Dependence of Climate Sensitivity Estimates on Internal Climate Variability During 1880-2020

doi:10.21203/rs.3.rs-2162757/v1

Download PDF

Research Article

Dependence of Climate Sensitivity Estimates on Internal Climate Variability During 1880-2020

https://doi.org/10.21203/rs.3.rs-2162757/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Observed rates of global-average deep-ocean and surface warming during 1880–2020 are matched with a 1D forcing-feedback model of vertical energy flow departures from assumed energy equilibrium driven by both anthropogenic and natural forcings. The monthly time resolution model ocean has a mixed layer, a second layer to 2,000 m depth, and a third layer extending to the ocean bottom. The model mixed layer temperature is radiatively forced with estimates of anthropogenic, volcanic, and direct solar forcings since 1765, and radiatively and non-radiatively forced with the observed history of El Niño Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO) activity since 1880. Model adjustable parameters are chosen to match observed sea surface temperature (SST) and deep ocean (0-2000 m) temperature trends during 1960–2020, as well as twenty years of lag regression relationships between sea surface temperature and satellite longwave and shortwave flux anomalies at the top of the atmosphere. The results support the dominant role of anthropogenic greenhouse gas forcing in ocean warming since 1880. Without ENSO and PDO effects, diagnosed climate sensitivity is 1.8 and 2.2 deg. C for two sea surface temperature datasets, respectively. Inclusion of ENSO and PDO improves agreement between model and observations, from 81–92% explained variance, and reduces the diagnosed sensitivity to 1.5 to 1.8 deg. C, depending on the surface temperature dataset. The global warming slowdown during 1998–2012 is also well matched after inclusion of ENSO and PDO effects.

ocean heat content

sea surface temperature

energy budget

1D model

El Niño Southern Oscillation

Pacific Decadal Oscillation

Accurate estimation of the rate of anthropogenic global warming is somewhat complicated by natural fluctuations in the climate system, especially when evaluating temperature trends over relatively short periods (e.g. IPCC 2013, Rose et al. 2014, Douville et al. 2014). This has led to some confusion among the public regarding the causes of warming of the climate system. The global warming slowdown or ‘hiatus’ seen in surface temperature data during 1998–2012 led to various explanations (e.g. Kosaka and Xie 2013; Yan et al. 2016; Xie et al. 2016; Hu and Federov 2017; Medhaug et al. 2017; Wei et al. 2021) regarding the role of natural climate variability in observed climate change and has highlighted the importance of using long time frames for estimating climate sensitivity.

The use of 3-D atmosphere-ocean general circulation models (AOGCMs, or ‘climate models’) for determining potential rates of future warming and climate change impacts do not include the observed history of interannual climate fluctuations since the models’ physics produce their own quasi-chaotic and quasi-periodic climate fluctuations, the timing of which cannot be expected to match the timing of such events in nature. It is therefore useful to employ alternative model frameworks which can empirically account for observed interannual variability in the climate system while also having some physical basis in terms of observed changes in energy flows in and out of the global atmosphere and ocean. One approach is with a simple 1-D time-dependent model framework of vertical energy flows in and out of the global-average oceans as well as in and out of the whole climate system at the top-of-atmosphere (TOA). Such a model is vastly simplified compared to AOGCMs and can be run in a spreadsheet program rather than a supercomputer.

Estimation of the equilibrium climate sensitivity (ECS) can be made from observed temperature changes between two periods separated by a sufficiently long period of time, and choosing those periods to avoid the effects of major volcanic eruptions. For example, Lewis and Curry (2018) estimated ECS by comparing two decadal time scale periods about 100 years apart when natural climate variations (including major climate eruptions) were considered to be a minimum. We expand upon this strategy by using a simple time-dependent 1-D model of vertical energy flows which includes the observed history of two known natural modes of climate variability, the El Niño Southern Oscillation (ENSO, Rasmussen and Carpenter 1982), and the Pacific Decadal Oscillation (PDO, Mantua et al. 1997). A simple 1-D model can have a minimum of adjustable parameters optimized to match the model output to a variety of observational metrics. This allows insight into the impact of natural climate variability on our interpretation of interannual and multi-decadal global temperature trends in terms of the sensitivity of the climate system to anthropogenic forcings. For example, ENSO has known impacts on radiative and non-radiative energy fluxes (Spencer and Braswell 2014, hereafter SB14), and the resulting mixture of forcing and feedback complicates the diagnosis of net climate feedback from the co-variations between radiative flux and surface temperature (e.g. Lindzen and Choi 2011; Spencer and Braswell 2011). By including the known history of ENSO activity in the 1D model as a forcing, we can estimate the extent to which ENSO causes multi-decadal variations in observed global warming trends, such as the recent global warming slowdown.

