1. Introduction
Incoming solar radiation is the ultimate energy source for our climate system. It is hence of fundamental importance to understand the impact of solar irradiance variation on Earth’s climate (Lean et al. 1995; Gray et al. 2010). Major anomalies in global surface temperature on centennial to millennial scales have been linked to changes in incoming solar energy based on proxy datasets (e.g., Jouzel et al. 2007; Swingedouw et al. 2011; Esper et al. 2012). On the other hand, sustained spaceborne measurements since 1978 have greatly facilitated the disentanglement of the role of short-term solar variation in modulating contemporary climate (Hickey et al. 1980; Fröhlich 2003; Kopp et al. 2012; Harder et al. 2019) from other various mechanisms that still have not been fully understood (Seppälä et al. 2014).
As far as solar irradiance is concerned, both the total solar irradiance (TSI) and spectral solar irradiance (SSI) matter for the climate. The TSI variations directly affect the amount of downward solar irradiance entering the climate system, which subsequently alters the surface energy budget and then affects the entire global climate through dynamical surface-air coupling and the circulation of the atmosphere (e.g., Meehl et al. 2008; Lean 2010; Jin et al. 2019; Misios et al. 2019). This is usually referred to as the “bottom-up” mechanism (Gray et al. 2010) with an emphasis on the broadband absorption of solar radiation (i.e., treating the TSI as an entity). The TSI variation during the 11-yr solar cycle has been found to be able to alter regional temperature and circulation in both the tropics (Meehl et al. 2008; Misios et al. 2019) and extratropics (Woollings et al. 2010; Ineson et al. 2011; Chen et al. 2015; Jin et al. 2019), but to hardly affect global mean surface temperature (Lockwood 2012; Solanki et al. 2013). TSI has also undergone a general increasing trend in the past century, which possibly has contributed approximately 10% to the total warming trend over the same period (Lean and Rind 2008; Imbers et al. 2013).
Besides studying the role of TSI as an entity in the climate system, SSI, which is the integrand of TSI, can also affect the climate because of the spectrally dependent atmospheric attenuation (Solanki et al. 2013) and surface reflection (Warren 2019). For instance, UV radiation is strongly absorbed by stratospheric ozone and thus radiatively heats the stratosphere. Such heating can induce circulation change, which could propagate to the troposphere through stratosphere–troposphere coupling, usually referred to as a “top-down” mechanism (Gray et al. 2010). During a solar cycle, UV variation in fraction can be larger, sometimes by one order of magnitude, than the fractional variations in the visible (VIS; 0.38–0.78 μm) or the near-infrared (NIR; 0.78–2.4 μm). As a result, most SSI–climate studies have focused on the top-down mechanism caused by the UV variation (e.g., DeLand and Cebula 2012; Ermolli et al. 2013; and references therein).
The SSI variation can also affect the climate in a bottom-up manner. The ice- and snow-covered surfaces can have considerably different albedos from the VIS to NIR (Curry et al. 2001; Briegleb and Light 2007; Warren 2019). For example, the measured spectral albedo of Antarctic snow (Grenfell et al. 1994) is well above 0.8 in the VIS, but drops to ~0.6 at 1.2 μm and to ~0.2 at 1.6 μm (Fig. 1c). While sea ice is usually covered with snow, the first-year sea ice with little snow coverage can have an albedo of ~0.7 in the VIS and ~0.4 or even less in the NIR (Brandt et al. 2005). The spectral reflectance of open water, however, is usually only ~0.06 and changes little from the VIS to the NIR (Fig. 1c). Figure 1c also shows measured spectral albedo for several other typical surfaces including sea ice, fresh snow, and barren desert. Such contrast between snowy or icy surface and open water, in terms of the VIS and NIR surface spectral albedo, implies that the variations of SSI in the VIS and NIR can directly affect the amount of solar radiance being absorbed by the surfaces (e.g., the contrasts between icy or snowy surfaces and open water), alter the surface energy budget, and consequently influence the atmospheric circulation. Moreover, the VIS and NIR have very different transmissivity in ice and snow and, thus, different efficacy in heating the ice or snow layer at various depths can be expected (Jin et al. 1994; Briegleb and Light 2007), which in turn can affect the heat stored in underlying ocean water (Light et al. 2008). In addition, the atmospheric gases and particles also exhibit spectrally dependent attenuations in the VIS and NIR (Bergstrom et al. 2003; Kindel et al. 2011; Schmidt and Pilewskie 2012). The most relevant fact here is that water vapor absorbs little radiation in the VIS but is a moderate absorber in the NIR. Such mechanisms for the SSI to influence climate have not been studied as extensively as the top-down mechanism, partly due to the lack of reliable observational constraints for the VIS and NIR SSI until 2003 (Rottman et al. 2005). In contrast, the continuous UV SSI measurements have been available since as early as 1978 (DeLand and Cebula 2008). While the interannual fractional variation of SSI in the VIS and NIR is not as large as that of UV SSI, the SSI in the VIS and NIR together consist of ~94% of the TSI. Thus, even a small fractional variation in the VIS and NIR can cause a difference of several watts per square meter or more in the SSI and, potentially, can influence the surface energy budget and climate.
(a) The solar spectral irradiance (SSI) integrated over each RRTMG_SW band, which is based on the SSI dataset used by the CESM2 as well as other climate models. More details can be found in text. (b) The difference between TSIS-1 SSI and the SSI used by the CESM2 for each RRTMG_SW band. The difference before the adjustment is shown in black and only available for 0.2–2.4 μm (the actual TSIS-1 SSI spectral coverage). The difference after adjustment is shown in red for all the bands used by RRTMG_SW. (c) Spectral albedo of different surface types, based on the surface spectral albedo database used in the MODTRAN5 (Anderson et al. 2007), a widely used radiative transfer modeling software. Solid lines are the spectral albedo at their native spectral grids, and the dash-dotted lines are the spectral albedo averaged onto the RRTMG_SW bands.
Citation: Journal of Climate 34, 10; 10.1175/JCLI-D-20-0743.1
The continuous spaceborne measurement of SSI started from the Spectral Irradiance Monitor (SIM) aboard the Solar Radiation and Climate Experiment (SORCE) satellite (Rottman et al. 2005). SORCE was then succeeded by the first Total and Spectral Solar Irradiance Sensor (TSIS-1; Carlisle et al. 2015; Pilewskie et al. 2018), which was launched in late December 2017 with a notably improved SIM instrument (Richard et al. 2011, 2020). Since March 2018, the TSIS-1 mission has been providing arguably the most accurate measurements of SSI in the VIS and NIR with a moderate spectral resolution. Thus, it is meaningful to examine how the TSIS-1 observed SSI differs from the SSI currently used by the climate models and how such difference in the SSI can affect the simulated climate. This study is motivated to address these questions. Given the length of available TSIS-1 observations, this study is focused on the static partitioning of SSI instead of its temporal variability. Specifically, we seek to answer the following question: even with the identical TSI, can the current discrepancies in our knowledge of the static partitioning between the VIS and NIR SSI cause discernable impact on the simulated surface climate? Addressing this will help us better understand the need to improve the SSI observations and monitoring, in addition to the long-term monitoring of TSI. Section 2 describes the data, model, and methods used in this study. Section 3 presents the simulation results, followed by the conclusions and further discussions in section 4.
