2. Release of NorESM2.1 – results from longer simulations

Author: Dirk Olivié

2.1. Abstract

We present results of some longer simulations with NorESM2.1, both in fixed-SST as fully-coupled setup. We compare NorESM2.1 and NorESM2, and also test how the individual four CAM-Nor modifications related to ice-delimiter, heterogeneous freezing, coagulation and dry deposition contribute to the observed changes.

The TOA imbalance under 1850 conditions in fixed-SST simulations becomes more positive (\(+0.52\,\textrm{W m}^{-2}\)). Although the impact of the individual changes does not add-up linearly, the largest individual contribution comes from the correction in the ice-delimiter (\(+0.21\,\textrm{W m}^{-2}\)). The correction in coagulation also gives a small positive contribution (\(+0.05\,\textrm{W m}^{-2}\)), whereas the correction in heterogeneous freezing (\(-0.04\,\textrm{W m}^{-2}\)) and dry deposition (\(-0.11\,\textrm{W m}^{-2}\)) give a negative contribution. The aerosol effective radiative forcing (ERF) is slightly less negative in NorESM2.1 (\(-1.22\,\textrm{W m}^{-2}\)) compared to NorESM2 (\(-1.39\,\textrm{W m}^{-2}\)).

In the coupled simulations, we see a relatively modest but significant change in the warming of the global ocean. This seems to be mainly caused by the change in coagulation. Tests have only been done with the LM-version of the model (\(1.9^\circ\) x \(2.5^\circ\) horizontal resolution in the atmosphere), but we expect that similar impacts can be found in the MM-version (\(0.9^\circ\) x \(1.25^\circ\)).

2.2. Introduction

NorESM2.1 is an updated version of NorESM2. Compared to NorESM2, NorESM2.1 contains 4 bug-fixes in CAM-Nor (the atmosphere component of NorESM), and some small bugfixes in iHAMOCC (the biogeochemistry of the ocean). Here we test the behaviour of NorESM2.1 with some longer simulations. We also individually investigate the 4 bug-corrections of CAM-Nor.

In addition to these answer-changing changes, there are also technical/estheitic changes to the model without impacting simulation results (e.g., how one can activity extra aerosol diagnostics output), but we do not focus on these latter changes here. The focus of this document only on the four answer changing modifications in CAM-Nor.

The four answer-chaning modifications in CAM-Nor are related to :

1. ice number delimiter (in src/NorESM/micro_mg2_0.F90),

2. heterogeneous freezing (in chemistry/oslo_aero/hetfrz_classnuc_oslo.F90),

3. coagulation of aerosol into cloud droplets (in src/chemistry/oslo_aero/koagsub.F90), and

4. dry-deposition velocity of aerosol (in src/chemistry/oslo_aero/oslo_aerosols_intr.F90).

The second and third correction are related to errors in the model code which lead to incorrect results, but additionally had the effect that simulation results depended on the number of CPUs used for the atmospheric component. In the corrected version of the model, results do not depend anymore on the number of CPUs chosen for the atmosphere. For the comparison here, we have used the NorESM2 CMIP6 results as reference (which had a specific number of CPUs for the atmosphere). Having run NorESM2 CMIP6 results with a different number of CPUs, would have given different results, leadint to a different reference simulation.

The ice-delimeter correction considerably modifies the behoviour of NorESM in the liquid/ice water path (see Sect. Section 2.4), impacting the short-wave and long-wave cloud forcing. Therefore we have tested some additional modifications (named SH/TS) changes to micro_mg2_0.F90 and clubb_intr.F90. We have not retained these additional modifications in NorESM2.1 – they will however probably be part of NorESM2.3, accompagnied by additional tuning of the parameters club_gamma and micro_mg_dcs.

The modification in iHAMOCC (biogeochemistry component of the ocean) are active in the coupled simulations.

Most of the compsets in NorESM2 are still available in NorESM2.1, but results will be (slightly) different. We have not spun-up NorESM2.1 to a new (quasi-) equilibriun state, and ask users therefore to be careful about possible drifts in fully-coupled simulations. The results below (see Sect. Section 2.4.3) will give an initial indication of how strong the drift and the change in climate state is. For fixed-SST simulations, there is less concern.

