Quantifying quasi-equilibrium breakdown in climate emulators
Name
ranjan-vranjan-sb-eaps-2026-thesis.pdf
Size
75.66 MB
Format
Adobe PDF
Checksum (MD5)
8fa46dfe979bee6896d345255cf64041
Author(s)
Ranjan, Vidya
Advisor(s)
Flierl, Glenn
Date Issued
May 2026
Publisher
Massachusetts Institute of Technology
Abstract
Accurate modeling of Earth’s future climate state is needed for addressing anthropogenic climate change. Earth System Models (ESMs) are the most accurate tools currently available to researchers, yet they are too computationally expensive to evaluate across the full range
of forcing scenarios needed for climate impact assessment. Climate emulators offer flexible approximations of ESM output tailored to specific research needs. The most common emula-
tion approach, linear pattern-scaling, approximates regional climate change as proportional to global mean temperature, yet this assumption breaks down under nonmonotonic forcing, nonlinear feedbacks, and heterogeneous forcing not captured by a single global predictor. This
thesis quantifies where and why these breakdowns occur across the CMIP6 ensemble using a first-order relaxation framework in which local temperature evolves toward a polynomial quasi-equilibrium response, with spatially varying inverse adjustment timescale λ(x) and
nonlinear feedback coefficient b(x) estimated in two sequential stages from standard CMIP6 experiments. Evaluated out-of-sample on the SSP5-3.4-Over overshoot scenario across twelve models, the framework achieves global-mean RMSE below 1 K in ten of twelve models, with tropical errors typically below 0.5 K. Systematic failures emerge at high latitudes and under reversed forcing: a near-universal warm bias traces to path-dependent hysteresis that a fixed adjustment timescale cannot capture, while polar errors of 1–2 K reflect slow ocean
adjustment and higher-order nonlinear feedbacks beyond the quadratic correction. These failure modes are structurally consistent across the ensemble and coincide with the regions of largest inter-model spread in CMIP6 projections, suggesting that the spatial structure of quasi-
equilibrium breakdown is governed by physical ocean dynamics rather than model-specific parameterization choices.
MIT Department
Massachusetts Institute of Technology. Department of Earth, Atmospheric, and Planetary Sciences
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link