2. Data, model, and methodology
a. SSI datasets and the SSIs used for this study
TSIS-1 was launched and mounted on the International Space Station in late December 2017. It carried two instruments, namely the Total Irradiance Monitor (TIM) and Spectral Irradiance Monitor (SIM), to measure TSI and SSI, respectively. The SIM on TSIS-1 has three channels and measures the SSI from 0.2 to 2.4 μm with varying spectral resolutions from 0.25 to 42 nm. It is a follow-up instrument of the SIM on the SORCE mission, with improved absolute accuracy (0.2%–0.4%), relative precision (0.01%), and long-term stability (0.01%–0.05% yr−1) (Coddington et al. 2016; Richard et al. 2020; Mauceri et al. 2020). In this study, we used the Level 3 (i.e., 24-hourly mean) SSI data measured by the SIM on TSIS-1 (Richard 2019) from 14 March 2018 to 21 August 2019 (i.e., ~1.5 years). The SSI data are further averaged to obtain the mean SSI over this period (referred to as TSIS-1 SSI hereafter). The SSI integrated from 0.2 to 2.4 μm, the entire spectral range covered by the SIM, is 1304.31 W m−2. The averaged TSI measured by TSIS-1 TIM (Kopp 2020) over the same period is 1361.37 W m−2.
The SSI data used by all climate modeling centers are largely based on a magnetic flux transport model that simulates solar magnetic field and irradiance (Wang et al. 2005). Multiple SSI datasets have been produced to serve different CMIP6 historical simulations (Matthes et al. 2017), which have also been used in the latest NCAR flagship climate model, the Community Earth System Model version 2.1 (CESM2 for brevity; Danabasoglu et al. 2020). As a case study here, we use a long-term average of the daily SSI data from 1978 to 2014 (referred to as CESM2 SSI hereafter). The spectral coverage of the SSI data provided to the modeling community is from 0.12 to 99.975 μm. The corresponding TSI, derived from the integration of the SSI here, is 1361.33 W m−2. The SSI integrated from 0.2 to 2.4 μm is 1310.37 W m−2, which is 6.06 W m−2 more than the counterpart from the TSIS-1 SIM observation mentioned in the previous paragraph.
Figure 1a shows the integrated SSI for each band used in the CESM2 shortwave radiation scheme, RRTMG_SW (Iacono et al. 2008). Among the 14 bands used in RRTMG_SW, one visible band (0.44–0.63 μm, hereafter referred to as the green band) and one NIR band (0.78–1.24 μm) both have more than 300 W m−2 irradiances. The third largest band-integrated irradiance is ~215 W m−2 from the other visible band (0.63–0.78 μm, the red band). In total the three bands here consist of 66.7% of the TSI. The temporal variation of the CESM2 SSI over each RRTMG_SW band is very small: among all the 14 bands, the standard deviation of daily (monthly) time series of band-integrated SSI is no more than 0.15 (0.11) W m−2, with the largest variation observed in the green band. The differences between the TSIS-1 SSI and the CESM2 SSI, without any adjustment, are shown as black bars in Fig. 1b for each RRTMG_SW band within the TSIS-1 SSI spectral coverage. The difference between TSIS-1 and CESM2 SSI is more than 4 W m−2 for the green band and the 0.78–1.24-μm NIR band, but with opposite signs. The magnitude differences are much larger than the temporal variation in the CESM2 SSI. As TSIS-1 measurements have by far the most accurate calibration of SSI and its results are adopted by the solar irradiance community as a reference standard, such a large difference between TSIS-1 SSI and CESM2 SSI warrants a close examination of the SSI used in climate models. Given the aforementioned different surface spectral reflectance in the VIS and NIR, especially the contrast between ice (or snow) surfaces and open water as shown in Fig. 1c, it is thus meaningful to investigate how such a difference in SSI can affect the surface radiation budget and the simulated climate, especially over high latitudes.
To make the TSIS-1 SSI usable for the climate model simulation and to best serve the purpose for this study, we have made further adjustments for the SSI datasets. First, to ensure that the two SSI datasets to be used in the climate simulations have the identical TSI, we scaled the long-term averaged CESM2 SSI by a factor of 1.000 03 for all wavelengths so its TSI is exactly identical to the TSI measured by the TSIS-1 TIM. Such scaled CESM2 SSI data are then used in the control run. For the perturbation run, the averaged TSIS-1 SSI is used as the SSI from 0.2 to 2.4 μm. For the wavelengths outside this spectral range (i.e., <0.2 μm or >2.4 μm), the CESM2 SSI is used but scaled with a factor so the entire integration (i.e., TSI) is identical to the TSI observed by the TSIS-1 TIM. In other words, the spectral shape from the CESM2 SSI is preserved for the spectral ranges not covered by the TSIS-1 SSI measurements but a scaling factor is used to achieve the identical TSI as used in the control run. The difference after the adjustment is shown as red bar in Fig. 1b. Between 0.2 and 2.4 μm, the difference is essentially the same as that before the adjustment, as the scaling factor for the CESM2 SSI is only 1.000 03. The difference for wavelengths longer than 2.4 μm is positive and in total 6.06 W m−2, which compensates for the difference from 0.2 to 2.4 μm and thus ensures the identical TSI. For both the control run and perturbation run, the SSI is time invariant. In this way, we ensure that the two simulations have identical TSI at all times but that the spectral partitions of such TSI between two runs are different and invariant over time. By doing so, any climatological differences between the two simulations can be attributed to the SSI differences, subject to the model’s own internal variability. Note that the influence of the model’s internal variability can be mitigated by averaging over multiple ensemble members and over sufficiently long simulation time (e.g., a decade or more).
b. Model and simulation setup
NCAR CESM2 (Danabasoglu et al. 2020) is used in this study for the climate simulations. The CESM2 configuration used here includes the atmospheric model (CAM6) coupled with a slab ocean model (DOCN-SOM) and the Los Alamos sea ice model version 5 (referred to as CICE5; Hunke et al. 2015), as well as the Community Land Model version 5 (CLM5). Interactive ocean and sea ice are warranted because, as articulated in section 1, the responses of sea surface temperature and sea ice fraction to the changes in SSI is a key mechanism through which SSI could affect surface energy balance and, subsequently, climate. The atmospheric model has a horizontal resolution of ~1.9° × 2.5° (latitude × longitude) and 32 vertical levels extending from the surface to a height of approximately 3 hPa. The ocean and sea ice components have a nominal 1° horizontal resolution.
The atmospheric radiation scheme in CESM2, as mentioned above, is the RRTMG_SW (Iacono et al. 2008) with 14 spectral bands used for the shortwave radiative transfer. The optical properties of clouds and aerosols are parameterized for each RRTMG_SW band. As far as the contrast between the visible and NIR is concerned, such optical properties of clouds and aerosols generally do not vary as drastically as does the surface reflectivity of a sea ice surface. For example, all types of clouds in the NIR are usually as reflective as in the visible, and such differences in the visible and NIR contrast between cloud and snowy or icy surface are widely used in shortwave remote sensing to distinguish cloudy pixels from snow or ice surface pixels.