For the test presented here, have we used the LM-version of the model, i.e., with the \(1.9 ^\circ\) x \(2.5^\circ\) horizontal resolution in the atmosphere.

For the results shown here, we have limited ourselves to global- and annual-mean values of specific model variables. However, the test simulations are available and can be used for more thorough analysis.

The experiments performed are described in Sect. Section 2.3, and results of the experiments are described in Section 2.4.

2.3. Experiments

2.3.1. Compsets

To characterize the difference between NorESM2 and NorESM2.1, we have looked at three types of simulations :

1. Fixed-SST experiments under 1850 conditions (NF1850norbc compset). These experiments are usefull as they are not affected by decadal variability or long-term drifts which hamper the analysis of the difference between NorESM2 and NorESM2.1. One is often specifically interested in the TOA imbalance.

2. Fixed-SST experiments under 1850 conditions but with 2014 aerosol emissions (NF1850norbc_aer2014 compset). Comparing experiments with 2014 and 1850 aerosol emissions allows to estimate the aerosol effective radiative forcing (ERF).

3. Fully-coupled simulations under 1850 conditions (NF1850norbc). If run over long enough time, they allow to get an idea of model drift in surface (2 m) air temperature, ocean volume temperate or AMOC.

2.3.2. Perturbation experiments

To see which change in NorESM2.1 was responsible for which behaviour change in the model, we have done experiments where the 4 modifications in CAM-Nor are tested seperately (or a combination of them). In addition have we tested an additional modification (SH/TS), which is not implemented in NorESM2.1. The different types of model setups are :

1. noresm2 (CMIP6) This reference experiment with NorESM2 has been run in 2019 (and used for CMIP6).

2. noresm2.1 (no CAM corr.) This experiment uses NorESM2.1, but where all the four CAM-Nor changes are set back to NorESM2. This setup should give results very close to NorESM2 results. The changes in BLOM/iHAMOCC are still active.

3. Experiments where only 1 NorESM2.1 modification is active :

   1. noresm2.1 (only ice-delimiter corr.)

   2. noresm2.1 (only het. freez. corr.)

   3. noresm2.1 (only koagsub corr.)

   4. noresm2.1 (only dry dep. corr.)

4. noresm2.1 (no koagsub corr.) This experiment uses 3 modifications but not a 4th one (no koagsub corr.).

5. noresm2.1 (no ice-delimiter corr.) This experiment uses 3 modifications but not a 4th one (no ice-delimiter corr.).

6. noresm2.1 These simulations have used the new NorESM2.1 code.

7. noresm2.1 + SH/TS NorESM2.1 code is used plus some additional changes. These changes should in principle be accompagnied by additional paramater modifications (micro_mg_dcs, clubb_gamma), to bring the TOA imbalance back to earlier values.

Table 2.1 Overview of the simulations

Code	Machine	NF1850norbc	NF1850norbc_aer2014	N1850
		fixed-SST	fixed-SST	fully-coupled
noresm2 (CMIP6)	Fram	ok	ok	ok
noresm2.1 (no CAM corr.)	Betzy	ok	ok	ok
noresm2.1 (only ice-delimiter corr.)	Betzy	ok		ok
noresm2.1 (only het. freez. corr.)	Betzy	ok		ok
noresm2.1 (only koagsub corr.)	Betzy	ok		ok
noresm2.1 (only dry dep. corr.)	Betzy	ok		ok
noresm2.1 (no koagsub corr.)	Betzy			ok
noresm2.1 (no ice-delimiter corr.)	Betzy	ok	ok	ok
noresm2.1	Betzy	ok	ok	ok
noresm2.1 + SH/TS	Betzy	ok		ok

2.3.3. Simulations

An overview of all the experiments can be found in Table Table 2.1. More details (name and length of the experiments) can be found in the Table Table 2.10.

The initial conditions of these simulations are the same as for the reference noresm2 (CMIP6) simulation. For the fixed-SST simulations (both NF1850norbc and NF1850norbc_aer2014) we started from 1751-01-01 of N1850_f19_tn14_20190621, and for the fully-coupled simulations we started from 1600-01-01 of N1950_f19_tn14_11062019.

