Current patterns of Earth's weather and climate are largely determined by the spatiotemporal distributions of energy exchanges between the surface, atmosphere, and space. Theoretically, the outgoing longwave radiation (OLR) is balanced by the incoming and reflected solar radiation at the top of the atmosphere (TOA) to produce an equilibrium climate. The incoming solar radiation can be scattered by clouds and aerosols or absorbed by the intermediary atmosphere, thereby contributing to the diverse energy transformation at the surface (Trenberth et al., 2009; Wild et al., 2013a). The Earth's surface energy balance is of particular significance because it is the key driver of atmospheric and oceanic circulations, hydrological cycles, and various surface processes (Wild et al., 2008; Mercado et al., 2009; Wild et al., 2013a; L'Ecuyer et al., 2015). Anthropogenic influences on climate change are driven by the uneven distribution of the TOA net radiation caused by forcings perturbed by variations of the atmospheric composition of greenhouse gases and aerosols as well as aerosol–cloud interactions (Trenberth et al., 2009; Stephens et al., 2012; Wild et al., 2013a; Trenberth et al., 2014; L'Ecuyer et al., 2015; Wild et al., 2019; Z. Wang et al., 2021).

Many efforts have been made to quantify the magnitudes of different radiative components or energy budgets in the climate system over a range of time–space scales, such as on global scales (Lin et al., 2008; Trenberth et al., 2009; Stephens et al., 2012; Wild et al., 2013b, 2015; L'Ecuyer et al., 2015; Wild et al., 2019; Wild, 2020), over land and ocean domains or the energy transport between them (Fasullo and Trenberth, 2008a, b; Trenberth et al., 2009; Wild et al., 2015; L'Ecuyer et al., 2015), over the Arctic (Previdi et al., 2015; Christensen et al., 2016), and over individual continents and ocean basins (L'Ecuyer et al., 2015; Kim and Lee, 2018; Thomas et al., 2020). The energy balance at the TOA can be accurately monitored by satellites from the most advanced Clouds and the Earth's Radiant Energy System (CERES) Energy Balanced and Filled (EBAF) data product (Loeb et al., 2018), while considerably larger uncertainties appear at the surface fluxes owing to weaker observational constraints (Raschke et al., 2016; Kato et al., 2018; Huang et al., 2019). These assessments mostly build upon complementary approaches from a combination of space and surface observations, climate models, and reanalyses. To date, the discrepancies of independent global mean surface radiative fluxes have been estimated to be within a few W m-2 (Wild, 2017a, b), enabling the accurate quantification of global surface budgets. Besides, the surface radiative components simulated by various climate models vary substantially in a range of around 10–20 W m-2 on global scales but exhibit greater inter-model discrepancies on regional scales (Li et al., 2013; Wild et al., 2013a; Boeke and Taylor, 2016; Wild et al., 2015; Wild, 2017a, b, 2020). Existing challenges in the surface energy estimates include considerable uncertainties from surface albedo and skin temperature as well as the partitioning of surface net radiation into sensible and latent heat (SH; LH) (Wild, 2017a, b).

Due to the large population and the largest emission source of aerosols and their precursors, East Asia, especially China, has long been a hotpot in climate change research. Aerosols can interact with radiation directly by scattering and absorbing solar/thermal radiation (Ghan et al., 2012) and indirectly by modifying cloud microphysical properties and lifetimes (Li et al., 2011), thereby influencing Earth's radiation balance. As the world's largest and highest plateau, the Tibet Plateau (TP) covers nearly one-fifth of the East Asian land area, significantly affecting the atmospheric circulation, energy budget, and water cycles of the climate system through its orographic and thermal effects (Liu et al., 2007; Xu et al., 2008a, b; Wu et al., 2015). Deeper insights into the energy budget differences over East Asian and global land against the background of high aerosol emissions and the TP role in East Asia are meaningful and essential attempts. Moreover, clouds play a key role in modulating global and regional energy budgets and hydrological cycles by increasing the reflected solar radiation and also the downward thermal radiation, leading to a cooling and warming of the climate system (Stephens, 2005; Wild et al., 2013a; Li and Mao, 2015; H. Wang et al., 2021). Therefore, our emphasis in this study is on the regional characterization of the East Asian energy balance under both all-sky and clear-sky conditions based on a combination of surface observations, satellite-derived products, reanalysis, and Coupled Model Intercomparison Project Phase 6 (CMIP6) models. The cloud influence on the radiative energy budgets at the TOA, within the atmosphere, and at the surface is further quantified over this region. Section 2 introduces the different data sources used in this study, including surface and satellite observations, climate models, and reanalyses. Sections 3 and 4 provide detailed analyses of the all-sky and clear-sky estimates of the energy balance components. The inferred cloud radiative effects (CREs) at the TOA, within the atmosphere, and at the surface are presented in Sect. 5. A summary and conclusions are given in Sect. 6. The present day in this study represents the years of 2010–2014, which correspond to the last 5 years of the historical simulations in CMIP6 climate models. East Asian land as considered in this study consists of five countries, i.e., China, Japan, South and North Korea, as well as Mongolia.