In the sea ice model used in the CESM2 (i.e., CICE5) the spectral albedo of sea ice is determined by modeling shortwave radiative transfer through each snow and ice layer with the optical properties of sea ice and snow prescribed according to physical measurements obtained from the Surface Heat Budget of the Arctic field experiment (SHEBA; Curry et al. 2001). A delta-Eddington solver is used to model the shortwave radiative transfer through multiple snow and sea ice layers (Briegleb and Light 2007). The direct and diffuse components are considered separately in the delta-Eddington solver. Both sea ice and snow albedos change little from the direct to the diffuse components. The surface spectral albedo is then decided by such radiative transfer calculation within the snow and sea ice layers. A typical bare sea ice albedo in the CICE is about 0.75 in the VIS, 0.57 from 0.7 to 1.2 μm, and 0.11 from 1.2 to 5.0 μm in the NIR. Snow spectral albedo varies with the snow grain radius. The smaller the snow grain is, the less contrast exists between the visible and NIR albedo. For example, for snow made of 30-μm grains, the albedo is 0.99 in the VIS and 0.93 in the NIR, but for 2000-μm grains it becomes 0.92 in the VIS and 0.77 in the NIR (Hunke et al. 2015). In the CESM2, the open water albedo is set to be the same for both the VIS and the NIR, which is 0.07 for direct solar radiation and 0.06 for diffusive solar radiation. Similarly, the surface spectral albedo over land snow and ice areas is computed by the Snow, Ice, and Aerosol Radiative Model (SNICAR; Flanner and Zender 2005; Flanner et al. 2007) embedded in the CLM5.
Two ensembles of simulations under the present-day conditions (with aerosols and greenhouse gases fixed at their 2000 levels) are conducted. As mentioned in section 2a, the control run (referred to as CESM2 hereafter) uses the CESM2 SSI scaled to the TSIS-1 TSI value and the perturbation run (referred to as TSIS-1 hereafter) uses the observed TSIS-1 SSI with the filled SSI outside the observational spectral coverage to ensure identical TSI as the control run. Each ensemble consists of four members and each member is integrated for 20 years, with results from the last 10 years used for analyses. The differences between two such ensemble 10-yr averages can be deemed as the response of the climate system to the prescribed SSI difference after all climate feedbacks have been considered (except the feedbacks associated with ocean circulation since only slab-ocean run is used here). To contrast the response of the climate system to the prescribed SSI differences before and after any intrinsic feedback mechanisms take effect, we also conduct a set of diagnostic simulations with each of the SSI datasets. Specifically, we run the model for five days in each calendar month, with the same initial condition used on the first day of each month for both the CESM2 and TSIS-1 experiments. Then the difference between the TSIS-1 and CESM2 5-day simulations can be deemed as the direct response of the climate system to the prescribed SSI difference before any climate feedbacks take effect. The use of 12 calendar months is to cover the entire seasonal cycle since the focus here is on annual-mean difference. The result from one 5-day diagnostic run is largely the same as the result averaged over the 12 diagnostic runs.
c. Simulated high-latitude surface albedo in the CESM2
It has been well known that snow cover over the sea ice is ubiquitous over all the seasons, with accumulation through autumn, winter, and spring and minimal snow cover in summer (Warren et al. 1999). Figures 2a and 2b show the fractional coverage of open water, snow ice and bare ice (defined as snow depth above sea ice < 1 cm) in the Arctic Ocean and the Southern Ocean, respectively. Consistently with the observations, the simulated coverage of snowy sea ice reaches its maximum in spring and minimum in summer for both polar oceans, with bare ice only discernible in the Arctic summer. The Antarctic continent surface is also always covered by snow in the simulations. While the change of spectral albedo of snow from the VIS to NIR is not as drastic as that of bare sea ice, the change is still noticeable. As a result, the surface albedo derived from the modeling of radiative transfer in the snow and ice layers in the CICE5 and SNICAR still exhibits a dichotomy between the VIS and NIR. Figure 2c shows the simulated Arctic annual-mean surface albedo in the green band, with the corresponding albedo difference between the green band and the 0.78–1.24-μm band shown in Fig. 2e. Over a dominant majority of the Arctic Ocean, the albedo difference is more than 0.24. A similar VIS–NIR albedo contrast can be seen in the Southern Ocean and Antarctic continent, as shown in Figs. 2d and 2f.
(a) The fractional coverage of open water (denoted as open_water), bare sea ice (denoted as bare_ice), and sea ice covered by snow (denoted as snow_ice) over the Arctic Ocean (55°–90°N) in each season. The sea ice region with monthly-mean snow depth less than 1 cm is defined as bare sea ice. The remaining sea ice region is deemed to be sea ice covered by snow. (b) As in (a), but for the Southern Oceans (southward to 55°S). (c) The Arctic annual-mean surface albedo in the green band, derived from the ratio of upward and downward surface flux in this RRTMG_SW band. (d) As in (c), but for the Antarctic. (e),(f) The Arctic and Antarctic annual-mean surface albedo difference between the green band and the first RRTMG_SW NIR band (0.78–1.24 μm), respectively. Results are based on the ensemble mean from the CESM2 SSI run.
Citation: Journal of Climate 34, 10; 10.1175/JCLI-D-20-0743.1
3. Simulation results
This section first describes the difference in the simulated global mean shortwave radiation budget at both the TOA and surface between the two sets of experiments, followed by a more detailed discussion of the differences in zonal-mean surface shortwave radiation budget. Then the discussion will dwell more on the simulated differences in the high-latitude surface climate between the two experiments.
a. Difference in TOA and surface shortwave radiation budget
Figure 3a shows the ensemble, multiple-year means of the net shortwave flux (downward for positive; i.e., downward minus upward) for each RRTMG_SW band at the TOA and surface from the control run (i.e., CESM2 SSI run). Among all the RRTMG_SW bands, the two bands with the largest incoming SSI as shown in Fig. 1 (i.e., the green band and the 0.78–1.24-μm NIR band) also have the largest net TOA shortwave (SW) radiation, each contributing > 50 W m−2 to the total SW radiation absorbed by our planet. The same two bands also contribute the most (each >40 W m−2) to the SW radiation absorbed by the surface (red bars in Fig. 3a). The red band is the third largest in terms of the band-by-band contribution to the net downward SW radiation at both the TOA and surface. The differences between the perturbation and control runs are shown in Fig. 3b. The globally averaged SW broadband radiation difference is −0.22 W m−2 at the TOA (−0.48 W m−2 in the NIR and 0.26 W m−2 in the UV+VIS) and −0.56 W m−2 at the surface (−0.87 W m−2 in the NIR and 0.31 W m−2 in the UV+VIS). The magnitudes of such net TOA SW flux differences are comparable to the magnitudes of several anthropogenic radiative forcings, such as direct effect of man-made aerosol (−0.27 ± 0.5 W m−2) and albedo change due to land use and change (−0.15 ± 0.10 W m−2) (Myhre et al. 2013). Such broadband difference is achieved by compensating differences among different spectral bands, as shown in Fig. 3b. For example, the difference in the 0.78–1.24-μm NIR band alone is −0.93 W m−2 at the TOA and −0.64 W m−2 at the surface; the difference in the green band is comparable (0.65 W m−2 at the TOA and 0.57 W m−2 at the surface) but with opposite sign. The adjustment made in Fig. 1b for the three bands with wavelength > 2.5 μm to ensure identical TSI can be proportionally seen in the TOA difference over the same three bands. But little difference can be seen at the surface over the same three bands because little radiation in those three bands reaches the surface (as shown in Fig. 3a). Figure 3b indicates that, even though the TSI is identical in the two sets of simulations, and the net SW broadband flux difference at the surface is no more than −0.56 W m−2, the perturbed and control runs do show a ~0.6 W m−2 difference in the global-mean net surface downward flux of the green band and the 0.78–1.24-μm NIR band, but with opposite signs. The TSIS-1 − CESM2 differences of the TOA net flux in the two bands are even larger, still with the opposite signs. In other words, as far as the fractional contribution to the TSI is concerned, the discrepancies between the TSIS-1 SSI and the current SSI used by the climate models can lead to nonnegligible differences in the net downward flux in both the VIS and NIR, at both the TOA and the surface.