2.4. Results

2.4.1. Fixed-SST simulations

In the tables we use values from the period year 5–30 (neglecting the first 4 years of the simulations).

2.4.1.1. TOA imbalance

Global-mean TOA imbalance is shown in Fig. Fig. 2.1, and the year 5–30 average is given in Table Table 2.2. In the reference similation noresm2 (CMIP6) there is a standard imbalance of around \(0.9\,\textrm{ W  m}^{-2}\) (if we derive boundary conditions from a fully-coupled N1850 simulation to be used in a NF1850norbc fixed-SST simulation, we find a TOA imbalance of around \(0.9 \,\textrm{W  m}^{-2}\)). The ice-delimiter corr. has the strongest impact and increases the TOA imbalance by around \(0.2\,\textrm{ W  m}^{-2}\). All 4 CAM-Nor modifications in NorESM2.1 (noresm2.1) together give an extra TOA imbalance of around \(0.5 \,\textrm{W  m}^{-2}\). Adding the SH/TS changes leads to a very strong reduction of the TOA imbalance by around \(-1.2 \,\textrm{W  m}^{-2}\).

Table 2.2 TOA imbalance \([\textrm{W  m}^{-2}]\) in fixed-SST simulations.

	Absolute	w.r.t. NorESM2 (CMIP6)
	\([\textrm{W m}^{-2}]\)	\([\textrm{W m}^{-2}]\)
noresm2 (CMIP6)	\(0.93 (\pm\ 0.06)\)	–
noresm2.1 (no CAM corr.)	\(0.92 (\pm\ 0.07)\)	-0.01
noresm2.1 (only ice-delimiter corr.)	\(1.14 (\pm\ 0.05)\)	0.21
noresm2.1 (only het. freez. corr.)	\(0.90 (\pm\ 0.08)\)	-0.04
noresm2.1 (only koagsub corr.)	\(0.98 (\pm\ 0.08)\)	0.05
noresm2.1 (only dry dep. corr.)	\(0.83 (\pm\ 0.09)\)	-0.11
noresm2.1 (no ice-delimiter corr.)	\(1.02 (\pm\ 0.08)\)	0.09
noresm2.1	\(1.46 (\pm\ 0.09)\)	0.53
noresm2.1 + SH/TS	\(-0.27 (\pm\ 0.11)\)	-1.21

Fig. 2.1 TOA imbalance \([\textrm{W  m}^{-2}]\) in fixed-SST simulations.

2.4.1.2. SWCF and LWCF

Global-mean shortwave and longwave cloud forcing are shown in Figs. Fig. 2.2 and Fig. 2.3, and the averages over year 5–30 are given in Tables Table 2.3 and Table 2.4. The ice-delimiter corr. leads to a positive contribution in both SWCF and LWCF by around \(0.32\) and \(0.22\,\textrm{ W  m}^{-2}\), respectively. Also dry dep. corr. leads to a positive contribution in both SWCF and LWCF by around \(0.25\) and \(0.15\,\textrm{ W  m}^{-2}\), respectively.

Fig. 2.2 Short wave cloud forcing \([\textrm{W m}^{-2}]\) in fixed-SST simulations.

Table 2.3 Short wave cloud forcing (SWCF) \([\textrm{W  m}^{-2}]\) in fixed-SST simulations.

	Absolute	w.r.t. NorESM2 (CMIP6)
	\([\textrm{W m}^{-2}]\)	\([\textrm{W m}^{-2}]\)
noresm2 (CMIP6)	\(-48.57 (\pm\ 0.06)\)	–
noresm2.1 (no CAM corr.)	\(-48.63 (\pm\ 0.07)\)	-0.06
noresm2.1 (only ice-delimiter corr.)	\(-48.25 (\pm\ 0.05)\)	0.32
noresm2.1 (only het. freez. corr.)	\(-48.56 (\pm\ 0.05)\)	0.01
noresm2.1 (only koagsub corr.)	\(-48.49 (\pm\ 0.06)\)	0.08
noresm2.1 (only dry dep. corr.)	\(-48.32 (\pm\ 0.07)\)	0.25
noresm2.1 (no ice-delimiter corr.)	\(-48.20 (\pm\ 0.05)\)	0.38
noresm2.1	\(-48.00 (\pm\ 0.06)\)	0.57
noresm2.1 + SH/TS	\(-52.01 (\pm\ 0.07)\)	-3.44

Fig. 2.3 Long wave cloud forcing \([\textrm{W m}^{-2}]\) in fixed-SST simulations.