It will be demonstrated that the primary reason for the inter-decadal variation in the rate of warming, including the recent warming slowdown, is the somewhat irregular nature of ENSO activity. Because of quasi-decadal periods when warm El Niños or cool La Niñas tend to be stronger, there are resulting low-frequency departures from the general warming trend putatively attributed to anthropogenic greenhouse gas accumulation in the atmosphere.

The model simulates monthly temperature changes over the global-average ice-free oceans between 60°N and 60°S latitude in three layers. Energy exchanges between land and ocean are neglected. Similar to the SB14 model study, the 1D model is primarily driven by the RCP6 anthropogenic and volcanic radiative forcing dataset of Meinshausen et al. (2011, hereafter M11). Significant differences with SB14 include: (1) extending the analyzed time period from 1955–2011 to 1880–2020; (2) extending the 0-750 m ocean measurements to the more recent 0-2000m deep-ocean temperature analysis of Cheng et al. 2017; (3) simplifying the model vertical domain to only three layers, extending it to the ocean bottom, with simple heat transfer between layers proportional to temperature departures away from an assumed equilibrium base state; and (4) updating the CERES (Wielicki et al. 1996) satellite record of radiative flux co-variations with the observed ENSO activity to 2000–2020.

The 1D model assumes there is an equilibrium climate state at the start of the model integration in 1765, when the M11 radiative forcing dataset begins, and that feedbacks in the climate system are constantly acting to restore energy balance in the system. The vertical structure in temperature departures away from the initial state cause vertical energy flows that act to restore the assumed initial vertical temperature profile shape. But unlike AOGCMs which develop natural climate variations from the underlying physics, the simple model uses the observed history of ENSO and the PDO as empirically-based forcing, including the observed lag-regression relationship between satellite observed TOA radiative flux and past ENSO activity. This provides quantitative insight into the reasons for observed temperature variations in terms of both anthropogenic and known natural influences.

As SB14 discussed, the ENSO related energy imbalances causing sea surface temperature SST (or mixed layer temperature) variations are both non-radiative (i.e. ocean upwelling variations) as well as radiative in origin. Regarding non-radiative forcing, it is well known that El Niño is associated with a reduction of the overturning of the upper ocean which, owing to the cold ocean depths, warms the surface layer on interannual time scales. The opposite happens during La Niña, when increased overturning causes cooler surface temperatures. In addition to this non-radiative forcing of surface layer temperature variations, we will also see that there is a radiative forcing component of ENSO which is required to explain satellite observed variations in TOA radiative flux. While the radiative component does not have a strong impact on the long-term temperature trends, it will be shown it does impact the diagnosis of radiative feedbacks, which has implications for the attempted diagnosis of climate sensitivity from observed variations in the Earth’s radiative budget, as well as for climate model validation.

Regarding ocean temperatures, the 3-layer model adjustable parameters will be chosen that produce the same temperature trend in global ice-free ocean (60°N-60°S) observed SSTs from either ERSSTv5 (Huang et al. 2017) or HadSST4 (Kennedy et al. 2019) data during 1960–2020, and (2) the same ocean temperature trend in the 0-2000 m layer from the Chang et al. (2017) dataset also during 1960–2020, and (3) the bottom model layer (2000–3688 m depth) is required to warm by a small amount (0.01 deg. C) in recent decades consistent with the very uncertain rate of deep ocean heat storage below 2000 m depth estimated by Purkey and Johnson (2010). The model results are not very sensitive to this last storage term, which is very small.

Additionally, the model will be required to approximate the observed time-lagged linear regression coefficients between satellite radiative flux versus the Multivariate ENSO Index (MEI, Wolter and Timlin 2011). The multi-satellite CERES-EBAF dataset of TOA radiative fluxes covering March 2000 through January 2021 was area-averaged over the global oceans (60N-60S) and monthly anomalies were computed, while the same was done for SST, before computing the lagged regression coefficients.

