Adaptation is often framed as marginally important to addressing climate change, and as socio-technically difficult and ineffectual. We combine theoretical and empirical analyses to show that adaptation—especially via economic development—is actually often the dominant driver of climate-sensitive societal outcomes, especially on smaller space and time scales. This aligns adaptation with markets and governance incentives. For these reasons, widely studied climate-sensitive outcomes such as crop yields, affluence, and damage and death rates from climate-related hazards have broadly and steadily improved over the past several decades, as have indirectly climate-sensitive outcomes such as mortality from violence and self-harm. These improvements provide important context to recent pessimistic studies of adaptation that focus on outcomes’ marginal sensitivities to climate. They also underscore the importance of economic development to human well-being, and they suggest that economically costly climate policies could harm climate-sensitive outcomes. Moreover, we show that the range of plausible greenhouse gas emissions scenarios has narrowed, providing greater clarity to the temperatures and types of impacts society must adapt to. Our analyses highlight where adaptation and development are currently underappreciated in climate change research and policy.

Increasing fuel aridity due to climate warming has and will continue to increase wildfire danger in California. In addition to reducing global greenhouse gas emissions, one of the primary proposals for counteracting this increase in wildfire danger is a widespread expansion of hazardous fuel reductions. Here, we quantify the potential for fuel reduction to reduce wildfire intensity using empirical relationships derived from historical observations with a novel combination of spatiotemporal resolution (0.375 km, instantaneous) and extent (48 million acres, 9 years). We use machine learning to quantify relationships between sixteen environmental conditions (including ten fuel characteristics and four temperature-affected aridity characteristics) and satellite-observed fire radiative power. We use the derived relationships to create fire intensity potential (FIP) maps for sixty historical weather snapshots at a 2 km and hourly resolution. We then place these weather snapshots in differing background climatological temperature and fuel characteristic conditions to quantify their independent and combined influence on FIP. We find that in order to offset the effect of climate warming under the SSP2-4.5 emissions scenario, fuel reduction would need to be maintained perpetually on similar to 3 million acres (or 600 000 acres per year, 1% of our domain, at a 5 year return frequency) by 2050 and similar to 8 million acres (or 1.6 million acres per year, 3% of our domain, at a 5 year return frequency) by 2090. Overall, we find substantial potential for fuel reduction to negate the effects of climate warming on FIP.

Large-scale capacity expansion models typically rely on estimates of the power transfer limits between modeled zones. Accurate estimation of these interface transfer limits (ITLs) requires modeling the underlying transmission network. Here we expand on a maximum flow optimization method that uses linearized power flow to estimate transfer limits. We apply this method to a data set of the U.S. transmission network to estimate ITLs between U.S. counties. By calculating ITLs using different subsets of the network, we evaluate how the size of the network used in the estimation affects the results. The results show diminishing returns to ITL accuracy after six hops, suggesting that a network subset can reasonably be used to approximate ITLs. The county-level estimates produced in this study will support more spatially resolved capacity expansion modeling and will help inform policy making at local and national levels.

California has experienced enhanced extreme wildfire behaviour in recent years 1 – 3 , leading to substantial loss of life and property 4 , 5 . Some portion of the change in wildfire behaviour is attributable to anthropogenic climate warming, but formally quantifying this contribution is difficult because of numerous confounding factors 6 , 7 and because wildfires are below the grid scale of global climate models. Here we use machine learning to quantify empirical relationships between temperature (as well as the influence of temperature on aridity) and the risk of extreme daily wildfire growth (>10,000 acres) in California and find that the influence of temperature on the risk is primarily mediated through its influence on fuel moisture. We use the uncovered relationships to estimate the changes in extreme daily wildfire growth risk under anthropogenic warming by subjecting historical fires from 2003 to 2020 to differing background climatological temperatures and aridity conditions. We find that the influence of anthropogenic warming on the risk of extreme daily wildfire growth varies appreciably on a fire-by-fire and day-by-day basis, depending on whether or not climate warming pushes conditions over certain thresholds of aridity, such as 1.5 kPa of vapour-pressure deficit and 10% dead fuel moisture. So far, anthropogenic warming has enhanced the aggregate expected frequency of extreme daily wildfire growth by 25% (5–95 range of 14–36%), on average, relative to preindustrial conditions. But for some fires, there was approximately no change, and for other fires, the enhancement has been as much as 461%. When historical fires are subjected to a range of projected end-of-century conditions, the aggregate expected frequency of extreme daily wildfire growth events increases by 59% (5–95 range of 47–71%) under a low SSP1–2.6 emissions scenario compared with an increase of 172% (5–95 range of 156–188%) under a very high SSP5–8.5 emissions scenario, relative to preindustrial conditions.