(a) The TOA and surface net (downward positive) shortwave fluxes for the control CESM2 simulation. (b) The differences between the TSIS-1 and CESM2 simulations. Vertical gray lines show ±1σ, where the standard deviation σ is derived from the 10 years of differences.
Citation: Journal of Climate 34, 10; 10.1175/JCLI-D-20-0743.1
The zonal-mean surface SW fluxes simulated by the control run, as well as the differences between the perturbation and control runs, are shown in Fig. 4. For the completeness of delineating the shortwave spectrum, we show the contrast between UV+VIS and NIR instead of VIS and NIR. As shown in Fig. 3b, there is little difference in the UV flux reaching the surface between the two runs, as the UV radiation is largely absorbed in the atmosphere. So the difference seen in the UV+VIS results of Fig. 4 is primarily from the VIS changes. Figure 4a shows that the upward fluxes are largest over the high latitudes, where the differences in reflected UV+VIS and NIR fluxes are also the largest. Over the latitudes with at least 10% sea ice coverage (vertical yellow shades in Figs. 4b–f), the perturbation run has less total SW flux absorbed by the surface than the control run in the 5-day diagnostic analysis (black line in Fig. 4b); such negative difference is further amplified in the coupled run (red line in Fig. 4b) due to the feedback mechanisms. The instantaneous response and feedback as reflected in the zonal-mean UV+VIS and NIR fluxes are further discussed below. In all our discussions, net flux is defined as downward positive (i.e., downward flux; upward flux is the net flux).
(a) The zonal-mean downward and upward fluxes over all the UV+VIS bands (black and green, respectively) and all the NIR bands (red and purple, respectively) from the CESM2 ensemble average. (b) The difference in ensemble zonal-mean surface net SW flux (downward positive) between the TSIS-1 and CESM2 experiments. The 5-day diagnostic run difference is in black and the long-term ensemble mean difference in red. The yellow shades indicate latitudes where zonal mean sea ice fraction > 0.1. (c),(d) The TSIS-1 and CESM2 zonal-mean differences in surface UV+VIS flux terms for the 5-day diagnostic run and long-term ensemble mean, respectively. The net flux is defined downward positive. (e),(f) As in (c) and (d), but for the differences in surface NIR terms. The thickened parts of the lines in (b), (d), and (f) indicate statistically significant difference (5% significance level).
Citation: Journal of Climate 34, 10; 10.1175/JCLI-D-20-0743.1
1) Instantaneous responses
Figures 4c and 4e summarize the instantaneous response based on the 5-day diagnostic analysis. The differences in surface downward UV+VIS and NIR flux (black solid line in Fig. 4c and red solid line in Fig. 4e) at each latitude have essentially the same magnitude but opposite signs. This indicates that, before any climate feedback mechanisms start to work, the changes in surface downward UV+VIS and NIR fluxes are consistent with what we prescribed at the TOA, namely an increase of UV+VIS flux and a corresponding decrease of NIR flux by the same amount at all latitudes. However, in the extrapolar region, the surface albedo for both UV+VIS and NIR bands is generally small such that the net flux changes (dashed lines in Figs. 4c and 4e) closely track the corresponding downward flux changes. As a result, the broadband SW flux changes in the extrapolar region is small due to the offset between the UV+VIS and NIR flux changes. However, over the high latitudes, the net UV+VIS flux difference (black dashed line in Fig. 4c) is essentially zero due to the high surface albedo but the net NIR flux difference (red dashed line in Fig. 4e) is noticeably negative due to the low surface albedo and reduced downward NIR flux. This fact implies that, before any feedback mechanisms start to work, over the polar regions the TSIS-1 run has less SW flux absorbed by the surface than the CESM2 run does. Such initial surface responses in the polar regions to prescribed SSI differences then lead to a change in surface energy budget and thus in surface temperature. Consequently, climate feedback mechanisms will respond to such changes in surface temperature.
2) Overall responses with feedbacks included
The differences in zonal-mean surface SW flux from the last 10 years of two ensemble means are shown in Figs. 4d and 4f. While the surface net UV+VIS (NIR) flux difference is still positive (negative) for the extrapolar regions, the surface net flux differences in the polar regions becomes negative for both the UV+VIS and NIR. Taking the Southern Ocean as an example, the upward UV+VIS flux differences between the TSIS-1 and CESM2 runs (green line in Fig. 4d) is indeed larger than the downward flux differences (black solid line in Fig. 4d); the upward NIR flux difference is positive in spite of the negative difference in the downward NIR flux (Fig. 4f). Such contrasts between upward and downward flux differences can only be explained by more sea ice coverage in the TSIS-1 run than in the CESM2 run, which is confirmed by differences in the simulated zonal-mean sea ice fraction as shown in Fig. 5d: the sea ice fraction difference between the two runs over the Southern Ocean can be as large as 2% and such a difference is statistically significant over most latitudes over the Southern Ocean. Similar but smaller sea ice coverage differences can be seen over the Arctic as well. Compared to Figs. 4c and 4e, it is clear that the high-latitude surface responds to the difference in the prescribed SSI, triggers the surface albedo feedback, and thus changes the surface climate.
(a),(b) Zonal-mean surface temperature and sea ice fraction climatology for the CESM2 ensemble simulation. (c) The differences in zonal-mean surface temperature climatology between the TSIS-1 and CESM2 simulations. Shaded areas denote ±1σ of annual-mean temperature differences. Thickened portions of the line indicate statistically significant differences (5% significance level). (d) As in (c), but for the sea ice fraction differences.
Citation: Journal of Climate 34, 10; 10.1175/JCLI-D-20-0743.1
The above analyses are in accordance with the increased sea ice fraction and reduced surface temperature (Figs. 5c and 5d; see more discussion in the next subsection). Together such results corroborate our reasoning in section 1; that is, even for identical TSI, the current discrepancies in the SSI partition between the VIS and NIR are large enough to cause differences in the simulated high-latitude surface mean climate by such bottom-up mechanisms due to surface spectral albedo contrasts between the VIS and NIR.
b. Difference in high-latitude surface and atmospheric fields
To further understand the differences caused by the prescribed SSI partitions between the VIS and NIR, this subsection examines the differences in surface temperature, sea ice fraction, and relevant atmospheric variables between the TSIS-1 and CESM2 simulations, with a focus on the high-latitude zones.
