8.1 Types of climate models and their applications

Climate Model Types and Uses

Climate models are the primary tools scientists use to understand and predict how Earth's climate system behaves. They translate physical laws, chemical processes, and biological interactions into mathematical equations that computers can solve. These models range from simple energy calculations to massive simulations that couple the atmosphere, oceans, land surfaces, ice sheets, and carbon cycle together.

Their applications are equally broad: informing IPCC assessment reports, guiding national climate policy, projecting extreme weather trends, and helping sectors like agriculture and public health plan for the future.

Main Categories of Climate Models

Climate models fall into three broad tiers, each trading off complexity against computational cost.

• Simple climate models reduce the climate system to a handful of equations. They're fast to run and useful for exploring how global temperature responds to different forcings (like doubling $CO_2$). Policymakers often rely on them for quick scenario evaluations, such as estimating remaining carbon budgets.
• Earth System Models of Intermediate Complexity (EMICs) sit in the middle. They include more processes than simple models but use simplified representations of components like ocean circulation. That efficiency makes them well suited for paleoclimate studies and multi-millennium simulations where running a full-complexity model would be impractical.
• General Circulation Models (GCMs) and Earth System Models (ESMs) are the most comprehensive. GCMs simulate atmospheric and oceanic circulation using three-dimensional grids, while ESMs go further by adding interactive carbon cycles, vegetation dynamics, and atmospheric chemistry. These are the workhorses behind detailed future climate projections, such as those in CMIP6 (Coupled Model Intercomparison Project, Phase 6).

Specialized Climate Models

• Regional Climate Models (RCMs) take output from a GCM and "nest" a higher-resolution simulation over a smaller area. This lets them capture local features like mountain ranges, coastlines, and land-use patterns that global models are too coarse to resolve. For example, an RCM might simulate Mediterranean Basin precipitation at 10–25 km resolution compared to a GCM's typical 50–100 km grid spacing.
• Energy Balance Models (EBMs) are among the simplest climate models. They represent Earth's temperature by balancing incoming solar radiation against outgoing longwave radiation. Despite their simplicity, EBMs are valuable for building intuition about fundamental processes like the greenhouse effect and how changes in albedo (surface reflectivity) shift global temperature.

Statistical vs. Dynamical Models

Fundamental Differences

Statistical climate models use historical observations to identify relationships between variables, then project those relationships forward. For instance, a statistical model might link sea surface temperatures in the tropical Pacific to rainfall patterns over East Africa. The core assumption is that past relationships remain valid in the future.

Dynamical models solve the fundamental equations of physics (conservation of mass, energy, and momentum) on a gridded representation of the Earth. Because they're built from physical principles rather than historical correlations, they can simulate climate states that have no precedent in the observational record, such as the response to $CO_2$ concentrations not seen in millions of years.

Computational Requirements and Applications

• Statistical models are computationally cheap and fast. They're commonly used for seasonal forecasting and for statistical downscaling, where coarse GCM output is translated into local-scale projections using observed relationships. The tradeoff: they need extensive, high-quality historical data and can break down when the climate moves outside the range of past experience.
• Dynamical models demand significant computing power. A single century-long ESM simulation can require millions of processor-hours on a supercomputer. They're preferred for long-term projections (e.g., end-of-century scenarios under different emissions pathways) precisely because they don't depend on the assumption that the future will resemble the past.

Emerging Hybrid Approaches

Hybrid models combine both approaches to leverage their respective strengths. A common strategy is dynamical-statistical downscaling, where a dynamical model provides large-scale climate fields and statistical methods refine those fields to local scales. These hybrids can improve short-term forecast skill while preserving the physical consistency needed for long-term projections.

Applications of Climate Models

Climate Change Assessment and Policy

Climate models are central to the IPCC assessment process. Scientists run coordinated experiments using multiple models under shared emissions scenarios (called Shared Socioeconomic Pathways, or SSPs) to project how temperature, precipitation, and sea level may change. These projections directly inform international agreements like the Paris Agreement, where targets such as limiting warming to 1.5°C or 2°C above pre-industrial levels depend on model-derived carbon budgets.