Weather and climate phenomena have outsized impacts on society when they are particularly extreme. Extreme Event Attribution (EEA) seeks to quantify the extent to which extreme weather and climate phenomena are the result of anthropogenic climate change (ACC), and thus it has implications for many pertinent climate change discussions, including those on potential legal claims of loss and damages and calculations of the social cost of carbon. The Fraction of Attributable Risk (FAR) is one metric that is used to quantify the proportion of an extreme weather or climate “event” associated with ACC. The FAR is typically applied to changes in the likelihood of exceeding some geophysical value chosen, post hoc, to represent the “event” (e.g., i.e., rainfall amounts, flood depths, drought measures, temperature values, etc.). The FAR has further been used to estimate the fraction of observed impacts (e.g., lives lost or economic damage) that can be associated with ACC by multiplying realized impacts by the FAR (IFAR = Impact×FAR). Here, we illustrate with a few stylized examples that this IFAR calculation only produces reliably useful results when the weather or climate phenomena in question can be easily conceived of as a discrete binary “event” (i.e., the entirety of the event either occurs or it does not). We show that the IFAR calculation can produce misleading results when the weather or climate phenomena in question are on a continuum, and ACC can be thought of as altering the intensity of the geophysical value that is used in the eventhood definition. Specifically, we show that the IFAR calculation inflates the impacts associated with ACC in these circumstances because it inaccurately assumes that there would have been zero impact had the geophysical value chosen to define eventhood not been exceeded. We illustrate that for weather and climate phenomena on a continuum (e.g., floods, droughts, temperatures, etc.), a clearer way of conceptualizing the impacts associated with ACC is to compare the expected value of the impact between the ACC and preindustrial conditions across the full continuum.

Abstract Estimating the human influence on extreme weather and climate events and their downstream impacts is relevant to many pertinent discussions around climate change, including potential legal claims of loss and damages and calculations of the social cost of carbon. Recently, a prominent method for doing this has emerged, referred to as the Attributable Costs method. In this method, the fraction of the risk of crossing a geophysical threshold attributable to Anthropogenic Climate Change (ACC) is multiplied by an impact (e.g., monetary economic damage) to calculate the ACC contribution to the impact. Here, I illustrate with a stylized example that the method produces misleading results when applied to weather phenomena on a continuum, where ACC alters the magnitude of the phenomena and/or the entire probability distribution of the phenomena. Specifically, the Attributable Costs method fails in these circumstances because impacts cannot be assumed to be zero when the geophysical threshold chosen to define the event is not crossed. This implies that the Attributable Costs method will systematically overestimate the contribution of ACC to impacts. We illustrate that a clearer way of conceptualizing the impacts attributable to ACC is to compare the expected losses of impacts using full probability distributions of the geophysical variable in question.

The growing climate emergency requires a dramatic and rapid reduction of greenhouse gas emissions in the United States and internationally. This study evaluates a variety of 100% clean electricity system scenarios in 2035 that could put the United States on a path to economy-wide net-zero emissions by 2050, specifically focusing on technical requirements, challenges, and cost implications. The results highlight there are multiple approaches to cost-effectively achieve a net-zero carbon grid in 2035.

Meeting the last increment of demand always poses challenges, irrespective of whether the resources used to meet it are carbon free. The challenges primarily stem from the infrequent utilization of assets deployed to meet high demand periods, which require very high revenue during those periods to recover capital costs. Achieving 100% carbon-free electricity obviates the use of traditional fossil-fuel-based generation technologies, by themselves, to serve the last increment of demand—which we refer to as the “last 10%.” Here, we survey strategies for overcoming this last 10% challenge, including extending traditional carbon-free energy sources (e.g., wind and solar, other renewable energy, and nuclear), replacing fossil fuels with carbon-free fuels for combustion (e.g., hydrogen- and biomass-based fuels), developing carbon capture and carbon dioxide removal technologies, and deploying multi-day demand-side resources. We qualitatively compare economic factors associated with the low-utilization condition and discuss unique challenges of each option to inform the complex assessments needed to identify a portfolio that could achieve carbon-free electricity. Although many electricity systems are a long way from requiring these last 10% technologies, research and careful consideration are needed soon for the options to be available when electricity systems approach 90% carbon-free electricity.

California has experienced increased instances of extreme wildfire behavior in recent years, but the extent to which this is due to anthropogenic warming has been difficult to determine. Here we quantify empirical relationships between temperature and the risk of extreme daily wildfire growth (>10,000 acres) in California and use these relationships to estimate how extreme growth risk is changing under anthropogenic warming. We subject fires from 2003 to 2020 to differing background climatological temperatures and aridity metrics and find that the fraction of the risk of extreme daily growth attributable to anthropogenic warming to date averages 19% but varies substantially depending on whether background warming pushed fires over critical aridity thresholds. When the historical fires from 2003 to 2020 are subjected to projected end-of-century temperatures, the expected frequency of extreme daily growth events increases by 59% under an emissions scenario in line with the Paris Agreement, compared to an increase of 172% under a very high emissions scenario.