The model uses time-dependent forcing-feedback equations of monthly temperature departures from assumed energy equilibrium. The model is only meant to account for the net impact of all physical processes controlling vertical energy flows which act to restore the system to energy equilibrium, without explicitly modeling those individual processes. These energy flows are assumed to be driven by temperature departures from their equilibrium values. So, for example, while clouds and water vapor are implicit components of radiative feedbacks, they are not explicitly separated. Instead, they are represented by adjustable LW and SW feedback parameters which multiply the first model layer temperature departure from equilibrium. Similarly, ocean processes such as wind-driven, deep-ocean tidal, and thermohaline transports are implicitly, not explicitly, included by the assumption they collectively act to reduce vertical temperature gradient departures from their equilibrium values.

The first model layer depth (approximately equivalent to the ocean mixed layer) is optimized to best match the time-dependent behavior of observed global-average sea surface temperature anomalies, as well as the satellite-observed TOA radiative flux variations, while also exchanging energy with the deep ocean in a realistic way. As such, the model mixed layer depth is that which produces global ocean-average temperature variability most consistent with specified time-varying forcing and feedback response.

The model equation for the mixed layer temperature departure from equilibrium (dΔT₁/dt) is

C _p1[dΔT₁/dt] = [ N(t) - λΔT₁] + S(t) + h_e1[ΔT₂ - ΔT₁] (1)

where C_p1 is the bulk heat capacity of the first (ocean mixed) layer; N represents all time-dependent external and internal radiative forcings; λ is the net feedback parameter (Forster and Taylor 2006; Forster and Gregory 2006); S represents time dependent non-radiative forcing of temperature change due to ENSO-induced changes in vertical ocean upwelling and overturning; and the last term in Eq. 1 represents the vertical transport heat between the mixed layer and the second model layer (to 2000 m depth), where h_e1 is the effective heat transfer coefficient (W m^− 2 K^− 1) between the first two model layers. Note that the ocean heat transfer coefficient has the same units as the net feedback parameter.

The second and third model layer temperatures are governed respectively by

C _p2[dΔT₂/dt] = h_e1[ΔT₁ - ΔT₂] + h_e2[ΔT₃ - ΔT₂] – S(t) (2)

and

C _p3[dΔT₃/dt] = h_e2[ΔT₂ - ΔT₃] (3)

Depending upon the model experiment, the radiative forcings N are represented by up to four component forcings

N(t) = RCP6(t) + α_LWMEI(t + 5) + α_SWMEI(t + 5) + βPDO(t), (4)

which include the M11 CMIP5 RCP6 yearly radiative forcing estimates interpolated to monthly values, ENSO-related LW and SW radiative forcings to approximately match the observed co-variations between satellite-observed global oceanic TOA radiative flux and the Multivariate ENSO Index (MEI), and PDO-related radiative forcings which when added to the other terms to reduce model residual errors during 1880–2020. The ENSO-related terms use a merged MEI index, which includes the extended MEI dataset during 1871–2005 intercalibrated with the MEI Version 2 dataset (1979–2020) during their period of mutual overlap, with the long-term mean value of 0.0019 over the full 1871–2020 period then removed. This removal of the mean MEI value over 1871–2020 assumes that there is no centennial time scale variations in ENSO activity. The monthly MEI values are then averaged to three months, and the PDO values are averaged to eleven months.

The non-radiative forcing of the mixed layer temperature due to the observed history of ENSO is

S(t) = γMEI(t + 3) (5)

which appears in Eqs. 1 and 2, causing simultaneous warming and cooling, respectively, of the first and second model layers during El Niño (depth-weighted so there is no net change in thermal energy across the two layers), and the opposite during La Niña.

The 1D model is initialized in 1765, which is when the RCP6 radiative forcings of M11 begin, and the time step is one month. The seven adjustable parameters in the model are varied to closely approximate a variety of observations. A model match to the observed SST temperature trend (1960–2020) is required to be within 0.001 C/decade, as is the model linear trend match to the 0-2000 m layer during the same period. After considerable experimentation, the adjustable depth of the model mixed layer was fixed at 50 m for all experiments presented here, a value which provided the highest correlations with the observational SST datasets during 1960–2020. Lag regression coefficients between CERES TOA LW and SW radiative flux anomalies (March 2000 through January 2021) and the MEI index were compared to the model results over the same time period and agreement was optimized by adjusting the MEI-related term coefficients in Eqs. 4 and 5. These adjustments were performed in an ad hoc fashion, and were not objectively optimized.

