Model Catalogue
ClimateCritters currently includes models from several families, organized here by the type of dynamics they exhibit. The Reference section has full API documentation for each; this page is meant as a starting point for choosing a model.
For method-evaluation purposes, the most relevant initial question is usually structural — does the model produce periodic output, chaotic output, intermittent transitions, or something else? Noise and sampling complications come later. The notes below try to flag those structural characteristics concisely.
Energy Balance Models
Equilibrium between absorbed shortwave and outgoing longwave radiation, with ice-albedo feedback.
EBM0D — Zero-dimensional global-mean surface temperature model. A single ODE tracking T from the balance C dT/dt = (1 - α) S₀/4 - OLR. With a temperature-dependent albedo function active, the model is bistable: a warm ice-free state and a fully glaciated (Snowball) state can both be stable for the same solar constant.
EBM1DLat — Budyko-Sellers type latitudinal EBM on a latitude grid. Diffusive meridional heat transport connects grid cells; OLR follows the Budyko linear form. The ice-line latitude emerges as a diagnostic. Useful for studying how meridional structure and polar amplification respond to changes in diffusivity or solar forcing.
Box Models
Composable multi-reservoir framework.
BoxModelSpec / GenericBoxModel — A framework rather than a specific model. Users define reservoirs and the fluxes connecting them; the integration machinery is handled automatically. Suitable for custom multi-box ocean, carbon, or tracer cycling experiments.
TwoBoxCarbon — Two-reservoir carbon cycle model tracking atmospheric CO₂ and surface ocean dissolved carbon. Exchange is proportional to the concentration gradient; a biological pump and weathering sink can be included.
Lorenz & Roessler
Low-dimensional chaotic systems.
Lorenz63 — Lorenz’s 1963 three-equation convection model. The canonical example of a strange attractor and sensitive dependence on initial conditions. Useful for testing anything that purports to detect chaos, nonlinearity, or memory in a time series.
Lorenz96 — Lorenz’s 1996 atmospheric ring model with N slow variables and, optionally, J fast variables per slow variable. Widely used as a testbed for data assimilation and ensemble forecasting. Complexity and number of active degrees of freedom scale with N and the forcing parameter F.
Roessler — Three-dimensional chaotic system with a simpler attractor geometry than Lorenz63. Often used as a comparison case when the structure of the attractor matters — the Rössler attractor has a cleaner folding topology than the butterfly.
Pendulums & Oscillators
Mechanical oscillators ranging from simple to chaotic.
SimplePendulum — Nonlinear pendulum with optional linear damping. Periodic at small amplitudes and anharmonic at larger ones. The natural frequency depends on amplitude, which can affect period-finding methods in non-obvious ways.
DrivenPendulum — Periodically forced damped pendulum. As the drive amplitude increases the system transitions from periodic to period-doubled to chaotic, making it a textbook example of the period-doubling route to chaos.
DoublePendulum — Conservative four-dimensional system. Chaotic for most initial conditions away from the trivial fixed point; sensitive dependence on initial conditions is especially pronounced.
DampedSpring — Damped spring-mass oscillator with optional periodic driving. The simplest oscillatory system in the library; exhibits resonance when driven near the natural frequency ω₀ = √(k/m). Useful as a baseline for testing methods that assume an underlying harmonic structure.
Individual Models
Standalone models drawn from the paleoclimate literature.
Daisyworld — Watson & Lovelock (1983). Two daisy species with different albedos colonize a planet and modulate its temperature. The system self-regulates temperature across a fairly wide range of solar luminosities, illustrating how biological feedbacks can stabilize climate. Multiple equilibria exist at the edges of the habitable range.
ENSORechargeOscillator — Jin (1997) recharge oscillator for ENSO. Two variables — sea surface temperature and thermocline heat content — produce quasi-periodic El Niño / La Niña cycles. Seasonal forcing can push the system into a more irregular or chaotic regime.
Model3 — Ganopolski (2024) glacial cycle model. Tracks ice volume and a discrete glacial/deglaciation regime under orbital forcing. A regime-switch structure with hysteresis produces the asymmetric sawtooth of glacial cycles; a time-varying critical ice volume (vc_func) can reproduce the Mid-Pleistocene Transition from 41 kyr to 100 kyr dominant periodicity.
Stocker2003BipolarSeesaw — Stocker & Johnsen (2003). Antarctic temperature evolves as a damped, lagged response to an imposed North Atlantic forcing record. A minimal representation of interhemispheric thermal coupling via ocean heat transport.
Stocker2003ExtendedSeaIceSeesaw — Four-equation extension of the seesaw model incorporating sea ice, NADW strength, and Antarctic temperature as prognostic variables. Richer internal dynamics than the basic seesaw while remaining analytically tractable.
Stommel — Stommel (1961) two-box thermohaline circulation model. Arguably the most influential minimal climate model: a warm-salty and a cold-fresh circulation mode can be simultaneously stable for the same boundary conditions. A classic demonstration of multiple equilibria in the ocean.