Detection and Attribution of Climate Change

The statistical methods that establish whether climate has changed beyond natural variability and identify the human and natural causes responsible.

Definition

Detection is the demonstration that climate has changed in a statistical sense beyond what internal variability would produce, and attribution is the process of establishing, with confidence, the most likely causes of that detected change.

Scope

This topic covers the formal framework for determining whether an observed climate change is real and what caused it. It treats detection, showing that a change is unlikely to be due to internal variability alone, and attribution, assigning the change to particular forcings using their distinct fingerprints. It includes the use of climate models to estimate internal variability and forced responses, the spatial and vertical fingerprints that distinguish drivers, and the emerging field of attributing individual extreme events.

Core questions

• How is a real climate change distinguished from natural variability?
• How are observed changes attributed to specific forcings?
• What fingerprints distinguish human from natural causes?
• Can individual extreme events be attributed to climate change?

Key theories

Fingerprinting

Each forcing produces a characteristic spatial and temporal pattern of response, so matching the observed pattern against modeled fingerprints attributes the change to its likely causes.

Probabilistic event attribution

Comparing the likelihood of an extreme event in model worlds with and without human influence quantifies how much human-caused change altered its probability or intensity.

Mechanisms

Detection compares the observed change with the range of internal variability estimated from long model runs and paleoclimate, judging it detected if it lies outside that range. Attribution then regresses the observations onto the modeled response patterns for each forcing, using the distinct fingerprints, such as tropospheric warming with stratospheric cooling, that separate greenhouse from solar forcing, while event attribution compares simulated event probabilities in factual and counterfactual climates.

Clinical relevance

Detection and attribution provide the rigorous evidence that recent warming is human-caused, underpin the IPCC's confidence statements, and increasingly inform legal and policy decisions and the assessment of climate-related risk and loss.

Evidence & guidelines

Following good-practice guidance, the IPCC Sixth Assessment Report attributes essentially all of the observed warming since the mid-twentieth century to human influence and reports growing confidence in attributing many extreme events.

History

Hasselmann's optimal fingerprinting framework in the late twentieth century put attribution on a firm statistical footing, Santer and others demonstrated the human fingerprint in the 1990s, and the 2000s saw the rise of probabilistic attribution of individual extreme events.

Debates

Reliability of model-estimated internal variability

Because attribution depends on how well models represent natural variability, debate continues over whether models capture enough low-frequency variability to support confident detection on regional scales.

Key figures

• Gabriele Hegerl
• Klaus Hasselmann
• Benjamin Santer
• Peter Stott

Seminal works

• hegerl2010
• ipccar6wg1

Frequently asked questions

What is climate fingerprinting?

It is the technique of matching the observed pattern of climate change against the distinct response patterns expected from each driver to identify which causes are responsible.

Can a single heatwave be blamed on climate change?

Event attribution cannot blame an event entirely on climate change but can quantify how much human warming changed its likelihood or intensity by comparing model worlds with and without it.