Figures 5a and 5b show the ensemble zonal-mean surface temperature and sea ice fraction climatology for the control run, with the corresponding TSIS-1 − CESM2 differences shown in Figs. 5c and 5d, respectively. The TSIS-1 experiment has colder temperature in both polar regions, especially in the Antarctic where zonal-mean temperature difference ranges from −0.5 to −0.7 K and is statistically significant (Fig. 5c). Correspondingly, a statistically significant increase of zonal-mean sea ice fraction up to 2.5% can be seen southward of 58°S (Fig. 5d). Such zonal-mean sea ice fraction difference is comparable to the difference caused by the long-term stratospheric ozone perturbations (Bitz and Polvani 2012; Xia et al. 2020). The covariation of sea ice fraction and surface temperature difference in the Southern Ocean can be clearly seen from their geographical distributions (Figs. 6b,d); that is, regions with large negative temperature differences are also where the large positive sea ice coverage differences occur. The Arctic also exhibits a negative difference in surface temperature and positive difference in sea ice fraction, but smaller and less statistically significant than their Antarctic counterparts. This is likely due to the large interannual variability in the Arctic. The spatial distributions of Arctic surface temperature and sea ice differences are also well correlated with each other (Figs. 6a,c).
(a) The differences in sea ice fraction climatology over the Arctic between the TSIS-1 and CESM2 simulations. The hatched areas are statistically significant at the 5% significance level. (b) As in (a), but for the Antarctic. (c),(d) As in (a) and (b), but for surface temperature (K).
Citation: Journal of Climate 34, 10; 10.1175/JCLI-D-20-0743.1
The long-term zonal-mean atmospheric temperature difference between the perturbation and control run is shown in Fig. 7. Consistent with the negative differences in high-latitude surface temperature, the entire tropospheric column in the high latitudes also exhibits negative differences between the two experiments, and such differences tend to be largest near the surface and gradually become smaller toward the tropopause. The negative air temperature difference near the surface of the Antarctic Plateau is ~−0.5 K and reduces to ~−0.1 K at 300 hPa (i.e., around the polar tropopause). Consistent with Fig. 5, the tropospheric temperature difference over the Arctic region is smaller than its Antarctic counterpart. Figure 7 suggests a reduced polar tropospheric lapse rate in response to the decrease of surface temperature from the control run to the perturbation run (i.e., a positive lapse-rate feedback in the polar region). This is consistent with our understanding of the lapse-rate feedback: although the global averaged lapse-rate feedback is negative, the polar lapse-rate feedback is positive (e.g., Pithan and Mauritsen 2014; Graversen et al. 2014). Figure 7 also shows a positive and statistically significant difference in the Northern Hemisphere subpolar lower stratosphere, which is likely caused by the meridional eddy heat flux deposited in that region in response to a change of the meridional temperature gradient from the surface through the troposphere. We note here that the performance of CESM2 in the lower stratosphere and above is not a primary focus in the CESM2 model development, as its primary focus is on the surface–troposphere climate.
The difference in zonal-mean atmospheric temperature climatology between the TSIS-1 and CESM2 simulations. The hatched areas are statistically significant at the 5% significance level.
Citation: Journal of Climate 34, 10; 10.1175/JCLI-D-20-0743.1
The ensemble zonal-mean differences in total precipitable water (TPW) and cloud water path (CWP) are plotted in Fig. 8c, with the corresponding CESM2 long-term means shown in Fig. 8a (note that TPW and CWP have different units and scales on the plots). In accordance with reduced surface temperature and tropospheric temperatures over the Antarctic, there is a negative difference in zonal-mean TPW. Such TPW difference can be largely explained by the Clausius–Clapeyron scaling with surface temperature difference (dotted line in Fig. 8c), ~7.5% change per 1 K difference in surface temperature (Held and Soden 2006). The CWP also shows a statistically significant negative difference in the Antarctic and Southern Ocean. For both quantities, the differences in the Arctic are also negative but not statistically significant. The cloud fraction differences, for high, middle, and low clouds, can be either positive or negative and are not statistically significant at all the high latitudes (Fig. 8d). In addition, we note here that the statistically significant positive difference in zonal-mean CWP and low cloud fraction around 38°S are consistent with the collocated maximum zonal-mean surface net SW flux change, which is ~−1.5 W m−2 (Fig. 4b). In other words, increased cloud absorption and reflection around 38°S result in less SW flux reaching the surface, for both UV+VIS and NIR (Figs. 4d and 4f, respectively). Note the changes in upward UV+VIS and NIR fluxes (green and purple lines in Figs. 4d and 4f, respectively) around 38°S are little because of the small surface albedo in open water and nonfrozen land.
(a) The zonal-mean climatology of total precipitable water (TPW; black) and cloud water path (CWP; red) from the CESM2 simulation. (b) As in (a), but for the high (black), middle (blue), and low (red) cloud fractions. (c) The differences in TPW and CWP between the TSIS-1 and CESM2 simulations, with thickened portions indicating the 5% significance level. The dotted line is the estimated TPW difference based on a simple Clausius–Clapeyron scaling, i.e., ~7.5% TPW change for every 1 K increase in surface temperature given relative humidity held unchanged. The yellow shades indicate latitudes where zonal mean sea ice fraction > 0.1. (d) As in (c), but for the differences in high, middle, and low cloud fractions.
Citation: Journal of Climate 34, 10; 10.1175/JCLI-D-20-0743.1
The statistically significant reductions of TPW and CWP in the Antarctic, as shown in Fig. 8c, also help explain a puzzle in Figs. 4c–f: the long-term surface downward UV+VIS flux difference (black solid line in Fig. 4d) is larger than the initial difference before the climate feedback mechanisms start to act (black solid line in Fig. 4c), and so is the contrast for the surface downward NIR flux (red solid lines in Figs. 4e and 4f). Note that water vapor has several absorption bands in the NIR, and liquid and ice clouds can attenuate the solar radiation in both VIS and NIR. Therefore, reduced TPW and CWP in the Antarctic in the TSIS-1 run favor more downward SW radiance at the surface, which indeed explains what has been shown in Figs. 4c–f.
c. Domain-averaged polar surface energy budget
The domain-averaged Arctic and Antarctic surface energy budgets are summarized in Table 1. Ensemble differences between the TSIS-1 and CESM2 experiments are shown in parentheses with the climatological values from the CESM2 experiment shown outside the parentheses. Compared to the CESM2 experiment, the TSIS-1 experiment shows increased surface upward SW flux in both domains, 0.6 W m−2 for the Arctic and 1.1 W m−2 for the Antarctic. The 0.8 W m−2 enhanced surface downward SW flux over the Antarctic domain is due to the decrease of water vapor absorption in the NIR and the decrease of cloud attenuation across the entire SW, as described in the previous subsection. The Arctic surface downward SW flux difference is only −0.1 W m−2. As for the longwave radiation at the surface, both upward and downward flux differences are negative and the net longwave flux difference is only ±0.1 W m−2 for both domains. As shown in Table 1, the reduction of downward LW fluxes is mainly due to the reduction of downward clear-sky LW fluxes in both regions. The reductions of upward and downward LW fluxes are consistent with the decreased surface and atmospheric temperatures as well as reduced TPW in both regions. As a result, the leading term to balance the difference in the surface net SW flux is the difference in latent heat, which is −0.4 W m−2 for both the Arctic and Antarctic. Such reductions in latent heat flux are also generally consistent with the increase of sea ice coverage, as sea ice behaves like a lid to shield direct evaporation from ocean water. Less evaporation is also consistent with the reduced TPW in both domains as shown in Fig. 8. The surface energy imbalance is not changed between the perturbation and control runs for both the Arctic and Antarctic.
