Mean‐State Dependence of CO2‐Forced Tropical Atlantic Sector Climate Change

Abstract

Twenty‐first‐century climate change projections are uncertain, especially on regional scales. An important source of uncertainty is that climate models exhibit biases, which limits their ability to predict climate. One of the largest biases is the too warm sea surface temperature (SST) in the eastern tropical Atlantic (TA), reflecting deficient atmospheric and oceanic circulation. Here, we show that CO2‐forced TA‐sector climate changes simulated by state‐of‐the‐art climate models exhibit a strong mean‐state dependence. In particular, models simulating largest SST warming in the eastern TA, consistent with the warming observed since the mid‐20th century, typically exhibit a more realistic mean state than models simulating largest warming in the western TA. The former models exhibit a larger climate sensitivity, and predict stronger and in part qualitatively different climate changes over the TA sector, for example in precipitation. These findings may help to reducing uncertainty in TA‐climate change projections.

Plain Language Summary: Twenty‐first‐century climate change projections are uncertain, especially on regional scales. An important source of uncertainty is that climate models exhibit biases, which limits their ability to predict climate. One of the largest biases is the too warm sea surface temperature in the eastern tropical Atlantic (TA), reflecting deficient atmospheric and oceanic circulation. Here, we show that CO2‐forced TA‐sector climate changes simulated by state‐of‐the‐art climate models exhibit a strong relationship to the quality of simulating the mean state. These findings may help to reducing uncertainty in climate change projections over the TA sector.

Key Points:

Climate projections for the tropical Atlantic sector depend on the quality of simulating present‐day conditions.

Less biased climate models provide more reliable projections.

Spread in CO2‐forced climate changes over the Tropical Atlantic region.