The experiments were in three groups: first, using just RCP6 forcings, then adding MEI forcings, then a PDO forcing. This was done separately for the HadSST4 and ERSSTv5 datasets. The MEI forcings have observational evidence, while the PDO forcings are more speculative and empirically act to reduce the unexplained residuals in the 1D model after RCP6 and MEI forcings have been included.

4.1 RCP6-only forcing

The model was first run with only the RCP6 forcings. To test for conservation of energy in the model, a net feedback parameter of 3.7 W m^− 2 K^− 1 was specified and the model run with RCP6 forcings until the total forcing reached 3.7 W m^− 2 (in 2044) and then that value was held constant for a multi-century period. This test verified that 1 deg. C of total-ocean warming was asymptotically approached by all three model layers.

Next, the model was run with various net feedback parameters and ocean heat transfer coefficients until the model produced the same SST trend as observations over 1960–2020, and the same 0-2000m temperature trend as observations. The resulting model fits to the observations during 1880–2020 are shown in Fig. 1a for the HadSST4 dataset, for the 0-2000m dataset in Fig. 1b, while statistics for this and the rest of the experiments are summarized in Table 1.

Table 1

1D model experiment specified parameters and model results using two different SST datasets for model optimization during 1960–2020. The top model layer (mixed layer) depth is 50 m in all simulations. Highest correlations between model and observations are shown in red for the periods 1880–2020, 1960–2020, and 1998–2012.
SST dataset to match	HadSST4	HadSST4	HadSST4	ERSSTv5	ERSSTv5	ERSSTv5
Forcing(s)	RCP6	RCP6 + MEI	RCP6 + MEI + PDO	RCP6	RCP6 + MEI	RCP6 + MEI + PDO
SW feedback (W m^− 2 K^− 1)		-0.87	-1.15		-0.43	-1.10
LW feedback (W m^− 2 K^− 1)		2.90	3.20		2.90	3.60
Net feedback parameter (W m^− 2 K^− 1)	1.70	2.03	2.05	2.04	2.47	2.50
Lyr1/Lyr2 effective heat transfer coeff. (W m^− 2 K^− 1)	1.10	0.97	1.02	1.47	1.30	1.35
Lyr2/Lyr3 effective heat transfer coeff. (W m-2 K-1)	0.30	0.30	0.30	0.30	0.30	0.30
MEI non-radiative forcing (W m^− 2 MEI^− 1)		0.83	0.83		0.83	0.83
MEI SW radiative forcing (W m^− 2 MEI^− 1)		0.13	0.13		0.13	0.13
MEI LW radiative forcing (W m^− 2 MEI^− 1)		0.08	0.08		0.08	0.08
PDO SW radiative forcing (W m^− 2 PDO^− 1)			-0.35			-0.35
Model ECS (deg. C)	2.18	1.82	1.80	1.81	1.50	1.48
Model trend 1880–2020 (C/decade)	0.063	0.071	0.070	0.052	0.058	0.057
SST trend 1880–2020 (C/decade)	0.066	0.066	0.066	0.058	0.058	0.058
Correlation 1880–2020	0.887	0.895	0.919	0.846	0.858	0.883
Model trend 1960–2020 (C/decade)	0.142	0.142	0.142	0.117	0.117	0.116
SST trend 1960–2020 (C/decade)	0.142	0.142	0.142	0.117	0.117	0.117
Correlation 1960–2020	0.902	0.947	0.960	0.892	0.943	0.954
Model trend 1998–2012 (C/decade)	0.161	-0.019	0.079	0.123	-0.022	0.046
SST trend 1998–2012 (C/decade)	0.029	0.029	0.029	0.059	0.059	0.059
Correlation 1998–2012	0.105	0.588	0.642	0.275	0.629	0.785