Table 2.4 Long wave cloud forcing (LWCF) \([\textrm{W  m}^{-2}]\) in fixed-SST simulations.

	Absolute	w.r.t. NorESM2 (CMIP6)
	\([\textrm{W m}^{-2}]\)	\([\textrm{W m}^{-2}]\)
noresm2 (CMIP6)	\(25.12 (\pm\ 0.03)\)	–
noresm2.1 (no CAM corr.)	\(25.17 (\pm\ 0.03)\)	0.05
noresm2.1 (only ice-delimiter corr.)	\(24.90 (\pm\ 0.03)\)	0.22
noresm2.1 (only het. freez. corr.)	\(25.11 (\pm\ 0.04)\)	-0.01
noresm2.1 (only koagsub corr.)	\(25.07 (\pm\ 0.02)\)	-0.04
noresm2.1 (only dry dep. corr.)	\(25.27 (\pm\ 0.04)\)	0.15
noresm2.1 (no ice-delimiter corr.)	\(25.29 (\pm\ 0.03)\)	0.18
noresm2.1	\(25.35 (\pm\ 0.05)\)	0.24
noresm2.1 + SH/TS	\(28.02 (\pm\ 0.04)\)	2.91

2.4.1.3. Surface wind and DMS emissions

Global-mean surface wind is shown in Fig Fig. 2.4 (upper panel). In NorESM2.1 the surface wind is around 1 % weaker than in NorESM2.1, mainly caused by ice delimiter corr.. The noresm2.1+SH/TS simulation shows a 1 % increase in surface (10 m) wind strength.

As a consequence of the surface wind changes, also DMS emissions are affected (see Fig Fig. 2.4, lower panel).

Fig. 2.4 Surface (10 m) wind \([\textrm{m s}^{-1}]\) and DMS emission strength \([\textrm{Tg yr}^{-1}]\).

2.4.1.4. Ice crystal and cloud droplet number

Global-mean vertically-integrated ice-crystal number is shown in Fig. Fig. 2.5 (upper panel). In NoreESM2.1, the ice crystal number is almost 50 % larger than in NorESM2. Around half of the change is caused by ice-delimiter corr., but also dry dep. corr. contributes to the change.

Global-mean vertically-integrated cloud droplet number is shown in Fig. Fig. 2.5 (lower panel). In NorESM2.1, the cloud droplet number is around 15 % smaller than in NorESM2. Again, around half of the change is caused by ice-delimiter corr., with some addtional contribution from dry dep. corr. and koagsub corr..

Fig. 2.5 Global-mean ice crystal number and cloud droplet number \([\textrm{# m}^{-2}]\) in fixed-SST simulations.

2.4.1.5. Cloud ice and liquid water path

Figure Fig. 2.6 (upper panel) shows the ice cloud water path (IWP), and it increases by around 12 % in NorESM2.1 w.r.t. NorESM2., and is mainly caused by ice-delimiter corr.. The liquid cloud water path (LWP) decreases by a bit less than 10 % in NorESM2.1 compared to NorESM2, mainly caused by ice-delimiter corr..

Fig. 2.6 Global-mean cloud ice and liquid water path \([\textrm{kg m}^{-2}]\) in fixed-SST simulations.

2.4.1.6. Aerosol optical depth

Table Table 2.5 shows the total aerosol optical depth in the fixed-SST simulations. In NorESM2.1, the aerosol optical depth is around 0.02 larger than in NorESM2. The increase is mainly caused by dry dep. corr., with an additional small contribution by ice-delimiter corr..

Table 2.5 Aerosol optical depth in fixed-SST experiments.