The clear-sky land energy balance budgets over East Asia are similarly evaluated as all-sky conditions. Detailed analyses are given in the Supplement if the reader is interested. The annual land mean SW clear-sky absorptions at the TOA and the surface over East Asia show larger variations among different models than those under all-sky conditions (Fig. 1a and b; Table 1), which is consistent with those reported by Wild et al. (2019) but is amazingly in contrast to the recognition that the representation of clouds is the largest uncertainty in climate models (Dolinar et al., 2015). In particular, the surface SW clear-sky absorptions simulated by various models still exhibit a larger uncertainty than the TOA counterparts despite the lower absolute values (Fig. 1b; Table 1). In contrast to the all-sky counterparts, the simulated clear-sky SSR among different models shows a notably smaller inter-model spread and SD than the surface SW absorptions (Table 1), with a much smaller model discrepancy compared to the all-sky conditions (Fig. 2a; Table 1).

To further constrain the outlined inter-model discrepancy of the simulated clear-sky SSR, surface observations from the CMA and CERES-interpolated estimates at the GEBA sites are utilized in this study. The high values of the station-based clear-sky SSR are mainly located in the TP but with an abnormally high value located in southern China (Fig. 3b). All the East Asian land mean clear-sky SSR estimates from CERES, the CMIP6 multi-model mean, and ERA5 agree reasonably well with the surface observations but with smaller correlation coefficients ranging from 0.78 to 0.82 compared to the all-sky conditions (Fig. 7b, d, and f). The CERES-derived clear-sky SSR is mainly overestimated in central and western China but with slight underestimations mainly located in northeastern, eastern, and southern China (Fig. 7a). Similar bias patterns can also be found in the clear-sky SSR from the CMIP6 multi-model mean and ERA5 compared to the surface observations, except for some individual sites over northeastern Inner Mongolia, eastern China, western Mongolia, and Japan (Fig. 7c and e) but with relatively smaller overestimations than the all-sky counterparts (Fig. 4c and e; Table 2). Specifically, the smallest station mean bias in the CERES-derived SSR compared to the multi-model mean and ERA5 (Table 2) can be attributed to its evenly distributed surface sites of overestimations and underestimations (Fig. 7b, d, f). Again, among all the aforementioned clear-sky SSR biases, more overestimations exist in urban stations than rural stations (Figs. 4b, d, f and 7b, d, f; Table 2). Consequently, all East Asian land mean area-weighted averages of clear-sky SSR from CERES, the CMIP6 multi-model mean, and ERA5 show higher overestimations of around 6, 12, and 8 W m-2, respectively, compared to the surface observed counterpart of 230 W m-2 (Table 3). Based on the similar method introduced in Wild et al. (2015), the best estimate for the East Asian land mean clear-sky SSR is determined to be 234 ± 1.1 W m-2 (2σ uncertainty), with a slightly smaller correlation coefficient of 0.94 and smaller deviations from the CERES and ERA5 estimates compared to the all-sky counterparts (Fig. 5b; Table 3). Besides, the overestimations still exist in the observed land mean clear-sky SSR for most climate models over East Asia, with a smaller multi-model mean overestimation of 9.1 W m-2 than the all-sky counterparts.

This clear-sky energy budget only represents the removal of cloud but maintains the same atmospheric conditions as the all-sky conditions, which is not balanced because it is not the equilibrium state the Earth would achieve when no clouds could form. Ultimately, the clear-sky East Asian land mean energy budget is not closed and has no quantifications of SH and LH as displayed in Fig. 6b. In addition to the analyses above, the clear-sky TOA energy budgets are derived from the CERES-derived product, with uncertainty ranges referring to Loeb et al. (2018), while the surface LW budgets are again from ERA5 reanalysis. Also, an additional clear-sky radiation-weighted surface albedo of 0.19 from CERES is obtained to estimate the surface-reflected and absorbed SW radiation. Apart from the TOA budget, all the remaining uncertainty ranges are given by different data sources from various CMIP6 models as well as the multi-model mean, CERES-derived, and ERA5-derived estimates.

