9.2 Observed temperature trends and attribution

Global and Regional Temperature Trends

Global mean surface temperature has risen approximately 1.1°C since pre-industrial times (roughly the 1850–1900 baseline), and the rate of warming has accelerated sharply since the mid-20th century. Understanding where and how this warming shows up is central to attributing it to specific causes.

Global Temperature Increase

• Land warms faster than oceans. Water has a much higher heat capacity than soil and rock, so oceans absorb more energy before their temperature rises. Land surfaces also lose moisture faster, reducing evaporative cooling.
• Higher latitudes warm faster than the tropics. This pattern, called Arctic amplification, is driven primarily by ice-albedo feedback: as ice and snow melt, darker land and ocean surfaces absorb more solar radiation, which drives further warming. Changes in atmospheric and oceanic heat transport also contribute.
• Urban areas show enhanced warming beyond the regional trend. The urban heat island effect results from replacing vegetation with heat-absorbing surfaces (asphalt, concrete), reduced evapotranspiration, and waste heat from buildings and vehicles. Climate scientists account for this when calculating global averages so it doesn't bias the record.

Temperature Pattern Changes

Warming isn't uniform across time of day, season, or region.

• Nighttime temperatures are rising faster than daytime temperatures, shrinking the diurnal temperature range. This matters for agriculture (many crops need cool nights) and human health (warm nights reduce the body's ability to recover from daytime heat).
• Temperature extremes are becoming more frequent and intense. Heat waves occur more often and last longer. The 2003 European heatwave, which killed tens of thousands, is a well-studied example. Warm nights are also increasing, raising energy demand for cooling and stressing vulnerable populations.
• Some regions buck the trend. Parts of the Southern Ocean and subpolar North Atlantic have shown localized cooling, likely linked to changes in ocean circulation, including a possible weakening of the Atlantic Meridional Overturning Circulation (AMOC) and increased upwelling of deep cold water around Antarctica.

Attributing Temperature Changes

Attribution is the process of determining why the climate is changing, not just that it's changing. It relies on comparing what climate models predict under different scenarios with what we actually observe.

Climate Modeling and Analysis

1. Run climate models with different forcings. Scientists simulate global temperatures using only natural factors (solar changes, volcanoes), only anthropogenic factors (greenhouse gases, aerosols), or both combined.
2. Compare model output to observations. If a simulation matches the observed temperature record well, the forcings used in that run are likely contributing to real-world change.
3. Identify climate "fingerprints." Different forcings produce distinct spatial and temporal patterns. For example, greenhouse gas warming produces tropospheric warming plus stratospheric cooling, while solar forcing would warm both layers. These fingerprints help separate causes.
4. Quantify contributions statistically. Techniques like optimal fingerprinting scale modeled patterns to best fit observations, estimating how much each forcing contributed. Bayesian methods assign probabilities to different attribution statements.

The key finding from these studies: models that include only natural forcings cannot reproduce the observed warming trend since the mid-20th century. You need anthropogenic forcings to match the data.

Climate Forcing Factors

Radiative forcing quantifies how much a given factor shifts Earth's energy balance, measured in watts per square meter ($\text{W/m}^2$). Positive forcing warms the planet; negative forcing cools it.

Natural factors considered in attribution:

• Solar variability — the 11-year sunspot cycle and longer-term output changes (contributes roughly $+0.1 \text{ W/m}^2$ since 1750, far too small to explain observed warming)
• Volcanic eruptions — inject sulfate aerosols into the stratosphere, causing short-term cooling (e.g., Mt. Pinatubo in 1991 cooled global temperatures by about 0.5°C for 1–2 years)
• Internal variability — oscillations like ENSO and the Pacific Decadal Oscillation redistribute heat but don't add net energy to the climate system over long periods

Anthropogenic factors considered:

• Greenhouse gas emissions — $\text{CO}_2$, methane ($\text{CH}_4$), and nitrous oxide ($\text{N}_2\text{O}$) are the dominant warming influence, with a combined forcing of roughly $+3.0 \text{ W/m}^2$ since 1750
• Aerosol emissions — sulfate aerosols reflect sunlight (cooling effect), while black carbon absorbs it (warming effect); the net aerosol effect is cooling, partially masking greenhouse gas warming
• Land-use changes — deforestation and urbanization alter surface albedo and moisture fluxes
• Ozone changes — stratospheric ozone depletion (cooling effect there) and tropospheric ozone increase (warming effect near the surface)

Evidence for Anthropogenic Influence

Observed Warming Patterns

Several lines of evidence converge to implicate human activity:

• Spatial patterns match predictions. Greater warming over land than oceans, Arctic amplification, and faster warming at night all align with what greenhouse gas physics predicts.
• Natural factors can't explain the trend. Solar output has been roughly flat or slightly declining since the 1980s, yet temperatures have continued rising. Volcanic eruptions cause temporary dips, not sustained warming. Internal variability like ENSO operates on timescales of years to decades and averages out.
• The vertical temperature structure is a smoking gun. The troposphere (lower atmosphere) is warming while the stratosphere is cooling. Increased greenhouse gases trap more outgoing infrared radiation in the troposphere, leaving less energy to warm the stratosphere. If the Sun were the primary driver, both layers would warm.

Energy Balance and Historical Context

• Satellite measurements show a growing energy imbalance at the top of the atmosphere: Earth is absorbing more energy than it radiates back to space, consistent with the enhanced greenhouse effect.
• Paleoclimate records from tree rings, ice cores, and ocean sediments show that the current rate of warming is unprecedented over at least the past 2,000 years. Natural variability in those records is far smaller than the warming observed since the Industrial Revolution.
• Timing aligns with emissions. The rapid temperature increase tracks closely with the rise in atmospheric $\text{CO}_2$ from fossil fuel combustion, which accelerated in the mid-20th century alongside the sharpest warming.

Uncertainties in Temperature Attribution

Attribution science is robust in its broad conclusions, but real uncertainties remain in the details.

Climate Variability and Data Limitations

• Internal variability complicates short-term attribution. Decadal oscillations like the Atlantic Multidecadal Oscillation (AMO) and Pacific Decadal Oscillation (PDO) can temporarily speed up or slow down warming in specific regions. The so-called "warming hiatus" of the early 2000s was partly linked to Pacific variability, not a pause in anthropogenic forcing.
• Historical temperature records have gaps. Coverage before the mid-20th century is sparse, especially over oceans and polar regions. Adjustments for changing instruments, station relocations, and measurement practices are necessary but introduce their own uncertainties.

Modeling Challenges and Knowledge Gaps

• Feedbacks are hard to pin down. Water vapor feedback amplifies warming, but cloud feedbacks remain the largest source of uncertainty in climate sensitivity estimates. Whether warming produces more low clouds (cooling) or fewer (warming) varies across models.
• Aerosol-cloud interactions are poorly constrained. Aerosols can change cloud brightness and lifetime, but these effects are difficult to observe and model. This uncertainty affects how well we can estimate the net cooling from aerosols, which in turn affects how much warming greenhouse gases are "responsible for."
• Non-linear responses and tipping points (e.g., ice sheet collapse, permafrost thaw releasing $\text{CH}_4$) are difficult to capture in models, adding uncertainty to future projections.
• Ongoing debates exist about specific contributions: for instance, how much of Arctic warming comes from black carbon deposited on snow versus greenhouse gas forcing, or how much regional temperature change is driven by land-use conversion versus large-scale atmospheric warming.

Improving model resolution and expanding observational networks (especially in data-sparse regions like the Arctic and deep oceans) are the main paths toward reducing these uncertainties.