	Absolute	W.r.t. noresm2 (CMIP6)
	[–]	[–]
noresm2 (CMIP6)	\(0.1404 (\pm\ 0.0003)\)	–
noresm2.1 (no CAM corr.)	\(0.1404 (\pm\ 0.0004)\)	0.0001
noresm2.1 (only ice-delimiter corr.)	\(0.1415 (\pm\ 0.0005)\)	0.0012
noresm2.1 (only het. freez. corr.)	\(0.1400 (\pm\ 0.0004)\)	-0.0003
noresm2.1 (only koagsub corr.)	\(0.1401 (\pm\ 0.0004)\)	-0.0003
noresm2.1 (only dry dep. corr.)	\(0.1594 (\pm\ 0.0006)\)	0.0190
noresm2.1 (no ice-delimiter corr.)	\(0.1591 (\pm\ 0.0004)\)	0.0187
noresm2.1	\(0.1610 (\pm\ 0.0005)\)	0.0207
noresm2.1 + SH/TS	\(0.1602 (\pm\ 0.0006)\)	0.0198

2.4.1.7. Aerosol burden

Figure Fig. 2.7 shows the global amount of one of the tracers to describe BC. It tracks the amount of BC within cloud droplets, having ended up there after coagulation with those cloud droplets. This quantity has increased more than 5-fold in NorESM2.1 compared to NorESM2, mainly caused by koagsub corr.. The dry dep. corr. slightly reduced this amount.

Fig. 2.7 Global total BC in cloud-droplets (after coagulation with could droplets) in fixed-SST simulations.

The global total aerosol burdens are shown in Fig. Fig. 2.8 and Table Table 2.6 for BC, OM, sulfate, dust and sea-salt. For the individual aerosol species, we observe the following:

1. The total BC aerosol burden has not much changed when going from NorESM2 to NorESM2.1, mainly due to compsensating impacts from ice-delimiter corr. on the one hand (increase), and koagsub corr. and dry dep. corr. on the other hand.

2. The second panel of the same figure shows that the OM burden has slightly increased (5 %) when going from NorESM2 to NorESM2.1, mainly due to ice-delimiter corr..

3. The third panel shows the sulfate burden, which increased by around 5 % when going from NorESM2 to NorESM2.1 – more than half of this change is caused by ice-delimter corr..

4. The fourth panel shows the dust aerosol burden, which increased by around 50 % when going from NorESM2 to NorESM2.1, principally caused by dry dep. corr..

5. The fifth panel shows the sea-salt aerosol burden, which increased by around 25-30 % when going from NorESM2 to NorESM2.1, principally caused by dry dep. corr..

Fig. 2.8 Global-mean aerosol burden [Tg] of black carbon, organic matter, sulfate, dust and sea-salt in fixed-SST simulations.

Table 2.6 Global total aerosol burden \([\textrm{Tg  yr}^{-1}]\) in fixed-SST simulations.

	BC	OM	Sulfate	Dust	Sea-salt
	Uncertainty
	\(\pm 0.0001\)	\(\pm 0.007\)	\(\pm 0.002\)	\(\pm 0.01\)	\(\pm 0.03\)
	Absolute values
noresm2 (CMIP6)	0.0418	2.064	0.571	8.61	9.03
noresm2.1 (no CAM corr.) (no CAM corr.)	0.0421	2.080	0.572	8.67	8.98
noresm2.1 (only ice-delimiter corr.)	0.0437	2.136	0.587	8.83	8.92
noresm2.1 (only het.freez. corr.)	0.0417	2.063	0.572	8.55	9.03
noresm2.1 (only koagsub corr.)	0.0401	2.058	0.570	8.49	9.05
noresm2.1 (only dry dep. corr.)	0.0403	2.065	0.572	12.86	11.51
noresm2.1 (no ice-delimiter corr.)	0.0388	2.054	0.576	13.08	11.45
noresm2.1	0.0412	2.147	0.596	13.23	11.40
noresm2.1 + SH/TS	0.0354	1.813	0.515	12.67	11.36
	Comparison w.r.t. noresm2 (CMIP6)
noresm2 (CMIP6)	–	–	–	–	–
noresm2.1 (no CAM corr.)	0.0003	0.016	0.001	0.06	-0.04
noresm2.1 (only ice-delimiter corr.)	0.0019	0.072	0.016	0.22	-0.10
noresm2.1 (only het. freez. corr.)	-0.00003	-0.001	0.001	-0.07	-0.002
noresm2.1 (only koagsub corr.)	-0.0016	-0.006	-0.001	-0.12	0.03
noresm2.1 (only dry dep. corr.)	-0.0015	0.001	0.001	4.25	2.48
noresm2.1 (no ice-delimiter corr.)	-0.0030	-0.010	0.005	4.47	2.43
noresm2.1	-0.0006	0.083	0.025	4.61	2.37
noresm2.1 + SH/TS	-0.0063	-0.251	-0.056	4.06	2.33

2.4.1.8. Aerosol lifetime

The global-mean aerosol lifetimes are shown in Fig. Fig. 2.9 and Table Table 2.7. The changes visible in the global aerosol burden are reflected in the lifetime changes (although emission changes mostly impact the burden but not the lifetime).