We double-check the energy balance components evaluated in this study by referring to the uncertainty ranges from the CERES-derived product given by Kato et al. (2018) (Table 5), which indicates that all estimated energy components fall within these uncertainty ranges, except for the all-sky surface downward LW radiation, with about 3 W m-2 lower than the corresponding lowest CERES range. This is in line with its much higher CERES-derived estimate compared to that of ERA5 (285 vs. 273 W m-2) (Table 1).

Overall, around 21.6 % and 56.9 % of the TOA incoming solar radiation are absorbed by the atmosphere and surface, respectively, for clear-sky conditions, while these absorptions are 23.1 % and 41.6 % for all-sky conditions. This implies that the existence of clouds results in more atmospheric SW absorption of around 1.5 % and much less surface solar absorption of around 15.3 % with respect to the TOA incoming solar radiation.

According to the annual land mean best estimates of radiative components over East Asia under all-sky and clear-sky conditions obtained in the previous sections, the present-day CREs can be inferred quantitatively over this region. The calculated SW, LW, and net CREs at the TOA, within the atmosphere, and at the surface are therefore presented in Fig. 8. Moreover, the corresponding calculation formulas are also given as follows. 1TOASWCRE=TOAoutgoingSWall-sky-TOAoutgoingSWclear-sky2TOALWCRE=TOAoutgoingLWall-sky-TOAoutgoingLWclear-sky3TOAnetCRE=TOASWCRE+TOALWCRE4SurfacenetSWCRE=surfacenetSWall-sky-surfacenetSWclear-sky5SurfacenetLWCRE=surfacenetLWall-sky-surfacenetLWclear-sky6SurfacenettotalCRE=surfacenetSWCRE+surfacenetLWCRE7AtmosphericSWCRE=TOASWCRE-surfacenetSWCRE8AtmosphericLWCRE=TOALWCRE-surfacenetLWCRE

The best estimates for the annual East Asian land mean reflected solar radiation at the TOA under all-sky and clear-sky conditions are -118 and -72 W m-2, respectively, differing by -46 W m-2, indicating that the clouds give rise to an extra 46 W m-2 solar reflection at the TOA, thus cooling the Earth–atmosphere system. Similarly, the TOA LW CRE, obtained as the difference between the TOA thermal radiation under all-sky and clear-sky conditions, is 24 W m-2, suggesting a warming effect of clouds on the system. Thus, the estimated TOA net CRE is -22 W m-2, pointing out that the overall effects of clouds result in an energy loss and net cooling to the system, not only in the global mean, but also over East Asian land.

At the Earth's surface, the shading effects of clouds are estimated to reduce the surface solar radiation by 60 W m-2, from 234 to 174 W m-2, while the surface solar absorption differs by 51 W m-2, from 190 to 139 W m-2; that is, the surface net SW CRE is -51 W m-2. In cloudy skies, the estimated surface downward LW radiation increases from 253 to 273 W m-2, with an increase of 20 W m-2, illustrating that the surface net LW CRE is 20 W m-2 and therefore leads to a surface warming. Thus, the surface net CRE, i.e., the sum of the surface net SW and LW CRE, is then -31 W m-2, indicating that clouds contribute more to the SW energy budgets. Eventually, the clouds lead to enhancements of the SW and LW absorptions within the atmosphere of around 5 and 4 W m-2, respectively, thus resulting in an atmospheric net CRE of 9 W m-2 over East Asian land.

The above CRE best estimates are compared with the corresponding estimates from different data sources (Fig. 9; Table 1). Generally, compared to the LW CREs (Fig. 9b), the simulated SW CREs show larger spreads and SDs amongst the models (Fig. 9a; Table 1). For the SW CREs at the TOA, within the atmosphere, and at the surface, the CERES-derived estimates match perfectly with the best estimates mentioned above, within 2 W m-2 of the biases, followed by the estimates from the multi-model means and ERA5 (Table 1). For the LW CREs, the calculated TOA LW CREs from the CMIP6 multi-model mean and CERES differ by no more than 1 W m-2 compared to the best estimate, while large differences are noted at the surface LW CREs, thereby leading to their opposite signs in the atmospheric LW CREs (Fig. 9b; Table 1). Specifically, since the ERA5-based TOA LW CRE deviates by no more than 3 W m-2 with the best estimate of 24 W m-2 with nearly the same surface LW CRE, the estimated atmospheric LW CRE is therefore the closest to the best estimate (Table 1). This is owing to the fact that we make use of the ERA5 data as the reference to estimate the surface LW radiation. Thus, the major reason for the large discrepancies in the atmospheric and surface LW CREs estimated from different data sources with respect to the best estimates in this study is the determination of the surface downward and upward LW radiation, which is also the reason for the large deviations in their net CREs (Fig. 9c).