In this experiment the optimum specified net feedback parameter was 1.70 W m^− 2 K^− 1, which for an assumed 2XCO2 radiative forcing of 3.7 W m^− 2 K^− 1 corresponds to an equilibrium climate sensitivity of 2.18 deg. C (see Table 1 for specified model parameters and experiment results). Note that the model with only RCP6 forcings (Fig. 1a) cannot capture the natural interannual fluctuations in the climate system. We next perform lag correlations between the model residuals (difference between model and observations in Fig. 1a) and a limited set of indicies potentially related to natural variations in global-average ocean temperatures. The model residuals are seen in Fig. 2 to be strongly correlated with ENSO activity (MEI) and the PDO, with little relationship to the North Atlantic Oscillation (NAO, Hurrell et al. 2003) or sunspot activity.

This provides the basis for including observed MEI and PDO variations in the model as pseudo-forcings in order to better account for natural variability when addressing climate sensitivity with observational data.

4.2 RCP6 plus MEI forcings

The lag correlation relationships in Fig. 2 suggest that the next level of model improvement could be accomplished by including ENSO as a forcing. As addressed by SB14, El Niño and La Niña activity have an impact on the TOA radiative variations measured by the CERES satellite instruments. Also, the impact of ENSO on SST variations (even after averaging over the global oceans) is well known, with El Niño causing SST warming and La Niña causing SST cooling. Updating the SB14 results with CERES data from March 2000 through January 2021, the lag-regression coefficients between the MEI index and CERES anomalies over the 60°N-60°S ocean areas (dashed lines in Fig. 3) show that there is a net radiative accumulation of energy in the climate system as El Niño approaches, and then radiative loss as El Niño fades.

As described in SB14, the satellite-observed relationships (dashed lines) in Fig. 3 represent some combination of ENSO-related internal radiative forcing and radiative feedback upon ENSO-related temperature changes. With no ENSO forcing terms, we see in Fig. 3a that the model does not show any relationship between observed ENSO activity and model radiative flux variability, as we would expect since the model does not know about ENSO at this point.

Experimentation with various assumed levels of radiative forcing coefficients α_LW and α_SW (see Eq. 4) and non-radiative forcing (Eq. 5) was performed until the model produced behavior like the observations in Fig. 3.

When we add these ENSO forcings to the 1D model we get much better agreement, especially since 1960, between the model and observations as seen in Fig. 3b, Fig. 4, and in Table 1 (results column 2).

The correlation between the modeled mixed layer temperatures and SST improves from 0.902 (81.4% explained variance) with only RCP6 forcings to 0.947 (89.5% explained variance) with ENSO forcings included. The modeled lag regression structures in Fig. 3b are due to a combination of the model-specified LW and SW feedback parameters acting on the MEI-induced temperature variations, as well as the LW and SW forcing terms which are proportional to the MEI index.

Note that the addition of ENSO forcings reduces the model specified ECS from 2.18 deg. C to 1.82 deg. C, a reduction of 0.36 deg. C. This suggests that some portion of the warming during 1960–2020 was due to multi-decadal fluctuations in ENSO activity. While the inter-calibrated MEI index we used has had the long-term average removed, Fig. 5 shows there can be multi-decadal periods where the MEI index of ENSO activity has low frequency excursions away from zero.

Those fluctuations are small, but as will be shown below, sufficient to explain the apparent short-term global warming “slowdown” during 1998–2012, which was also the conclusion of Hu and Federov (2017).

These results suggest that the subtle shift to stronger El Niño activity during the late 20th Century caused a small portion of the observed SST warming, thus reducing the model-diagnosed equilibrium climate sensitivity necessary to explain the surface and deep-ocean warming during 1960–2020 in response to increasing atmospheric CO₂. It should also be kept in mind that the ENSO radiative forcing in Fig. 5 is not the only ENSO-related effect in the model. There is also the non-radiative forcing of SST (Eq. 5) which occurs as warm surface waters build up during El Niño as upper-ocean upwelling and overturning is reduced.