Fig. 2.9 Global-mean aerosol lifetime [day] of black carbon, organic matter, sulfate, dust and sea-salt in fixed-SST simulations.

Table 2.7 Global-mean aerosol life time [day] in fixed-SST simulations.

	BC	OM	Sulfate	Dust	Sea-salt
	Uncertainty
Uncertainty	\(\pm\,0.01\)	\(\pm\,0.01\)	\(\pm\,0.01\)	\(\pm\,0.01\)	\(\pm\,0.002\)
	Absolute values
noresm2 (CMIP6)	5.91	5.29	3.725	1.90	0.978
noresm2.1 (no CAM corr.)	5.96	5.32	3.734	1.89	0.980
noresm2.1 (only ice-delimiter corr.)	6.20	5.50	3.855	1.97	0.988
noresm2.1 (only het. freez. corr.)	5.91	5.30	3.730	1.88	0.982
noresm2.1 (only koagsub corr.)	5.69	5.28	3.711	1.88	0.979
noresm2.1 (only dry dep. corr.)	5.71	5.29	3.723	2.83	1.253
noresm2.1 (no ice-delimiter corr.)	5.51	5.27	3.762	2.84	1.251
noresm2.1	5.86	5.55	3.921	2.98	1.272
noresm2.1 + SH/TS	5.02	4.67	3.294	2.71	1.186
	Change w.r.t. NorESM2 (CMIP6)
noresm2 (CMIP6)	–	–	–	–	–
noresm2.1 (no CAM corr.)	0.05	0.03	0.01	-0.002	0.002
noresm2.1 (only ice-d elimiter corr.)	0.28	0.21	0.13	0.07	0.010
noresm2.1 (only het. freez. corr.)	-0.005	0.006	0.005	-0.01	0.004
noresm2.1 (only koagsub corr.)	-0.22	-0.01	-0.01	-0.01	0.001
noresm2.1 (only dry dep. corr.)	-0.20	-0.0002	-0.001	0.94	0.275
noresm2.1 (no ice-delimiter corr.)	-0.40	-0.02	0.04	0.94	0.273
noresm2.1	-0.05	0.26	0.20	1.09	0.294
noresm2.1 + SH/TS	-0.89	-0.62	-0.43	0.81	0.208

2.4.2. ERF estimates

The aerosol effective radiative forcing (ERF) is obtained by the difference in the TOA imbalance in an NF1850norbc_aer2014 simulation (fixed-SST, 1850 conditions except for 2014 aerosol emissions) and the correspondign N1850norbc simulation (fixed-SST, 1850 conditions).

When going from NorESM2 to NorESM2.1 the aerosol ERF becomes slightly weaker, changing form -1.37\(\pm\)0.06 \(\textrm{W m}^{-2}\) to -1.22\(\pm\)0.09 \(\textrm{W m}^{-2}\). The aerosol ERF in NorESM2 was known to be reasonably strong compared to other model estimates.

Table 2.8 Aerosol effective radiative forcing (ERF) for a limited number of experiments.