A better comparison with the global annual mean best estimates of CREs by Wild et al. (2019) is given in Fig. S5. At the TOA, slightly lower and much lower East Asian land-mean SW and LW CREs of 1 and 4 W m-2 result in 3 W m-2 more energy loss at the TOA compared to the globe. At the surface, much lower annual East Asian land-mean SW and LW CREs of 3 and 8 W m-2 are estimated compared to the values over the globe, leading to a net CRE deviation of 5 W m-2 indicative of 5 W m-2 more energy loss at the surface. However, lower and higher annual East Asian land-mean SW and LW CREs of 2 and 4 W m-2 within the atmosphere contribute to the nearly close net CRE with a deviation of no more than 2 W m-2 compared to the global mean estimates. On the whole, lower annual East Asian land-mean best estimates in the absolute values of surface SW and LW CREs as well as the TOA LW CRE compared to their global mean counterparts give rise to the CRE differences between them.

This study aims to explore how the energy budgets are interrupted by the complex orographic and thermal effects of the TP as well as the high anthropogenic aerosol emissions over East Asian land compared to global land, based on complementary data sources from space and surface observations as well as the CMIP6 climate models and ERA5 reanalysis. A further quantitative investigation of CREs at the TOA, within the atmosphere, and at the surface is also conducted.

Comparisons between all-sky and clear-sky energy budgets indicate that the overall effects of clouds greatly reduce the surface solar absorption by about 15.3 % and enhance that within the atmosphere by 1.5 %. Compared to the global land energy budget estimates from Wild et al. (2015), for the SW budgets, a notably more atmospheric SW reflection of 5.2 % but with a slightly less atmospheric SW absorption of 0.6 % with respect to their respective TOA incident solar radiation is estimated over East Asian land, possibly indicating that the lower water vapor content effects due to the TP overcompensate for the aerosol and cloud effects over East Asian land. For the LW budgets, a substantially lower surface LW emission of around 25 W m-2 and a smaller relative surface downward LW radiation of around 3.6 % with respect to their respective surface emissions can be noticed over East Asian land compared to global land, which possibly result from the lower regional surface skin temperature as well as the weaker greenhouse effect and fewer low clouds mainly induced by the high altitude and thinner air over the TP, thus leading to a higher percentage of regional OLR of 2.7 %.

The CREs over East Asian land are inferred through the energy budget differences between all-sky and clear-sky conditions. The clouds reduce the solar absorption at the TOA by 46 W m-2 and enhance the TOA thermal radiation by 24 W m-2, respectively, leading to a TOA net CRE of -22 W m-2, a more cooling effect on the regional climate system than that over the globe (-19 W m-2). At the surface, the net CRE is estimated to be -31 W m-2 according to less solar absorption of 51 W m-2 and more downward thermal radiation of 20 W m-2, indicative of larger cloud impacts on SW radiation. Within the atmosphere, the estimated net CRE is 9 W m-2 due to the increases in solar absorption and net thermal radiation of 5 and 4 W m-2, respectively. Compared to the global mean best estimates of CREs as introduced by Wild et al. (2019), relatively lower East Asian land mean best estimates of surface SW and LW CREs as well as the TOA LW CRE contribute to the CRE differences between them.

On the whole, all the estimated land mean energy balance components over East Asia in this study fall within the uncertainty ranges of the CERES-derived assessments, except for the all-sky surface downward LW radiation. More accurate and reliable datasets should be utilized to reduce the substantial uncertainties in the regional energy balance estimates, particularly in the surface budgets, and more widespread temporal and spatial representations of energy budget research are recommended for more comprehensive comparisons in future. For example, newly published surface radiation products with high resolutions based on satellite datasets (e.g., Letu et al., 2022; Xu et al., 2022) are expected to make sense in improving the accuracy of the regional/global surface radiation budget studies.