4.3 RCP6 plus MEI and PDO forcings

Finally, we examine the inclusion of PDO-related forcings to reduce model residuals further. This is admittedly more speculative than the MEI forcings, which had a twenty-year record of short-term co-variations with satellite radiative fluxes. While partly related to ENSO activity (e.g. Newman et al. 2003), the PDO is a lower-frequency phenomenon without sufficient satellite radiative flux data to capture several cycles in order to estimate its radiative and non-radiative impacts. For the purposes of the 1D model, we simply include a PDO forcing term as a SW forcing proportional to the monthly PDO index (Eq. 4). The resulting model temperature time series (Fig. 6) are the result of assuming that one unit change in the PDO index causes a 0.35 W m-2 change in radiative forcing, which provides a marginally better fit to the observations, as reflected in the third column of results in Table 1, and in the lag regression relationships in Fig. 3c.

The improved fit in Fig. 3c suggests that both ENSO and the PDO (which is partly a low-frequency manifestation of ENSO) are involved in the observed relationship between MEI and satellite-observed variations in TOA radiative fluxes since 2000. The resulting ECS with the PDO included (1.80 deg. C) is not significantly less than with only ENSO included (1.82 deg. C).

Also shown in Table 1 are the results when the ERSSTv5 dataset is used to match the model mixed layer temperature trend to. That SST dataset generally produces slightly lower correlations, and lower ECS values than does the HadSSTv4 dataset, with 1.81, 1.50, and 1.48 deg. C ECS for RCP6-only, RCP6 + ENSO, and RCP6 + ENSO + PDO, respectively. These results from ERSSTv5 are included to illustrate how much diagnosed ECS values can vary with different SST datasets, even when the same deep-ocean (0-2000m) data are used.

4.4 The 1998–2012 warming slowdown period

For the warming slowdown period of 1998–2012, the ERSSTv5 data turn out to produce better agreement between the model and observations. As seen in Fig. 7 (and the last three rows of Table 1), the model with only RCP6 forcing produces too much warming, with a linear trend of + 0.12 deg. C/decade versus + 0.06 deg. C/decade from the ERSSTv5 observations, with a low correlation between model and observations (0.28).

Addition of the MEI forcing to RCP6 in the model (Fig. 7b) actually creates a slight negative trend in the model (-0.02 deg. C/decade) during 1998–2012, while the correlation improves to 0.63. Finally, addition of the PDO produces nearly the same trend as the observations (+ 0.05 versus + 0.06 deg C/decade) and increases the correlation to 0.79.

A simple time-dependent 1D model of vertical energy exchanges away from assumed energy equilibrium in the ocean is presented. Optimization of the model energy flows to match observed global average temperature variability allow determination of the equilibrium climate sensitivity best supporting those temperature changes. Observed temperature variability in global ice-free ocean surface and deep-layer is well explained with the RCP6 history of radiative forcings along with forcing terms proportional to observed ENSO and PDO activity during 1880–2020. The results are generally consistent with the findings of others (e.g. Kosaka and Xie 2013; Douville et al. 2014, Hu and Federov 2017; Lewis and Curry 2017; Medhaug et al. 2017; Yan et al. 2016, and others).

The model provides a relatively easy method for evaluating the impact of natural climate variability on the quantification of anthropogenic warming rates. For the sixty-year period 1960–2020 when anthropogenic forcing of the climate system was at a maximum, the model suggests that increasing El Niño activity during that period contributed somewhat to the observed warming, thus reducing the inferred equilibrium climate sensitivity based upon those observations, from 2.18 deg. C to 1.82 deg. C. Lower ECS values are obtained if the ERSSTv5 data are used rather than the HadSST4 data.

The further inclusion of the history of PDO as an assumed radiative forcing does not change the diagnosed ECS substantially, but it does improve the model agreement with satellite observations of how TOA radiative fluxes vary with ENSO activity. Restricting analysis to the global warming slowdown period of 1998–2012 gives good agreement between model and the ERSSTv5 observations, with nearly equal trends and high correlation (0.79) at monthly time resolution.

The current study does not address the possibility of diagnosed climate sensitivity increasing over time, as is suggested by recent AOGCM experiments (Meehl et al. 2020). It instead provides a baseline that allows diagnosis of ECS from simple energy considerations during any chosen period (here, 1960–2020) for the warming rates in SST and 0-2000m ocean temperature while also accounting for selected natural climate variations during that time. Of course, to the extent that errors in either the observed warming rates or assumed radiative forcing histories exist (e.g. uncertainties in the RCP6 direct and indirect aerosol forcing), the diagnosed ECS values will change. There is also the possibility that the particular patterns of ocean heat uptake could be different in the future (e.g. Rose et al. 2014) leading to a climate sensitivity different from those diagnosed here, or that climate sensitivity to anthropogenic forcing is different from that to natural forcings (Xie et al. 2016).