Models also help evaluate mitigation strategies. For example, they can estimate how quickly global temperatures would respond to a transition from fossil fuels to renewable energy, or how much carbon removal would be needed to meet specific targets.

Extreme Weather and Climate Feedbacks

Climate models are used to study how extreme events change in a warming world. Researchers can compare simulations with and without human-caused greenhouse gas increases to assess how the frequency and intensity of heat waves, tropical cyclones, and droughts shift under different warming levels.

Models also help identify climate feedbacks and tipping points. Key examples include:

• Arctic sea ice loss: reduced ice lowers surface albedo, which amplifies warming (ice-albedo feedback)
• Amazon rainforest dieback: warming and drying could push the forest past a threshold where it transitions to savanna, releasing stored carbon
• Atlantic Meridional Overturning Circulation (AMOC) slowdown: freshwater input from melting ice sheets could weaken this ocean current, with major consequences for European and global climate

Interdisciplinary Applications

Climate model output feeds into impact models across many sectors:

• Agriculture: crop models use projected temperature and precipitation to estimate future yields for staple crops like wheat and maize
• Water resources: hydrological models translate climate projections into river flow and groundwater recharge estimates
• Ecosystems: species distribution models project how habitat ranges may shift as climate zones move
• Human health: models track how warming may expand the range of disease vectors like mosquitoes carrying malaria or dengue

Beyond impact assessment, climate models contribute to early warning systems for floods and droughts, and they inform long-term infrastructure planning, from coastal defense design to urban heat island mitigation.

Limitations and Strengths of Climate Models

Model-Specific Characteristics

Each model type has characteristic strengths and weaknesses tied to its level of complexity:

• Simple models give fast answers but can't represent spatial patterns or regional detail. They're best for global-mean estimates and quick sensitivity tests.
• EMICs handle long timescales well but may oversimplify key processes. For instance, an EMIC might use a simplified ocean model that doesn't fully capture deep-water formation patterns.
• GCMs/ESMs provide the most complete picture of the climate system, but their complexity makes them expensive to run and sometimes difficult to interpret. Small differences in how they represent sub-grid processes can lead to divergent projections.

Comparative Strengths and Weaknesses

• Statistical models are efficient for well-observed regions and established climate patterns, but they can fail under truly novel conditions, such as rapid ice sheet collapse or abrupt shifts in ocean circulation, where historical relationships no longer apply.
• Regional Climate Models add valuable local detail, but they inherit biases from the global model that provides their boundary conditions. If the driving GCM misrepresents large-scale circulation, the RCM will carry that error into its regional simulation.
• All climate models struggle with certain processes that occur at scales smaller than their grid cells. Cloud formation, aerosol-cloud interactions, and turbulent mixing in the ocean are parameterized (approximated with simplified equations) rather than explicitly resolved. These parameterizations are a major source of uncertainty in projections, particularly for regional precipitation and climate sensitivity (how much warming results from doubling $CO_2$).

Overall Value and Ongoing Improvements

The core strength of climate models is their ability to integrate physical, chemical, and biological processes into a coherent framework, providing insights that no single observation or experiment could deliver on its own.

Models are steadily improving through several avenues:

• Higher resolution: as computing power grows, models can use finer grids, explicitly resolving processes like convective storms that previously had to be parameterized
• Better parameterizations: ongoing research refines how sub-grid processes like cloud microphysics and land-surface interactions are represented
• Multi-model ensembles: projects like CMIP6 coordinate dozens of modeling groups worldwide, running the same experiments so that the spread across models can be used to quantify projection uncertainty

No single model is "correct." The value comes from understanding the range of outcomes across models and identifying where they agree (high confidence) versus where they diverge (areas needing further research).