The domain-averaged surface radiation budget over the Arctic and Antarctic. The values outside and inside of the parentheses are the climatological values from the CESM2 ensemble simulations and the differences between the TSIS-1 and CESM2 simulations, respectively. SW, LW, SH, and LH refer to shortwave, longwave, sensible heat, and latent heat, respectively. The arrows indicate the directions (downward and upward) of the radiation fields. The unit is W m−2 for all the terms. The surface energy imbalance is defined positive if the surface gains heat (i.e., downward positive).
In summary, the results above delineate a coherent picture of how the high-latitude simulations respond to the different SSI partition between VIS and NIR when the TSI is exactly identical. The TSIS-1 experiment has more incoming SW radiation in the VIS and less in the NIR than the CESM2 experiment. Initially, the surface downward SW flux differences at each latitude are consistent with such TOA differences (Figs. 4c,e). Snow and ice surfaces can reflect more in the VIS than in the NIR while open water reflects little in both the VIS and NIR. As a result, more downward VIS flux and less downward NIR flux together lead to a reduction of SW absorption by the snow and ice surfaces. This then triggers the spectral surface-albedo feedback to reduce surface temperature and lead to an expansion of sea ice coverage. The troposphere responds to such a decrease of polar surface temperature with cooler tropospheric temperature and reduced TPW. The surface energy imbalance in both polar regions remains unchanged by reduced surface SW absorption, reduced latent heat flux, and reduction of both upward and downward LW fluxes by nearly the same magnitudes. Although the differences exist in the details of responses between the Arctic and Antarctic regions, the delineation above largely holds for both polar regions.
4. Conclusions and discussion
The SSI used in climate models has not been constrained by observations as much as the TSI. Comparing the most recent TSIS-1 observations (Carlisle et al. 2015; Pilewskie et al. 2018) with the SSI used by the climate modeling communities, their TSI difference is no more than 1 W m−2 but, as shown in this study, the SSI difference in a given VIS or NIR band can be as large as 4 W m−2. This study carries out a set of idealized simulation experiments to answer one question: If two simulations are forced by the identical TSI but different SSI partitions between the VIS and NIR (as derived from the TSIS-1 observation and the current values used by the climate modeling community), to what extent can such a difference in SSI partitioning affect the simulated high-latitude climate? Our results show that indeed such SSI partitioning difference can cause discernible differences in the simulated high-latitude surface climate, primarily through a bottom-up mechanism due to the changes in surface SW absorption caused by the VIS and NIR surface albedo contrasts in the high latitudes. Comparisons between the initial 5-day response and the long-term mean difference clearly show the role played by such bottom-up mechanism. Compared to the CESM 2 experiment, the TSIS-1 experiment exhibits colder surface temperature and larger sea ice coverage in both polar regions. The polar troposphere also responds to such surface changes with colder temperature in the troposphere and less TPW. The ensemble difference is more statistically significant over the Antarctic than the Arctic, likely due to the large Arctic internal variability.
Multiple feedbacks exist in the polar region, which can amplify the original forcing signal and lead to “polar amplification” (Moritz et al. 2002; Taylor et al. 2013; Pithan and Mauritsen 2014). Although the temperature and sea ice differences shown in this study are not as large as the counterparts in the 2 × CO2 experiments (Stroeve and Notz 2015), the differences are nevertheless discernible. Therefore, the long-term SSI partition between VIS and NIR discussed here could leave its fingerprint in the decadal to centennial variations of our climate. The implications are twofold: first, in addition to the TSI, the SSI and its temporal variations need to be monitored; second, in order to faithfully simulate the observed climate changes and to attribute the changes to different natural and anthropogenic causes, the SSI partition between VIS and NIR used in climate simulations needs to be as accurate as possible.
Our simulations and physical explanations are self-consistent within the framework of surface energy budget and spectral dependence of ice and snow surface albedo. It should be noted that ocean circulation and salinity feedbacks, which are also important for sea ice evolution (Bintanja et al. 2013, 2015; Pellichero et al. 2017), are not included in the slab ocean model used here. It has been demonstrated that, at least in some regions, sea ice amount change could induce positive feedbacks from alterations to vertical salinity gradient and ocean buoyancy, leading to further sea ice change (Joly et al. 2011; Shi and Lohmann 2017). Follow-up studies can further investigate how the ocean circulation can interact with such changes induced by the SSI partitions between the VIS and NIR.
Acknowledgments
The authors are thankful to two anonymous reviewers and Prof. Stephen Warren for their insightful and constructive comments, which improved the clarity of the presentation. Particularly we want to thank Prof. Warren for pointing out the ubiquitous existence of snow cover on the sea ice. The TSIS-1 SSI and TSI datasets are available from the TSIS website (https://lasp.colorado.edu/home/tsis/data/). The SSI data used by the climate modelers can be found at https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/atm/cam/solar/. The study was supported by NASA Grant 80NSSC19K1098 awarded to the University of Michigan. We would like to acknowledge high-performance computing support from Cheyenne (doi:10.5065/D6RX99HX) provided by NCAR’s Computational and Information Systems Laboratory, sponsored by the National Science Foundation.
REFERENCES
Anderson, G. P., and Coauthors, 2007: Using the MODTRAN5 radiative transfer algorithm with NASA satellite data: AIRS and SORCE. Proc. SPIE, 6565, 65651O, https://doi.org/10.1117/12.721184.
Bergstrom, R. W., P. Pilewskie, B. Schmid, and P. B. Russell, 2003: Estimates of the spectral aerosol single scattering albedo and aerosol radiative effects during SAFARI 2000. J. Geophys. Res., 108, 8474, https://doi.org/10.1029/2002JD002435.
Bintanja, R., G. J. van Oldenborgh, S. S. Drijfhout, B. Wouters, and C. A. Katsman, 2013: Important role for ocean warming and increased ice-shelf melt in Antarctic sea-ice expansion. Nat. Geosci., 6, 376–379, https://doi.org/10.1038/ngeo1767.
Bintanja, R., G. J. van Oldenborgh, and C. A. Katsman, 2015: The effect of increased fresh water from Antarctic ice shelves on future trends in Antarctic sea ice. Ann. Glaciol., 56, 120–126, https://doi.org/10.3189/2015AoG69A001.
Bitz, C. M., and L. M. Polvani, 2012: Antarctic climate response to stratospheric ozone depletion in a fine resolution ocean climate model. Geophys. Res. Lett., 39, L20705, https://doi.org/10.1029/2012GL053393.
Brandt, R. E., S. G. Warren, A. P. Worby, and T. C. Grenfell, 2005: Surface albedo of the Antarctic sea ice zone. J. Climate, 18, 3606–3622, https://doi.org/10.1175/JCLI3489.1.
Briegleb, B. P., and B. Light, 2007: A delta-Eddington multiple scattering parameterization for solar radiation in the sea ice component of the Community Climate System Model. NCAR Tech. Note NCAR/TN-472+STR, 100 pp., https://doi.org/10.5065/D6B27S71.
Carlisle, C., R. Wedge, D. Wu, H. Stello, and R. Robinson, 2015: Total and Spectral Solar Irradiance Sensor (TSIS) project overview. 7 pp., https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20150023359.pdf.
Chen, H., H. Ma, X. Li, and S. Sun, 2015: Solar influences on spatial patterns of Eurasian winter temperature and atmospheric general circulation anomalies. J. Geophys. Res. Atmos., 120, 8642–8657, https://doi.org/10.1002/2015JD023415.