Acknowledgments: This research was funded though U.S. Department of Energy contract DE-SC0019296 and the Alabama Office of the State Climatologist.

Author Declarations

- Funding: This research was funded through U.S. Department of Energy contract DE-SC0019296 and through the Alabama Office of the State Climatologist.

-Conflicts of interest/Competing Interests: The authors have no relevant financial or non-financial interests to disclose.

-Ethics approval/declarations: Not applicable.

-Consent to participate: RS and JC consented to participate in research represented here.

-Consent for publication: RS and JC consent to publish the manuscript as submitted.

-Data availability: The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

-Code availability: The software generated during the current study is available from the corresponding author on reasonable request.

-Authors' contributions: RS and JC contributed to the study concept and design. Material preparation, data collection and analysis were performed by RS. The first draft of the manuscript was written by RS and both authors commented on all versions of the manuscript. Both authors read and approved the final manuscript.

Cheng L, Trenberth KE, Fasullo J, Boyer T, Abraham J, Zhu J (2017) Improved estimates of ocean heat content from 1960-2015. Sci Adv 3. https://doi.org/10.1126/sciadv.1601545
Douville H, Voldoire A, Geoffroy O (2014) The recent global warming hiayus: What is the role of Pacific variability? Geophys. Res. Lett. 42. https://doi:10.1002/2014GL062775
Forster PM., Gregory JM (2006) The climate sensitivity and its components diagnosed from Earth Radiation Budget data. J Clim 19:39–52. https://doi.org/10.1175/JCLI3611.1
Forster PM, Taylor KE (2006) Climate forcings and climate sensitivities diagnosed from coupled climate model integrations. J Clim 19:6181–6194. https://doi.org/10.1175/JCLI3974.1
Hu S, Federov AV (2017) The extreme El Niño of 2015–2016 and the end of global warming hiatus. Geophys. Res. Lett. 44: 3816–3824. https://doi.org/10.1002/2017GL072908
Huang B, Thorne PW, Banzon VF, Boyer T, Chepurin G, Lawrimore JH, Menne MJ, Smith TM, Vose RS, Zhang H-M (2017) Extended reconstructed sea surface temperature version 5 (ERSSTv5), upgrades, validations, and intercomparisons. J Clim 30:8179-8205. https://doi.org/10.1175/JCLI-D-16-0836.1
Hurrell JW, Kushnir Y, Ottersen G, Visbek M (2003) The north atlantic oscillation: Climatic significance and environmental impact. Geophys Mono 134 AGU. https://doi.org/10.1029/2003EO080005
IPCC (2013) Climate change 2013: The physical science basis. Contribution of working group I to the fifth assessment report of the Intergovernmental Panel on Climate Change. Stocker TF,Qin D, Plattner G-K, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (eds). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA https://doi.org/10.1017/CBO9781107415324
Kennedy JJ, Rayner NA, Atkinson CP, Killick RE (2019) An ensemble data set of sea‐surface temperature change from 1850: the Met Office Hadley Centre HadSST.4.0.0.0 data set. J Geophys Res: Atmospheres 124. https://doi.org/10.1029/2018JD029867
Kosaka Y, Xie S-P (2013) Recent global-warming hiatus tied to equatorial Pacific surface cooling. Nature 501:403-407. https://doi.org/10.1038/nature12534
Lewis N., Curry J (2018) The impact of recent forcing and ocean heat uptake data on estimates of climate sensitivity. J Clim 31:6051–6071. https://doi.org/10.1175/JCLI-D-17-0667.1
Lindzen RS, Choi Y-S (2011) On the observational determination of climate sensitivity and its implications. Asia-Pac J Atmos Sci 47(4):377-390. https://doi.org/10.1007/s13143-011-0023-x
