exceeded. Thus, there is emerging interest in solar radiation modification, which is the process of deliberately increasing Earth's albedo to cool the planet. Stratospheric aerosol injection (SAI)-the theoretical deployment of particles in the stratosphere to enhance reflection of incoming solar radiation-is one strategy to slow, pause, or reverse global warming. If SAI is ever pursued, it will likely be for a specific aim, such as affording time to implement mitigation strategies, lessening extremes, or reducing the odds of reaching a biogeophysical tipping point. Using an ensemble climate model experiment that simulates the deployment of SAI in the context of an intermediate greenhouse gas trajectory, we quantified the probability that internal climate variability masks the effectiveness of SAI deployment on regional temperatures. We found that while global temperature was stabilized, substantial land areas continued to experience warming. For example, in the SAI scenario we explored, up to 55% of the global population experienced rising temperatures over the decade following SAI deployment and large areas exhibited high probability of extremely hot years. These conditions could cause SAI to be perceived as a failure. Countries with the largest economies experienced some of the largest probabilities of this perceived failure. The potential for perceived failure could therefore have major implications for policy decisions in the years immediately following SAI deployment.

Plant-water sensitivity regulates wildfire vulnerability.

Extreme wildfires extensively impact human health and the environment. Increasing vapour pressure deficit (VPD) has led to a chronic increase in wildfire area in the western United States, yet some regions have been more affected than others. Here we show that for the same increase in VPD, burned area increases more in regions where vegetation moisture shows greater sensitivity to water limitation (plant-water sensitivity; R2 = 0.71). This has led to rapid increases in human exposure to wildfire risk, both because the population living in areas with high plant-water sensitivity grew 50% faster during 1990-2010 than in other wildland-urban interfaces and because VPD has risen most rapidly in these vulnerable areas. As plant-water sensitivity is strongly linked to wildfire vulnerability, accounting for ecophysiological controls should improve wildfire forecasts. If recent trends in VPD and demographic shifts continue, human wildfire risk will probably continue to increase.

Contribution of historical precipitation change to US flood damages.

Precipitation extremes have increased across many regions of the United States, with further increases anticipated in response to additional global warming. Quantifying the impact of these precipitation changes on flood damages is necessary to estimate the costs of climate change. However, there is little empirical evidence linking changes in precipitation to the historically observed increase in flood losses. We use >6,600 reports of state-level flood damage to quantify the historical relationship between precipitation and flood damages in the United States. Our results show a significant, positive effect of both monthly and 5-d state-level precipitation on state-level flood damages. In addition, we find that historical precipitation changes have contributed approximately one-third of cumulative flood damages over 1988 to 2017 (primary estimate 36%; 95% CI 20 to 46%), with the cumulative impact of precipitation change totaling $73 billion (95% CI 39 to $91 billion). Further, climate models show that anthropogenic climate forcing has increased the probability of exceeding precipitation thresholds at the extremely wet quantiles that are responsible for most flood damages. Climate models project continued intensification of wet conditions over the next three decades, although a trajectory consistent with UN Paris Agreement goals significantly curbs that intensification. Taken together, our results quantify the contribution of precipitation trends to recent increases in flood damages, advance estimates of the costs associated with historical greenhouse gas emissions, and provide further evidence that lower levels of future warming are very likely to reduce financial losses relative to the current global warming trajectory.

Verification of extreme event attribution: Using out-of-sample observations to assess changes in probabilities of unprecedented events.

Independent verification of anthropogenic influence on specific extreme climate events remains elusive. This study presents a framework for such verification. This framework reveals that previously published results based on a 1961-2005 attribution period frequently underestimate the influence of global warming on the probability of unprecedented extremes during the 2006-2017 period. This underestimation is particularly pronounced for hot and wet events, with greater uncertainty for dry events. The underestimation is reflected in discrepancies between probabilities predicted during the attribution period and frequencies observed during the out-of-sample verification period. These discrepancies are most explained by increases in climate forcing between the attribution and verification periods, suggesting that 21st-century global warming has substantially increased the probability of unprecedented hot and wet events. Hence, the use of temporally lagged periods for attribution-and, more broadly, for extreme event probability quantification-can cause underestimation of historical impacts, and current and future risks.