The share of variable renewable energy (VRE) on India's grid has surpassed 100 GW, and the government has ambitious plans reach 450 GW by 2030. One strategy to increase wind and solar PV deployment is through the co-location of wind and solar PV plants to form a single hybrid power plant. Hybrid plants have the potential to reduce transmission infrastructure costs and variability in the output power profile compared to a standalone plant with a single technology, and this resource analysis aims to take a first step towards quantifying the potential savings from hybridizing wind and solar PV plants in India and the size of this opportunity. We utilize a brute-force optimization to minimize the levelized cost of energy (LCOE) for standalone wind, standalone solar PV, and hybrid wind/solar PV plants across all of India. By comparing these LCOEs, we determine that locations where hybrid plants exhibit potential cost savings and grid benefits exhibit both; a high interconnection cost and; a wind capacity factor between roughly 34% and 38%. However, because our work does not capture the value of the electricity generated by looking at energy prices, nor does it quantify the potential of hybrids to provide other value streams such as firm capacity and reserves. Further, because the work does not compare solar PV and wind hybrids to alternative generation technologies or storage systems, it cannot be considered a holistic cost-benefit analysis.

Wind and solar electricity generation is projected to expand substantially over the next several decades due both to rapid cost declines as well as regulation designed to achieve climate targets. With increasing reliance on wind and solar generation, future energy systems may be vulnerable to previously underappreciated synoptic-scale variations characterized by low wind and/or surface solar radiation. Here we use western North America as a case study region to investigate the historical meteorology of weekly-scale “droughts” in potential wind power, potential solar power and their compound occurrence. We also investigate the covariability between wind and solar droughts with potential stresses on energy demand due to temperature deviations away human comfort levels. We find that wind power drought weeks tend to occur in late summer and are characterized by a mid-level atmospheric ridge centered over British Columbia and high sea level pressure on the lee side of the Rockies. Solar power drought weeks tend to occur near winter solstice when the seasonal minimum in incoming solar radiation co-occurs with the tendency for mid-level troughs and low pressure systems over the U.S. southwest. Compound wind and solar power drought weeks consist of the aforementioned synoptic pattern associated with wind droughts occurring near winter solstice when the solar resource is at its seasonal minimum. We find that wind drought weeks are associated with high solar power (and vice versa) both seasonally and in terms of synoptic meteorology, which supports the notion that wind and solar power generation can play complementary roles in a diversified energy portfolio at synoptic spatiotemporal scales over western North America.

Global climate change mitigation is often framed in public discussions as a tradeoff between environmental protection and harm to the economy. However, climate-economy models have consistently calculated that the immediate implementation of greenhouse gas emissions restriction (via e.g. a global carbon price) would be in humanity’s best interest on purely economic grounds. Despite this, the implementation of global climate policy has been notoriously difficult to achieve. This evokes an apparent paradox: if the implementation of a global carbon price is not only beneficial to the environment, but is also ‘economically optimal’, why has it been so difficult to enact? One potential reason for this difficulty is that economically optimal greenhouse gas emissions restrictions are not economically beneficial for the generation of people that launch them. The purpose of this article is to explore this issue by introducing the concept of the break-even year, which we define as the year when the economically optimal policy begins to produce global mean net economic benefits. We show that in a commonly used climate-economy model (DICE), the break-even year is relatively far into the future—around 2080 for mitigation policy beginning in the early 2020s. Notably, the break-even year is not sensitive to the uncertain magnitudes of the costs of climate change mitigation policy or the costs of economic damages from climate change. This result makes it explicit and understandable why an economically optimal policy can be difficult to implement in practice.