Coddington, O., J. L. Lean, P. Pilewskie, M. Snow, and D. Lindholm, 2016: A solar irradiance climate data record. Bull. Amer. Meteor. Soc., 97, 1265–1282, https://doi.org/10.1175/BAMS-D-14-00265.1.
Curry, J. A., J. L. Schramm, D. K. Perovich, and J. O. Pinto, 2001: Applications of SHEBA/FIRE data to evaluation of snow/ice albedo parameterizations. J. Geophys. Res., 106, 15 345–15 355, https://doi.org/10.1029/2000JD900311.
Danabasoglu, G., and Coauthors, 2020: The Community Earth System Model version 2 (CESM2). J. Adv. Model. Earth Syst., 12, e2019MS001916, https://doi.org/10.1029/2019MS001916.
DeLand, M. T., and R. P. Cebula, 2008: Creation of a composite solar ultraviolet irradiance data set. J. Geophys. Res., 113, A11103, https://doi.org/10.1029/2008JA013401.
DeLand, M. T., and R. P. Cebula, 2012: Solar UV variations during the decline of Cycle 23. J. Atmos. Sol.-Terr. Phys., 77, 225–234, https://doi.org/10.1016/j.jastp.2012.01.007.
Ermolli, I., and Coauthors, 2013: Recent variability of the solar spectral irradiance and its impact on climate modelling. Atmos. Chem. Phys., 13, 3945–3977, https://doi.org/10.5194/acp-13-3945-2013.
Esper, J., and Coauthors, 2012: Orbital forcing of tree-ring data. Nat. Climate Change, 2, 862–866, https://doi.org/10.1038/nclimate1589.
Flanner, M. G., and C. S. Zender, 2005: Snowpack radiative heating: Influence on Tibetan Plateau climate. Geophys. Res. Lett., 32, L06501, https://doi.org/10.1029/2004GL022076.
Flanner, M. G., C. S. Zender, J. T. Randerson, and P. J. Rasch, 2007: Present day climate forcing and response from black carbon in snow. J. Geophys. Res., 112, D11202, https://doi.org/10.1029/2006JD008003.
Fröhlich, C., 2003: Long term behaviour of space radiometers. Metrologia, 40, S60–S65, https://doi.org/10.1088/0026-1394/40/1/314.
Graversen, R. G., P. G. Langen, and T. Mauritsen, 2014: Arctic amplification in CCSM4: Contributions from the lapse rate and surface albedo feedbacks. J. Climate, 27, 4433–4450, https://doi.org/10.1175/JCLI-D-13-00551.1.
Gray, L. J., and Coauthors, 2010: Solar influences on climate. Rev. Geophys., 48, RG4001, https://doi.org/10.1029/2009RG000282.
Grenfell, T. C., S. G. Warren, and P. C. Mullen, 1994: Reflection of solar radiation by the Antarctic snow surface at ultraviolet, visible, and near-infrared wavelengths. J. Geophys. Res., 99, 18 669–18 684, https://doi.org/10.1029/94JD01484.
Harder, J. W., S. Beìland, and M. Snow, 2019: SORCE-based solar spectral irradiance (SSI) record for input into chemistry-climate studies. Earth Space Sci., 6, 2487–2507, https://doi.org/10.1029/2019EA000737.
Held, I. M., and B. J. Soden, 2006: Robust responses of the hydrological cycle to global warming. J. Climate, 19, 5686–5699, https://doi.org/10.1175/JCLI3990.1.
Hickey, J. R., L. L. Stowe, H. Jacobowitz, P. Pellegrino, R. H. Maschhoff, F. House, and T. H. Vonder Haar, 1980: Initial solar irradiance determinations from Nimbus 7 cavity radiometer measurements. Science, 208, 281–283, https://doi.org/10.1126/science.208.4441.281.
Hunke, E. C., W. H. Lipscomb, A. K. Turner, N. Jeffery, and S. Elliott, 2015: CICE: The Los Alamos Sea Ice Model documentation and software user’s manual version 5.1. Doc. LA-CC-06–012, 116 pp., https://github.com/CICE-Consortium/CICE-svn-trunk/blob/master/cicedoc/cicedoc.pdf.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, https://doi.org/10.1029/2008JD009944.
Imbers, J., A. Lopez, C. Huntingford, and M. R. Allen, 2013: Testing the robustness of the anthropogenic climate change detection statements using different empirical models. J. Geophys. Res. Atmos., 118, 3192–3199, https://doi.org/10.1002/jgrd.50296.
Ineson, S., A. A. Scaife, J. R. Knight, J. C. Manners, N. J. Dunstone, L. J. Gray, and D. H. Joanna, 2011: Solar forcing of winter climate variability in the Northern Hemisphere. Nat. Geosci., 4, 753–757, https://doi.org/10.1038/ngeo1282.
Jin, C., B. Wang, J. Liu, N. Liang, and Y. Mi, 2019: Decadal variability of northern Asian winter monsoon shaped by the 11-year solar cycle. Climate Dyn., 53, 6559–6568, https://doi.org/10.1007/s00382-019-04945-4.
Jin, Z., K. Stamnes, W. F. Weeks, and S.-C. Tsay, 1994: The effect of sea ice on the solar energy budget in the atmosphere–sea ice–ocean system: A model study. J. Geophys. Res., 99, 25 281–25 294, https://doi.org/10.1029/94JC02426.
Joly, S., S. Senneville, D. Caya, and F. J. Saucier, 2011: Sensitivity of Hudson Bay Sea ice and ocean climate to atmospheric temperature forcing. Climate Dyn., 36, 1835–1849, https://doi.org/10.1007/s00382-009-0731-4.
Jouzel, J., and Coauthors, 2007: Orbital and millennial Antarctic climate variability over the past 800,000 years. Science, 317, 793–796, https://doi.org/10.1126/science.1141038.
Kindel, B. C., P. Pilewskie, K. S. Schmidt, O. Coddington, and M. D. King, 2011: Solar spectral absorption by marine stratus clouds: Measurements and modeling. J. Geophys. Res., 116, D10203, https://doi.org/10.1029/2010JD015071.
Kopp, G., 2020: TSIS TIM level 3 total solar irradiance 24-hour means V03. Goddard Earth Sciences Data and Information Services Center (GES DISC), accessed 24 August 2020, https://doi.org/10.5067/TSIS/TIM/DATA306.
Kopp, G., A. Fehlmann, W. Finsterle, D. Harber, K. Heuerman, and R. Willson, 2012: Total solar irradiance data record accuracy and consistency improvements. Metrologia, 49, S29–S33, https://doi.org/10.1088/0026-1394/49/2/S29.
Lean, J. L., 2010: Cycles and trends in solar irradiance and climate. Wiley Interdiscip. Rev.: Climate Change, 1, 111–122, https://doi.org/10.1002/wcc.18.
Lean, J. L., and D. H. Rind, 2008: How natural and anthropogenic influences alter global and regional surface temperatures 1889 to 2006. Geophys. Res. Lett., 35, L18701, https://doi.org/10.1029/2008GL034864.
Lean, J. L., J. Beer, and J. Bradley, 1995: Reconstruction of solar irradiance since 1610: Implications for climate change. Geophys. Res. Lett., 22, 3195–3198, https://doi.org/10.1029/95GL03093.
