Observations of tropical land (10S-10N) show surface temperature trends from 1996-2014 lower than predictions from the recent Coupled Model Intercomparison Project Phases 6 (CMIP6). Recent studies have shown that a potential source of discrepancies between model predicted and observed temperature trends is the natural variability of the climate system (Po-Chedley et al., 2022). In this GitHub, we use a machine learning method to show that the tropical land trend from 1996-2014 is significantly impacted by natural variability. After removing this natural variability from the observational record, observed trends align much better with model simulations of surface temperature trends.
aodhansweeney / surfacetrendlearing Goto Github PK
View Code? Open in Web Editor NEWA ML project aimed at removing the natural variability from forced surface temperature trends.
License: MIT License