Efforts to mitigate global warming are often justified through calculations of the economic damages that may occur absent mitigation. The earliest such damage estimates were speculative mathematical representations, but some more recent studies provide empirical estimates of damages on economic growth that accumulate over time and result in larger damages than those estimated previously. These heightened damage estimates have been used to suggest that limiting global warming this century to 1.5 °C avoids tens of trillions of 2010 US$ in damage to gross world product relative to limiting global warming to 2.0 °C. However, in order to estimate the net effect on gross world product, mitigation costs associated with decarbonizing the world's energy systems must be subtracted from the benefits of avoided damages. Here, we follow previous work to parameterize the aforementioned heightened damage estimates into a schematic global climate-economy model (DICE) so that they can be weighed against mainstream estimates of mitigation costs in a unified framework. We investigate the net effect of mitigation on gross world product through finite time horizons under a spectrum of exogenously defined levels of mitigation stringency. We find that even under heightened damage estimates, the additional mitigation costs of limiting global warming to 1.5 °C (relative to 2.0 °C) are higher than the additional avoided damages this century under most parameter combinations considered. Specifically, using our central parameter values, limiting global warming to 1.5 °C results in a net loss of gross world product of roughly forty trillion US$ relative to 2 °C and achieving either 1.5 °C or 2.0 °C require a net sacrifice of gross world product, relative to a no-mitigation case, though 2100 with a 3%/year discount rate. However, the benefits of more stringent mitigation accumulate over time and our calculations indicate that stabilizing warming at 1.5 °C or 2.0 °C by 2100 would eventually confer net benefits of thousands of trillions of US$ in gross world product by 2300. The results emphasize the temporal asymmetry between the costs of mitigation and benefits of avoided damages from climate change and thus the long timeframe for which climate change mitigation investment pays off.

Abstract Global mean surface air temperature (Tglobal) variability on subdecadal timescales can be of substantial magnitude relative to the long‐term global warming signal, and such variability has been associated with considerable environmental and societal impacts. Therefore, probabilistic foreknowledge of short‐term Tglobal evolution may be of value for anticipating and mitigating some course‐resolution climate‐related risks. Here we present a simple, empirically based methodology that utilizes only global spatial patterns of annual mean surface air temperature anomalies to predict subsequent annual Tglobal anomalies via partial least squares regression. The method's skill is primarily achieved via information on the state of long‐term global warming as well as the state and recent evolution of the El Niño–Southern Oscillation and the Interdecadal Pacific Oscillation. We test the out‐of‐sample skill of the methodology using cross validation and in a forecast mode where statistical predictions are made precisely as they would have been if the procedure had been operationalized starting in the year 2000. The average forecast errors for lead times of 1 to 4 years are smaller than naïve benchmarks on average, and they perform favorably relative to most dynamical Global Climate Models retrospectively initialized to the observed state of the climate system. Thus, this method can be used as a computationally efficient benchmark for dynamical model forecast systems.

The El Niño Southern Oscillation (ENSO) is the dominant mode of variability in the climate system on seasonal to decadal timescales. With foreknowledge of the state of ENSO, stakeholders can anticipate and mitigate impacts in climate-sensitive sectors such as agriculture and energy. Traditionally, ENSO forecasts have been produced using either computationally intensive physics-based dynamical models or statistical models that make limiting assumptions, such as linearity between predictors and predictands. Here we present a deep-learning-based methodology for forecasting monthly ENSO temperatures at various lead times. While traditional statistical methods both train and validate on observational data, our method trains exclusively on physical simulations. With the entire observational record as an out-of-sample validation set, the method’s skill is comparable to that of operational dynamical models. The method is also used to identify disagreements among climate models about the predictability of ENSO in a world with climate change.

Today, most global economic production depends on energy produced from burning fossil fuels, which emit carbon dioxide as a byproduct. Although the costs of carbon-free energy such as wind and solar have come down dramatically over recent decades, there are substantial challenges to completely decarbonizing our electricity system, and even greater challenges to completely decarbonizing the transportation and industrial sectors (1). Thus, economic activity is projected to produce greenhouse gas emissions throughout this century. These emissions of greenhouse gases are causing Earth to warm, and, in aggregate, the effects of global warming are expected to be deleterious. These deleterious effects are expected to harm global welfare and diminish economic productivity. This diminution of production, other things being equal, would lead to a reduction in greenhouse gas emissions and, thus, would lessen the anticipated warming. In PNAS, Woodard et al. (2) find that the reduction in emissions through damage to economic activity is roughly the same magnitude as, but opposite in sign to, natural carbon cycle feedbacks that are projected to increase carbon dioxide levels relative to a world without carbon–climate feedbacks. The net effect of the socioeconomic carbon–climate feedbacks is estimated to be about the same magnitude as, but opposite in sign to, natural biogeophysical carbon–climate feedbacks. As a result, the level of greenhouse gases in the atmosphere in 2100 is projected to be about the same as if neither feedback (socioeconomic or natural biogeophysical) were active. Woodard et al. (2) analyze various influences on carbon dioxide emissions using the Kaya identity, which represents these emissions as the product of population, per-capita productivity, energy used per unit of production [energy intensity of gross domestic production (GDP)], and carbon emitted per unit of energy used (carbon intensity of energy). Reduced per-capita economic productivity due to climate change is projected to be … [↵][1]1To whom correspondence should be addressed. Email: kcaldeira{at}carnegiescience.edu. [1]: #xref-corresp-1-1