	ERF	ERF(ari)	ERF(aci)	ERF(alb)
	\([\textrm{W  m}^{-2}]\)	\([\textrm{W  m}^{-2}]\)	\([\textrm{W  m}^{-2}]\)	\([\textrm{W  m}^{-2}]\)
noresm2 (CMIP6)	\(-1.37\,\pm\,0.06\)	\(0.04\,\pm\,0.01\)	\(-1.35\,\pm\,0.06\)	\(-0.06\,\pm\,0.03\)
noresm2.1 (no ice-delimiter corr.)	\(-1.17\,\pm\,0.08\)	\(0.05\,\pm\,0.01\)	\(-1.21\,\pm\,0.06\)	\(-0.02\,\pm\,0.03\)
noresm2.1	\(-1.22\,\pm\,0.09\)	\(0.08\,\pm\,0.01\)	\(-1.37\,\pm\,0.07\)	\(0.06\,\pm\,0.04\)
noresm2.1 + SH/TS	\(-1.43\,\pm\,0.10\)	\(0.01\,\pm\,0.01\)	\(-1.44\,\pm\,0.07\)	\(-0.00\,\pm\,0.04\)

2.4.3. Coupled simulations

In the figures shown here, we have used an 11-year running average. All figures show global mean values, except for the surface (2 m) air temperature (additionally also northern and southern hemisphere are shown), and the figure on AMOC.

2.4.3.1. Surface (2 m) air temperature

Figure Fig. 2.10 shows the surface (2 m) air temperature averaged globally (upper panel), in the northern hemisphere (middle panel), and in the southern hemisphere (lower panel). NorESM2.1 shows a slighly higher surface (2 m) air temperature, mainly in the SH. noresm2+SH/TS shows colder surface (2 m) air temperatures, mainly in the SH.

Fig. 2.10 Global, northern hemisphere and southern hemispshere surface (\(2\,\textrm{ m}\)) air temperature \([^\circ C]\) in fully-coupled simulations.

2.4.3.2. Mean ocean volume temperature

Figure Fig. 2.11 shows the volume-mean ocean temperature. NorESM2 and no CAM corr. show very similar behaviour, i.e., a small negative trend. The simulations containing koagsub corr. (i.e, NorESM2.1, no ice-delimiter corr. and only koagsub corr.) all show a mitigated positive trend.

Fig. 2.11 Global-mean ocean volume temperature \([^\circ C]\) in fully-couple simulations.

2.4.3.3. TOA imbalance

Figure Fig. 2.12 shows the TOA radiative imbalance in the fully-coupled simulations.

Fig. 2.12 Global-mean TOA imbalance \([\textrm{W m}^{-2}]\) in fully-coupled simulations.

Table Table 2.9 shows the TOA radiative imbalance averaged over the first 100 yr of the simulations (1600–1699).

Table 2.9 Global-mean TOA imbalance \([\textrm{W m}^{-2}]\) and surface (2 m) air temperature \([^\circ C]\) averaged over year 1600–1679 of the fully-coupled simulations.

Code	TOA imbalance	Surface (2 m) air temperature
	\([\textrm{W m}^{-2}]\)	\([^\circ C]\)
noresm2 (CMIP6)	-0.01	14.45
noresm2.1 (no CAM corr.)	-0.01	14.47
noresm2.1 (no ice-delimiter corr.)	0.09	14.57
noresm2.1	0.12	14.51
noresm2.1 + SH/TS	-0.40	13.97

2.4.3.4. AMOC strength

Figure Fig. 2.13 shows the AMOC at \(26.0^\circ\ \textrm{ N}\).

Fig. 2.13 Atlantic meridional overturning circulation [Sv] in fully-coupled simulations.

2.4.3.5. SWCF and LWCF in coupled simulations

Figure Fig. 2.14 shows the global-mean SWCF (upper panel) and LWCF (lower panel).

Fig. 2.14 Global-mean SWCF and LWCF \([\textrm{W m}^{-2}]\) in fully-coupled simulations.

2.4.3.6. Surface wind and DMS emissions

Figure Fig. 2.15 shows the global averaged surface (10 m) wind (left panel) and the global total DMS emissions (right panel).

Fig. 2.15 Global-mean surface (10 m) wind \([\textrm{m s}^{-1}]\) and DMS emission strenght \([\textrm{Tg yr}^{-1}]\) in fully-coupled simulations.

2.4.3.7. Ice and cloud droplet number

Figure Fig. 2.16 shows the vertically-integrated ice crystal number (left) and cloud droplet number (right).

Fig. 2.16 Global-mean ice crystal and cloud droplet number \([\#\textrm{ m}^{-2}]\) (column integrated) in fully-coupled simulations.

2.4.3.8. Liquid and ice water path

Figure Fig. 2.17 shows the IWP and LWP.

