CAS-ESM2.0 dataset for the Carbon Dioxide Removal Model Intercomparison 2 Project (CDRMIP)

ABSTRACT


Introduction
Continued anthropogenic greenhouse gas emissions are changing the Earth system in a severe, pervasive, and irreversible manner.The concentration of carbon dioxide (CO2)-the most important greenhouse gas-has increased from 280 ppm in the preindustrial period to a present-day figure above 419 ppm, due to anthropogenic activities (Dlugokencky and Tans, 2023).This CO2 increase has been proven to increase radiative forcing in the atmosphere, resulting in global warming (Song et al., 2022).Global mean surface temperature (GMST), as averaged over 2011-2020, has risen by approximately 1.1 ℃, compared with the pre-industrial period (Morice et al., 2012;Allen et al., 2018;IPCC, 2021;Cao et al., 2023).Continued warming will affect biogeochemical and hydrological cycles, ecosystems, and biodiversity (Collins et al., 2013;Wattes et al., 2015), resulting in decreased ocean pH, more extreme weather and climate, rises in sea level and other consequences, all of which pose a risk to natural ecosystems and to human society.
To mitigate global warming, international efforts have been made to limit global warming to below 2 °C through removing CO2 from the atmosphere during the second half of this century (Fuss et al., 2014;Rogelj et al., 2018).At present, there is little consensus on the response of the Earth system to different scenarios of CO2 removal.To address this need, the Carbon Dioxide Removal Model Intercomparison Project (or CDRMIP), one of 23 Model Intercomparison Projects (MIPs) in CMIP6 (Eyring et al., 2016), has been proposed to help assess the potential and risks of using CDR to address climate change (Keller et al., 2018).

i n p r e s s
In the Institute of Atmospheric Physics, Chinese Academy of Sciences (IAP/CAS) it has a long history (since 1980s) in the climate/earth model development (Zhou et al., 2020).CAS-ESM2.0 is the newest version of the earth system model developed in IAP/CAS, which participated in several CMIP6 simulations, e.g., Ocean Model Intercomparison Project Phase 1 (OMIP1), Flux-Anomaly-Forced Model Intercomparison Project (FAFMIP) in addition to the DECK experiments (Zhang et al., 2020;Dong et al., 2021;Jin et al., 2021).Due to the importance of the CDR experiments, CAS-ESM2.0also completed the CDRMIP experiment following the standard protocol of Keller et al. (2018); the corresponding data have been uploaded to the ESGF data server for CMIP6 users to download and can be found at https://esgfnode.llnl.gov/projects/cmip6/.This paper is intended to release the CDRMIP experiment data by CAS-ESM2.0and provide a preliminary evaluation of CAS-ESM2.0for CDR experiments simulations which may be useful for associated data users..The remainder of the paper is organized as follows: Section 2 is the model and experiments description; followed by a basic technical validation of the CAS-ESM2.0 in CDR experiments in Section 3 and Usage notesin Section 4; finally Section 5 offers a summary.

Experimental design
i n p r e s s In this study, we conduct an idealized Tier 1 experiment that was designed to investigate CDR-induced climate "reversibility" (Keller et al., 2018).This experiment investigates the "reversibility" of the climate system by leveraging the prescribed 1% yr−1 CO2 concentration increase experiment that was conducted for generations of CMIPs.The CDRMIP experiment starts from the 1% yr −1 CO2 concentration increase experiment, 1pctCO2, and then, at the 4×CO2 concentration level, prescribes a −1% yr −1 removal of CO2 from the atmosphere to pre-industrial levels.Using CAS-ESM2.0,we conduct the 1pctCO2 experiment from year 200 of the piControl simulation.It takes about 140 years for the carbon dioxide to reach four times the pre-industrial level; thus, 200-339 is considered as the ramp-up period.Then, from year 340, the model prescribes a −1% yr −1 removal of CO2 from the atmosphere; this is considered as the ramp-down period.After the CO2 reaches its original level, the model is run for a further ~80 years.An et al. (2021) called this period the "cooling hiatus".In the following analysis, we refer to the 150-200-year period in the piControl experiment as the "piControl period", the 320-340-year period as the "CO2 peak period", and the 520-540-year period as "the cooling hiatus period" or "CO2 stabilization".The prescribed CO2 is shown in each panel of Fig. 1 in black.One ensemble experiment simulated by CAS-ESM2.0has been conducted with monthly output for all uploaded variables.