Mantua NJ, Hare SR, Zhang Y, Wallace JM, Francis RC (1997) A Pacific interdecadal climate oscillation with impacts on salmon production. Bull Amer Meteor Soc 78:1069-1079. https://doi.org/10.1175/1520-0477(1997)078<1069:APICOW>2.0.CO;2
Medhaug I, Stolpe MB, Fischer EM, Knutti R (2017) Reconciling controversies about the ‘global warming hiatus’. Nature 545:41–47. https://doi.org/10.1038/nature22315
Meehl GA, Senior CA, Eyring V, Flato G, Lamarque J-F, Stouffer RJ, Taylor KE, Schlund M (2020) Context for interpreting equilibrium climate sensitivity and transient climate response from the CMIP6 Earth system models. Sci. Adv 6(26). https://doi.org/10.1126/sciadv.aba1981
Meinshausen M, Smith SJ, Calvin KV, Daniel JS, Kainuma MLT, Lamarque J-F, Matsumoto K, Montzka SA, Raper SCB, Riahi K, Thomson AM, Velders GJM, van Vuuren D (2011) The RCP greenhouse gas concentrations and their extension from 1765 to 2300. Clim Chg 109. https://doi.org/10.1007/s10584-011-0156-z
Newman M, Compo GP, Alexander MA (2003) ENSO-forced variability of the pacific decadal oscillation. J Clim 16 (23):3853–3857. https://doi.org/10.1175/15200442(2003)016<3853:EVOTPD>2.0.CO;2
Purkey S, Johnson G (2010) Warming of global abyssal and deep southern ocean waters between the 1990s and 2000s: Contributions to global heat and sea level rise budgets. J Clim 23:6336– 6351. https://doi.org/10.1175/2010JCLI3682.1
Rasmussen EM, Carpenter TH (1982) Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon Wea Rev 110:354-384. https://doi.org/10.1175/1520-0493(1982)110<0354:VITSST>2.0.CO;2
Rose BEJ, Armour KC, Battisti DS, Feldl N, Koll DDB (2014) The dependence of transient climate sensitivity and radiative feedbacks on the spatial pattern of ocean heat uptake. Geophys. Res. Lett. 41:1071-1078. https://doi.org/10.1002/2013GL058955
Spencer RW, Braswell WD (2014) The role of ENSO in global ocean temperature changes during 1955-2011 simulated with a 1D climate model. Asia-Pac J Atmos Sci 50(2):229-237. https://doi.org/10.1007/s13143-014-0011-z
Spencer RW, Braswell WD (2011) On the misdiagnosis of surface temperature feedbacks from variations in Earth’s radiant energy balance. Rem Sens 3:1603-1613. https://doi.org/10.3390/rs3081603
Wei M, Shu Q, Song Z, Song Y, Yang X, Guo Y, Li X, Qiao F (2021) Could CMIP6 climate models reproduce the early-2000s global warming slowdown? Science China: Earth Sciences 64:853– 865. https://doi.org/10.1007/s11430-020-9740-3
Wielicki BA, Barkstrom BR, Harrison EF, Lee RB III, Smith GL, Cooper JE (1996) Clouds and the Earth’s Radiant Energy System (CERES): An Earth Observing System experiment, Bull Am Meteor Soc 77:853–868. https://doi.org/10.1175/1520-0477(1996)077<0853:CATERE>2.0.CO;2
Wolter K, Timlin MS (2011) El Niño/Southern Oscillation behaviour since 1871 as diagnosed in an extended multivariate ENSO index (MEI.ext). Int J Climatol 31:1074–1087. https://doi.org/10.1002/joc.2336
Xie S-P, Kosaka Y, Okumura YM (2016) Distinct energy budgets for anthropogenic and natural changes during global warming hiatus. Nature Geoscience 9:29-33. https://doi.org/10.1038/ngeo2581
Yan X-H, Boyer T, Trenberth K, Karl TR, Xie S-P, Nieves V, Tung K-K, Roemmich D (2016) The global warming hiatus: Slowdown or redistribution? Earth’s Future 4:11. https://doi.org/10.1002/2016EF000417

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Dependence of Climate Sensitivity Estimates on Internal Climate Variability During 1880-2020

Status:

Version 1

Abstract

Figures

1. Introduction

2. Methodology

3. The 3-layer Model

4. Model Experiment Results

4.1 RCP6-only forcing

4.2 RCP6 plus MEI forcings

4.3 RCP6 plus MEI and PDO forcings

4.4 The 1998–2012 warming slowdown period

5. Discussion

Declarations

References

Additional Declarations

Status:

Version 1