How Attribution Science Uses Models to Uncover Climate Change’s Effects on Weather

   

 The Equation Read More [[{“value”:”

Scientists are increasingly able to determine how human-caused emissions have increased the risks of particular events and the harms they produce—advances that are critical as communities experience ever more costly and dangerous extreme weather. This work is known as attribution science: a robust and growing field that examines the role fossil-fueled climate change plays in specific weather events, long-term climate trends, and their real-world consequences.

As my colleague Dr. Delta Merner explains, attribution science includes several connected areas of research. The first, “event attribution,” asks whether climate change made a particular weather event more likely or more intense. The second, “source attribution,” traces climate change and associated hazards back to emissions sources or sectors. And thirdly, “impact attribution” examines the resulting harms, including damage to homes and infrastructure, lost livelihoods, health impacts, and deaths.

Attribution science utilizes a suite of tools to understand these connections: earth observations, statistical methods, emissions data, and climate models. Climate models are a central tool of attribution science through their ability to develop modeled counterfactuals—that is, simulations of what our world might have been without human-induced warming.

Climate models and attribution science have both advanced rapidly over recent decades, making both critical tools for understanding how climate change is progressing and fossil fuel companies’ responsibility for its worsening effects.

So, how exactly are climate models used in attribution science, and why can we have confidence in the conclusions they help produce?

Climate models are among the main tools scientists use to simulate how Earth will respond to increasing concentrations of heat-trapping pollutants in the atmosphere. When scientists refer to climate models, they generally talk about a general circulation model, or GCM. These models simulate how the atmosphere, oceans, land, and cryosphere respond to changes in internal climate dynamics, such as the El Niño-Southern Oscillation (ENSO) and Atlantic Multidecadal Oscillation, and external drivers, including greenhouse gas emissions.

Climate models are some of the most reliable tools for understanding our earth-climate system because they have been improved, tested, and corrected for decades. 

Modern climate models are made up of connected component models for the atmosphere, ocean, land surface, and cryosphere. An atmospheric model, for example, simulates circulation, radiation, clouds, and aerosols. Ocean models simulate circulation and heat uptake. Land models represent processes such as soil moisture, vegetation, snow cover, evapotranspiration, and carbon storage.

When these components are connected, or coupled, they exchange information with one another, allowing models to represent the feedbacks and interactions that shape the real-world climate system. These coupled models are often compared through coordinated international efforts such as the Coupled Model Intercomparison Project, or CMIP, which allows scientists to evaluate results across many models, institutions, and experimental designs.

CMIP simulations are also a major source of evidence for national and international climate assessments, including reports from the Intergovernmental Panel on Climate Change (IPCC), where they help scientists assess past climate change, project future risks, and evaluate confidence across different concentration scenarios and model results.

At their core, climate models use well-established physical laws to calculate how energy, momentum, water, and other properties move through the climate system. They divide the planet into a three-dimensional grid and calculate how conditions such as temperature, moisture, winds, and rainfall change over time and from place to place.

For attribution science, knowing how each of these factors contributes to our climate matters enormously. Event attribution compares the climate in which an event occurred (usually today’s climate) with a counterfactual climate that is identical except for the added influence of human-caused warming. By running climate model simulations with—and without—observed human influences, researchers can quantify how greenhouse gas emissions changed the probability or intensity of an event.

For example, a heatwave may have occurred because of a particular weather pattern, but human-caused warming can raise the baseline temperature on which that weather pattern acts, making an extreme heat event more likely, hotter, longer lasting, or all three. The 2021 Pacific Northwest heatwave illustrates this clearly: the region has already warmed by 1°-1.5°C, and event-specific attribution studies found that the heatwave would have been virtually impossible without human-caused climate change and was about 1°C hotter because of it. The relevant question is therefore how much climate change shifted the odds, intensity, or consequences of any given event.

Climate models are some of the most reliable tools for understanding our earth-climate system because they have been improved, tested, and corrected for decades. Their development spans generations of scientists and computational advances, and they continue to improve.

In fact, a recent study found that climate models that were developed over 50 years ago have correctly predicted the Earth’s warming. The fact that models developed decades ago broadly projected the warming trajectory we are now living through is strong evidence that they capture the core physics of global warming.

Furthermore, climate models have evolved in both scale and resolution, improving our ability to understand regional and localized changes to our earth-system. While global climate models are essential for understanding large-scale warming and circulation patterns, regional models can provide more detailed information about how those changes interact with local geography, land surfaces, coastlines, mountains, cities, and weather systems. This is especially useful when studying events such as convective storms, storm surge, flooding, and wildfires, where local and regional processes can strongly influence the severity of an event.

The advances by climate models and their proven record for simulating our earth system are important in the attribution of science. Climate models allow scientists to translate that physical understanding into quantitative estimates of how human-caused warming has changed the likelihood, intensity, or impacts of specific weather extremes.

Attribution scientists use multiple robust methods to test, constrain, and quantify uncertainty. One of the most important methods is the use of climate-model ensembles, which help scientists distinguish the role of natural variability from the influence of human-caused warming when assessing extreme weather events and their impacts.

First, scientists use initial-condition ensembles to understand internal climate variability. In these experiments, researchers run the same climate model many times with nearly identical conditions, changing only very small differences in the initial state of the atmosphere or ocean. Because the climate system is complex and chaotic, these tiny differences can lead to different weather patterns and different sequences of natural variability over time.

This allows attribution scientists to simulate the range of outcomes that could occur under the same external conditions. When researchers compare a world with human-caused emissions to a counterfactual world without them across many simulations for each scenario, they can see whether the climate-change signal consistently shifts the likelihood or severity of an event beyond the range expected from natural variability alone.

Second, scientists use multi-model ensembles to evaluate structural uncertainty, which originates in the scientific choices made by developers of each model. Different modeling centers make different scientific and technical choices about numerical schemes, parameterizations, spatial resolution, and how to represent processes such as clouds, aerosols, land-atmosphere interactions, and ocean circulation because the climate system is complex and no single model can perfectly capture every process at every scale. Comparing results across models helps scientists determine whether a finding is robust across multiple plausible representations of the climate system.

That said, some models can be similar in their representation of certain interactions in the climate system, mainly due to the fact that a few were developed using the same computer code and parameterizations. Regardless of the choice, researchers use methods that account for both model performance and model interdependence, giving less weight to structurally similar models or models that perform less well against observations.

The purpose of ensembles is to understand and constrain uncertainty in a transparent way. And what is clear is that, with both types of ensembles, there is greater confidence in the climate-change signals on a variety of weather extremes and impacts.

Climate models are essential tools for attribution science because they are grounded in well-established physics, tested against observations, and used alongside multiple lines of evidence. Attribution science brings together physical understanding, observations, statistical analysis, emissions data, and climate-model experiments to test how human-caused warming is changing the risks people face today. The use of multiple datasets, models and methods is what makes the field robust, allowing scientists to compare the world as it exists now to a counterfactual world without human-caused warming, quantify uncertainty, and assess how much climate change has shifted the likelihood, intensity, or impacts of specific events.

“}]]