2m temperature
The 2m temperature in the simulation is shown first.When the carbon dioxide reaches four times the pre-industrial level, the temperature increases by ~5K (Fig. 1a).
Changes in temperature largely correspond to changes in CO2, suggesting a reversible i n p r e s s The prescribed CO2 variation in the simulation is also shown in each panel (black line).
Fig. 2 shows the spatial pattern of the 2m temperature anomaly in the CO2 peak period and the cooling hiatus period, compared with the piControl period, as well as the difference between the cooling hiatus period and CO2 peak period.In the CO2 peak period (320-340 years), the warming occurs throughout the globe, with the greatest warming in the higher latitudes of both hemispheres.The maximum warming can reach up to 10℃ (Fig. 2a).After the ramp-down period, the global warming magnitude weakens, compared with the CO2 peak (Fig. 2b); however, the polar amplification i n p r e s s phenomenon still exists, with greater magnitudes of warming in the Arctic and Antarctic.
The warming hole in the Atlantic Ocean exists in both of the two periods.In low latitudes, the warming is less than 2℃, with some slight regional cooling in tropical oceans.
Considering the difference between the cooling hiatus period and the CO2 peak period, the cooling is obvious from the CO2 peak to the stable period after ramp-down, with a greater magnitude of cooling on land, compared with ocean (Fig. 2c).

Precipitation
With respect to precipitation variation, its overall evolution follows that of 2m temperature, i.e., global mean precipitation increases in the ramp-up period and then decreases when CO2 falls.However, as previous studies have suggested (e.g., Kug et al., 2022), the maximum precipitation after peak CO2 appears to be delayed by ~10 years, compared with the corresponding maximum of 2m temperature (Fig. 1a).Compared with the piControl period (150-200 years), the global mean precipitation in CO2 peak increases by ~0.2 mm/d, in line with the findings of Kug et al. (2022).After CO2 returns to its original value, global mean precipitation continues to decline for decades afterwards.Compared to the original state, global mean precipitation is ~0.07 mm/d higher in the final stable period.
i n p r e s s The spatial pattern of precipitation change is shown in Fig. 3.In contrast to changes in 2m temperature, which are of greater magnitude in higher latitudes, precipitation change is mainly located in the tropical regions.Compared with the cooling hiatus period, in the CO2 peak period there are increases of greater magnitude in middleto-high latitudes.This may be associated with the higher warming in the CO2 peak i n p r e s s period.In the CO2 peak period, precipitation mainly increases in the central equatorial and eastern Pacific and the western Indian Ocean but decreases in the northwest and southeast equatorial Pacific and the western Atlantic.In the CO2 stabilization period, the precipitation change is similar in the Indian and Pacific Oceans but is of slightly weaker magnitude, compared with the CO2 peak period.However, in the tropical Atlantic, there is a dipole pattern in the CO2 stabilization period, with increases in the north and decreases in the south.This seems to be different from the CO2 peak period.The precipitation change is largely the result of the local SST change which will be examined in the next subsection (Fig. 4).
i n p r e s s Fig. 3. Same as Fig. 2, but for precipitation (units: mm/day).

Ocean temperature
In previous studies, the warmer-get-wetter mechanism has been proposed to interpret patterns of precipitation change against a background of global warming (Xie et al., 2010).Precipitation changes in tropical regions are positively correlated with spatial deviations of SST warming relative to the tropical mean, because global moist instability i n p r e s s is determined by relative SST changes.From the SST changes shown in Fig. 4, it can be seen that, in tropical regions, precipitation changes are correlated closely with local SST changes, i.e. precipitation tends to increase more if the local SST increases by a greater magnitude.
From the time series shown in Fig. 1, it can be seen that SST closely follows global mean 2m temperature.However, the whole-volume mean temperature of the ocean largely behaves differently from SST.The peak in ocean temperature occurs around year 420, about 80 years later than the CO2 peak.By the time of this peak, the ocean temperature has increased by approximately 0.6℃.Afterwards, temperature slowly declines; however, at the end of the simulation, there is still a ~0.5℃ increase in ocean temperature, compared with the original value before the ramp-up period (Fig. 1b).That is to say, although the CO2 decreases and returns to the original value, the ocean has taken in so much energy that it cannot return to its original state in just one or two centuries.
Zonally averaged cross-sections of ocean temperature anomalies in ocean basins are shown in Fig. 5.In each of the two periods, the vertical ocean temperature anomaly is similar in all ocean basins, and is of larger magnitude in the CO2 peak period.With respect to the difference between the two periods (lower panels), it can be seen that although after the ramp-down period the surface ocean temperature decreases following the decline in CO2, the deeper ocean continues to warm in all ocean basins, with the greatest warming in the Atlantic and Arctic Oceans at depths of about 1 km.This may be associated with the variation of the Atlantic meridional overturning circulation (AMOC).
i n p r e s s Fig. 4. Same as Fig. 2, but for sea surface temperature (units: ℃).

