Geoscientific Model Development | 2019

Limitations of the 1 % experiment as the benchmark idealized experiment for carbon cycle intercomparison in C 4 MIP

 

Abstract


Abstract. Idealized climate change simulations are used as benchmark experiments to facilitate the\ncomparison of ensembles of climate models. In the fifth phase of the Coupled Model\nIntercomparison Project (CMIP5), the 1\u2009% per yearly compounded change in atmospheric\n CO2 concentration experiment was used to compare Earth system models with full\nrepresentations of the global carbon cycle in the Coupled Climate–Carbon Cycle Model\nIntercomparison Project (C 4 MIP). However, this “1\u2009% experiment” was never\nintended for such a purpose and implies a rise in atmospheric CO2 concentration\nat double the rate of the instrumental record. Here, we examine this choice by using an\nintermediate complexity climate model to compare the 1\u2009% experiment to an idealized\n CO2 pathway derived from a logistic function. The comparison shows three key\ndifferences in model output when forcing the model with the logistic experiment. (1)\xa0The\nmodel forced with the logistic experiment exhibits a transition of the land biosphere\nfrom a carbon sink to a carbon source, a feature absent when forcing the model with the\n1\u2009% experiment. (2)\xa0The ocean uptake of carbon comes to dominate the carbon cycle as\nemissions decline, a feature that cannot be captured when forcing a model with the\n1\u2009% experiment, as emissions always increase in that experiment. (3)\xa0The permafrost\ncarbon feedback to climate change under the 1\u2009% experiment forcing is less than half\nthe strength of the feedback seen under logistic experiment forcing. Using the logistic\nexperiment also allows smooth transition to zero or negative emissions states, allowing\nthese states to be examined without sharp discontinuities in CO2 emissions. The\nprotocol for the CMIP6 iteration of C 4 MIP again sets the 1\u2009% experiment as the\nbenchmark experiment for model intercomparison; however, clever use of the Tier\xa02\nexperiments may alleviate some of the limitations outlined here. Given the limitations of\nthe 1\u2009% experiment as the benchmark experiment for carbon cycle intercomparisons,\nadding a logistic or similar idealized experiment to the protocol of the CMIP7 iteration\nof C 4 MIP is recommended.

Volume 12
Pages 597-611
DOI 10.5194/GMD-12-597-2019
Language English
Journal Geoscientific Model Development

Full Text