Fig. 2.17 Global-mean ice and liquid cloud water path \([\textrm{kg  m}^{-2}]\) in fully-coupled simulations.

2.5. Conclusion

We have tried to describe some differences between NorESM2 and NorESM2.1. We find that the TOA imbalance in fixed-SST simulations is aroud 0.5 \(\textrm{W m}^{-2}\) stronger in NorESM2.1 than in NorESM2 – the strongest contribution comes from the ice-delimeter corr.. We observe a lower LWP in NorESM2.1 compard to NorESM2 (and it was already reasonably low in NorESM2). The atmospheric burden of dust and sea-salt aerosol increases considerably in NorESM2.1 compared to NorESM2. The strenght of the aerosol ERF (negative) is slightly reduced in NorESM2.1 (\(-1.22\,\textrm{ W m}^{-2}\)) compared in NorESM2 (\(-1.37\, \textrm{W m}^{-2}\)). The fully-coupled version of NorESM2.1 shows a mitigated but clear warming trend in the ocean volume temperature, mainly caused by koagsub corr.. A test with additional changes (SH/TS) appeared to give unsatisfactory results. These changes will however be implemented in NorESM2.3 including additional parameter tuning.

We think that NorESM2.1 is an interesting model version as several bugs have been corrected (4 in the atmospheric component CAM-Nor, and a few minor ones in iHAMOCC). We think that this version is able to be used – however in fully-coupled simulations one should be aware of possible drifts. The tests have been run with the LM version only (\(1.9^\circ\,\textrm{x}\,2.5^\circ\) atmospheric resolution), but we expect that the impact of the changes will be similar in the MM version (\(0.9^\circ\,\textrm{x}\,1.25^\circ\) atmospheric resolution).

2.6. Experiments

Table 2.10 Overview of all experiments

Code	Case name	Length
noresm2 (CMIP6)	NF1850norbc_f19_20191025	1–30
noresm2.1 (no CAM corr.)	NF1850norb c_f19_f19_20231204_test04	1–30
noresm2.1 (only ice-delimiter corr.)	NF1850norb c_f19_f19_20231204_test19	1–30
noresm2.1 (only het. freez. corr.)	NF1850norb c_f19_f19_20231204_test11	1–30
noresm2.1 (only koagsub corr.)	NF1850norb c_f19_f19_20231204_test12	1–30
noresm2.1 (only dry dep. corr.)	NF1850norb c_f19_f19_20231204_test13	1–30
noresm2.1 (no ice-delimiter corr.)	NF1850norbc_ f19_f19_20231204_test02	1–30
noresm2.1	NF1850norbc_ f19_f19_20231204_test17	1–30
noresm2.1 + SH/TS	NF1850norb c_f19_f19_20231204_test18	1–30
noresm2 (CMIP6)	NF1850n orbc_aer2014_f19_20191025	1–30
noresm2.1 (only ice-delimiter corr.)	NF1850norb c_f19_f19_20231204_test19	1–30
noresm2.1 (no ice-delimiter corr.)	NF1850norbc_ f19_f19_20231204_test02	1–30
noresm2.1	NF1850norbc_ f19_f19_20231204_test17	1–30
noresm2 (CMIP6)	N1850_f19_tn14_20190621	1600–2099
noresm2.1 (no CAM corr.)	N1850 _f19_tn14_20231204_test05	1600–1749
noresm2.1 (only ice-delimiter corr.)	N1850 _f19_tn14_20231204_test15	1600–1699
noresm2.1 (only het. freez. corr.)	N1850 _f19_tn14_20231204_test16	1600–1699
noresm2.1 (only koagsub corr.)	N1850 _f19_tn14_20231204_test14	1600–1699
noresm2.1 (only dry dep. corr.)	N1850 _f19_tn14_20231204_test17	1600–1699
noresm2.1 (no koagsub corr.)	N1850 _f19_tn14_20231204_test13	1600–1749
noresm2.1 (no ice-delimiter corr.)	N1850 _f19_tn14_20231204_test03	1600–1749
noresm2.1	N1850 _f19_tn14_20231204_test11	1600–1749
noresm2.1 + SH/TS	N1850 _f19_tn14_20231204_test12	1600–1699