Light, B., T. C. Grenfell, and D. K. Perovich, 2008: Transmission and absorption of solar radiation by Arctic sea ice during the melt season. J. Geophys. Res., 113, C03023, https://doi.org/10.1029/2006JC003977.
Lockwood, M., 2012: Solar influence on global and regional climates. Surv. Geophys., 33, 503–534, https://doi.org/10.1007/s10712-012-9181-3.
Matthes, K., and Coauthors, 2017: Solar forcing for CMIP6 (v3.2). Geosci. Model Dev., 10, 2247–2302, https://doi.org/10.5194/gmd-10-2247-2017.
Mauceri, S., E. Richard, P. Pilewskie, D. Harber, O. Coddington, S. Béland, M. Chambliss, and S. Carson, 2020: Degradation correction of TSIS SIM. Sol. Phys., 295, 152, https://doi.org/10.1007/s11207-020-01707-y.
Meehl, G. A., J. M. Arblaster, G. Branstator, and H. van Loon, 2008: A coupled air–sea response mechanism to solar forcing in the Pacific region. J. Climate, 21, 2883–2897, https://doi.org/10.1175/2007JCLI1776.1.
Misios, S., L. J. Gray, M. F. Knudsen, C. Karoff, H. Schmidt, and J. D. Haigh, 2019: Slowdown of the Walker circulation at solar cycle maximum. Proc. Natl. Acad. Sci. USA, 116, 7186–7191, https://doi.org/10.1073/pnas.1815060116.
Moritz, R. E., C. M. Bitz, and E. J. Steig, 2002: Dynamics of recent climate change in the Arctic. Science, 297, 1497–1502, https://doi.org/10.1126/science.1076522.
Myhre, G., and Coauthors, 2013: Anthropogenic and natural radiative forcing. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 675–688.
Pellichero, V., J.-B. Sallée, S. Schmidtko, F. Roquet, and J.-B. Charrassin, 2017: The ocean mixed layer under Southern Ocean sea-ice: Seasonal cycle and forcing. J. Geophys. Res. Oceans, 122, 1608–1633, https://doi.org/10.1002/2016JC011970.
Pilewskie, P., G. Kopp, E. Richard, O. Coddington, T. Sparn, and T. Woods, 2018: TSIS-1 and continuity of the total and spectral solar irradiance climate data record. Proc. 20th EGU General Assembly, Vienna, Austria, European Geosciences Union, 5527, https://ui.adsabs.harvard.edu/abs/2018EGUGA.20.5527P/abstract.
Pithan, F., and T. Mauritsen, 2014: Arctic amplification dominated by temperature feedbacks in contemporary climate models. Nat. Geosci., 7, 181–184, https://doi.org/10.1038/ngeo2071.
Richard, E., 2019: TSIS SIM level 3 solar spectral irradiance 24-hour means V04 (TSIS_SSI_L3_24HR) at GES DISC, accessed 6 April 2020, https://catalog.data.gov/dataset/tsis-sim-level-3-solar-spectral-irradiance-24-hour-means-v04-tsis-ssi-l3-24hr-at-ges-disc.
Richard, E., and Coauthors, 2011: Future long-term measurements of solar spectral irradiance by the TSIS spectral irradiance monitor: Improvements in measurement accuracy and stability. Proc. 11th Int. Conf. on New Developments and Applications in Optical Radiometry, Maui, Hawaii, INV004, 5–6, http://newrad2011.aalto.fi/attachments/File/NewRAD2011_AbstractCollection_20110704.pdf.
Richard, E., D. Harber, O. Coddington, G. Drake, J. Rutkowski, M. Triplett, P. Pilewskie, and T. Woods, 2020: SI-traceable spectral irradiance radiometric characterization and absolute calibration of the TSIS-1 Spectral Irradiance Monitor (SIM). Remote Sens., 12, 1818–1844, https://doi.org/10.3390/rs12111818.
Rottman, G., J. Harder, J. Fontenla, T. N. Woods, O. R. White, and G. Lawrence, 2005: The Spectral Irradiance Monitor (SIM): Early observations. Sol. Phys., 230, 205–224, https://doi.org/10.1007/s11207-005-1530-7.
Schmidt, S., and P. Pilewskie, 2012: Airborne measurements of spectral shortwave radiation in cloud and aerosol remote sensing and energy budget studies. Light Scattering Reviews, Vol. 6, A. Kokhanovsky, Ed., Springer, 239–288, https://doi.org/10.1007/978-3-642-15531-4_6.
Seppälä, A., K. Matthes, C. E. Randall, and I. A. Mironova, 2014: What is the solar influence on climate? Overview of activities during CAWSES-II. Prog. Earth Planet. Sci., 1, 24, https://doi.org/10.1186/s40645-014-0024-3.
Shi, X., and G. Lohmann, 2017: Sensitivity of open-water ice growth and ice concentration evolution in a coupled atmosphere–ocean–sea ice model. Dyn. Atmos. Oceans, 79, 10–30, https://doi.org/10.1016/j.dynatmoce.2017.05.003.
Solanki, S. K., N. A. Krivova, and J. D. Haigh, 2013: Solar irradiance variability and climate. Annu. Rev. Astron. Astrophys., 51, 311–351, https://doi.org/10.1146/annurev-astro-082812-141007.
Stroeve, J., and D. Notz, 2015: Insights on past and future sea-ice evolution from combining observations and models. Global Planet. Change, 135, 119–132, https://doi.org/10.1016/j.gloplacha.2015.10.011.
Swingedouw, D., L. Terray, C. Cassou, A. Voldoire, D. Salas-Mélia, and J. Servonnat, 2011: Natural forcing of climate during the last millennium: Fingerprint of solar variability. Climate Dyn., 36, 1349–1364, https://doi.org/10.1007/s00382-010-0803-5.
Taylor, P. C., M. Cai, A. Hu, J. Meehl, W. Washington, and G. J. Zhang, 2013: A decomposition of feedback contributions to polar warming amplification. J. Climate, 26, 7023–7043, https://doi.org/10.1175/JCLI-D-12-00696.1.
Wang, Y.-M., J. Lean, and N. Sheeley, 2005: Modeling the sun’s magnetic field and irradiance since 1713. Astrophys. J., 625, 522–538, https://doi.org/10.1086/429689.
Warren, S. G., 2019: Optical properties of ice and snow. Philos. Trans. Roy. Soc., 377A, 20180161, https://doi.org/10.1098/rsta.2018.0161.
Warren, S. G., I. G. Rigor, N. Untersteiner, V. F. Radionov, N. N. Bryazgin, Y. I. Aleksandrov, and R. Colony, 1999: Snow depth on Arctic sea ice. J. Climate, 12, 1814–1829, https://doi.org/10.1175/1520-0442(1999)012<1814:SDOASI>2.0.CO;2.
Woollings, T., M. Lockwood, G. Masato, C. Bell, and L. Gray, 2010: Enhanced signature of solar variability in Eurasian winter climate. Geophys. Res. Lett., 37, L20805, https://doi.org/10.1029/2010GL044601.
Xia, Y., Y. Y. Hu, J. P. Liu, Y. Huang, F. Xie, and J. T. Lin, 2020: Stratospheric ozone-induced cloud radiative effects on Antarctic sea ice. Adv. Atmos. Sci., 37, 505–514, https://doi.org/10.1007/s00376-019-8251-6.
