Skip to content

MIMOSA 1.3.2

Status: released 10 August 2026

MIMOSA 1.3.2 replaces the previous sea-level-rise (SLR) component with a deterministic, reduced-complexity formulation calibrated against IPCC AR6. The module remains small enough to run directly in both simulation and optimisation mode: it does not depend on an external climate model or soft-linked SLR projections.

This release deliberately changes the default model equations and projected SLR. Existing analyses should therefore not expect results from MIMOSA 1.3.1 and 1.3.2 to be numerically identical, even when all user-supplied parameters are unchanged.

Highlights

  • All SLR contributions use 1900 as a common reference and initialise to a total of 0.23 m in 2025.
  • Ocean thermal expansion is represented by fast and slow response boxes.
  • Glacier loss approaches a finite, temperature-dependent equilibrium.
  • Greenland follows a SIMPLE-style delayed equilibrium response.
  • Antarctic loss responds to a lagged subsurface-ocean temperature proxy and can include a smooth fast-response mechanism.
  • Future land-water-storage changes are represented explicitly.
  • Users can select coherent low, central or high physical-response parameter sets.
  • The default central response and uncertainty cases are documented against IPCC AR6 ranges.
  • The AR6 benchmark tables are calculated from the current equations and checked for freshness by the test suite.

Updated sea-level-rise equations

The old component represented thermal expansion, glaciers and Greenland with three compact equations. The new component retains a deterministic equation-based design but represents the different response times and finite ice reservoirs more explicitly.

Thermal expansion

Thermal expansion now consists of fast upper-ocean and slow deep-ocean boxes. Both relax towards a temperature-dependent equilibrium using an exact exponential update. This avoids timestep-dependent integration error for a constant forcing and captures both near-term and committed ocean response.

The first-order relaxation structure follows the reduced-complexity approach in BRICK, while the upper- and deep-ocean distinction is motivated by SURFER. The MIMOSA response coefficients are calibrated against the thermal-expansion contribution assessed in IPCC AR6 rather than copied directly from either model.

Glaciers and Greenland

The glacier contribution now relaxes towards a capped, temperature-dependent equilibrium. This removes the previous behaviour in which all available glacier ice eventually melted under any sustained temperature above the fixed equilibrium temperature.

Greenland now uses a SIMPLE-style delayed equilibrium. Its equilibrium contribution is bounded by the ice sheet's sea-level potential, while its response time decreases smoothly with warming.

Antarctica and land-water storage

Antarctica is now represented explicitly. A lagged Antarctic subsurface-ocean temperature proxy is driven by global temperature, following the scaling-and-delay concept used in LARMIP-2. The ordinary Antarctic response is calibrated against AR6 component projections. A smooth threshold term allows rapid loss to be examined without introducing a discontinuity into the optimisation model.

The high parameter set uses this mechanism as a low-likelihood, high-impact sensitivity case. It must not be interpreted as the upper endpoint of the AR6 likely range.

Land-water storage adds a scenario-independent future anomaly of 0.03 m between 2025 and 2100. Its historical contribution is already implicit in the common 2025 initial total and is not added a second time.

See Sea-level rise for the equations, scientific references, parameter interpretation and component-level results.

Low, central and high responses

The new fixed-component option selects one of three coherent deterministic parameter sets:

params = load_params()
params["model structure"]["sealevelrise options"]["projection"] = "high"
model = MIMOSA(params)

The available values are:

  • low: approximately follows the lower edge of the AR6 likely range;
  • central: the default, calibrated to remain near AR6 central estimates; and
  • high: a low-likelihood, high-impact case with faster ice-sheet response.

All three use the same historical initial conditions. They describe uncertainty in the future physical response rather than uncertainty in observed present-day sea level.

AR6 calibration and reproducibility

The documentation compares all three response sets with AR6 warming-level estimates at 2, 3 and 4 degrees C. The comparison prescribes a linear temperature path from 1.27 degrees C in 2025 to the selected warming level in 2100. It is a controlled diagnostic of the SLR equations, not an assertion that a warming level uniquely determines SLR independently of the preceding temperature pathway.

The benchmark and central component tables are generated with:

python docs/scripts/generate_slr_benchmark.py

The resulting CSV files are included in the documentation. A model test recalculates both tables and fails if a change to the SLR equations or parameters leaves the committed documentation assets out of date.

Compatibility notes

The aggregate result remains available as total_SLR, so damage components and analysis scripts that use only total sea-level rise do not require a variable-name change. Its numerical values do change because it now includes recalibrated contributions from thermal expansion, glaciers and Greenland as well as explicit Antarctic and land-water contributions.

The existing component results slr_thermal, slr_cumgsic and slr_cumgis retain their names, but their equations and calibration have changed. New output variables include:

  • slr_thermal_fast and slr_thermal_slow;
  • slr_antarctic_ocean_temp and slr_cumais; and
  • slr_cumlws.

Model extensions that accessed the old internal SLR parameters directly must be updated. In particular, the old thermal-equilibrium, fixed glacier-melt-rate and Greenland melt-rate parameters have been replaced by component sensitivities, equilibrium functions and response times. Use the documented component results or the new parameter names rather than relying on the removed internal parameters.

Validation

The SLR tests cover:

  • the common 1900 reference and 2025 initial total;
  • exact first-order relaxation across different timesteps;
  • finite glacier and ice-sheet responses;
  • smooth activation and finite-stock limitation of Antarctic loss;
  • configuration and ordering of the three projection sets;
  • the 0.03 m land-water contribution in 2100; and
  • agreement between freshly calculated AR6 benchmarks and the documentation CSV files.