AMOC
The Atlantic Meridional Overturning Circulation (AMOC) plays an important role in modulating global climate change (Caesar et al., 2018).As can be seen in Fig. 1c, the AMOC decreases during the ramp-up period ~4Sv and stabilizes until the year ~380 in the middle of the ramp-down period, with a recorded strength of ∼14 Sv (Sv = 1 million i n p r e s s m3 s−1).Then after the year 380, AMOC increases quickly up to ∼17 Sv until the cooling hiatus period; subsequently, it decreases slightly again.Our simulation results are similar to those obtained by An et al. (2021), except that their model simulates a recovery of AMOC in the ramp-down period until it is even stronger than its original value.The AMOC change pattern is also shown in Fig. 6.There is obvious weakening of AMOC in the peak CO2 period and a slightly less-visible weakening in the cooling hiatus i n p r e s s period.The maximum weakening in the first period reaches -10 Sv; in the second period the maximum weakening is ~8 Sv.When considering the difference between the two periods, it is notable that the recovery of AMOC can be as much as 5 Sv.
i n p r e s s

Net flux at TOA
The net flux at TOA (top of atmosphere) is an essential variable to show the longterm energy budget of the whole earth system.The time series of the global mean net flux at TOA is shown in Fig. 1d.Before the ramp-up period, the model has a net energy increase of ~0.5 W/m2.The variation of the net flux at TOA is similar with other quantities.That is, first increase during the ramp-up period, and then decrease during the ramp-down period.A different feature is the tipping point between ramp-up and rampdown is almost immediate at the peak of the CO2 concentration, rather than a delay.
During the ramp-down period the net flux at TOA decreases quickly, with a larger rate than during the ramp-up period.Thus after stabilization of CO2, the net flux at TOA is ~0.5 W/m2 with a reversed sign compared with at the beginning of the ramp-up.This is interesting and deserves further research.

Usage notes
The

Summary
This study provides a preliminary evaluation of the CDRMIP experiment which we carried out using CAS-ESM2.0.The model is integrated from 200-340 years as a 1% yr−1 CO2 concentration increase experiment, and then to ~478 years as a carbon dioxide removal experiment, until CO2 returns to its original value.Finally, another 80 years is integrated in which CO2 is kept constant.Changes in 2m temperature, precipitation, sea surface temperature, ocean temperature, Atlantic meridional overturning circulation (AMOC), and sea surface height are all analyzed.Although the CO2 concentration is reversible from the ramp-up to ramp-down period, it is notable that there are many aspects of climate system that is irreversible.In the ramp-up period, global mean 2m temperature and precipitation both increase while the AMOC weakens.Values of all the above variables change in the opposite direction in the ramp-down period, but only after peaks which are delayed by several years, relative to the CO2 peak.After CO2 returns to i n p r e s s its original value, global mean 2m temperature is still ~1K higher than in the original state, and precipitation is ~0.07 mm/d higher.In addition, at the end of the simulation, there is a ~0.5 ℃ increase in ocean temperature, and a 1 Sv weakening in AMOC, compared with original values before the ramp-up period.Our model simulation produces similar results to those of comparable experiments previously reported in the literature.
The data have been uploaded for the community to use.
It is noteworthy that a single Earth system model is here used to investigate the Earth system (especially for atmosphere and ocean components) response to idealized increases and reductions in atmospheric CO2 concentrations.The use of simulations by multiple Earth system models is needed in future studies to better demonstrate the response of different components of the Earth system to CDR.Finally, we seek in our own future work to better understand the mechanisms involved in regional climate response to CDR scenarios, and accurately evaluate any irreversible behaviors.