Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 Results of comprehensive long-term simulations of surface all-sky and clear-sky ultraviolet (UV) radiation through 1960–2100 are presented. A new earth system model, MIROC-ESM-CHEM, is used for the simulation, which considers key processes that change the surface UV radiation: atmospheric dynamics and chemistry affecting ozone in the stratosphere and troposphere, aerosols and clouds in the troposphere, and changes in surface albedo with sea ice and snow cover. In contrast to previous assessments considering only the effect of long-term change in stratospheric ozone, the simulated long-term behavior of UV radiation in this study is strongly affected by other processes. In one of two simulations, all-sky UV radiation in the northern midlatitudes is projected to increase in the 21st century despite the expected recovery of the stratospheric ozone layer. Reductions in aerosols and clouds are expected to overcompensate for the effect of ozone recovery. The results are sensitive to the future socioeconomic scenario, describing GHG concentrations and emissions of aerosol and ozone precursors in the troposphere. The interannual variability of UV radiation associated with the 11 year solar cycle and local processes is also discussed.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The solar ultraviolet (UV) radiation reaching the Earth's surface is known to damage animals and plants living on the ground surface as well as those living in water near the surface [see Bais et al., 2007, and references therein]. Solar UV accounts for only a small percentage of the total solar irradiance, and ozone in the stratosphere effectively absorbs the UV-B (280–315 nm) part of the spectrum, which has the potential to damage DNA. Since the discovery of ozone depletion in the Antarctic lower stratosphere, the increase in the surface UV-B radiation induced by ozone depletion has received wide attention. After several pioneering works [e.g., Austin et al., 1994; Burrows et al., 1994], meteorological agencies around the world have developed operational monitoring and prediction systems for surface erythematic UV radiation [see Bais et al., 2007, Appendix 7b].
 Several intensive studies have simulated the long-term evolution of stratospheric ozone through time, typically 1960–2100, using a number of coupled Chemistry Climate Models (CCMs) including detailed chemistry in the stratosphere. Several papers on multi-CCM assessments have been published to date, in which all models simulate qualitatively similar long-term evolution of stratospheric ozone [e.g., Eyring et al., 2006, 2007, 2010a, 2010b; Austin et al., 2010; Butchart et al., 2010]. As observed, stratospheric column ozone in these CCMs decreased through 1960–2000 due to increasing anthropogenic emission of ozone depleting substances (ODSs): halocarbons including chlorine and bromine atoms, and their accumulation in the atmosphere. The ozone reduction due to ODSs occurred globally in the upper stratosphere through gas-phase reactions, which caused a quasi-homogeneous reduction of column ozone of a few Dobson Units (DU). More severe ozone depletion occurred in the Antarctic lower stratosphere, where polar stratospheric clouds (PSCs) develop in extremely low temperature regions in the winter polar vortex. On the surface of PSCs, reactive chlorine and bromine are rapidly extracted from reservoir species such as HCl and ClONO2, accelerating catalytic destruction of ozone.
 The anthropogenic emission of ODSs has been well regulated since the Montreal Protocol in 1987, and ODSs in the stratosphere is expected to decrease through the 21st century [e.g., Daniel et al., 2007]. Based on this future scenario, CCMs have projected a recovery of stratospheric ozone to the 1980 level by around 2045–2055 in the extratropics and 2060 in the Antarctic [e.g., Austin et al., 2010; Eyring et al., 2010a]. On the other hand, increasing concentrations of greenhouse gases (GHGs) are likely to change ozone transport in the lower stratosphere, causing column ozone to decrease in the tropics and increase in the extratropics [e.g., Eyring et al., 2010a; Butchart et al., 2010]. The dynamical effect will depend on the future scenario of GHG concentrations [Eyring et al., 2010b].
Tourpali et al.  estimated changes in radiative transmission of UV using time evolving ozone and temperature profiles obtained from multimodel CCM simulations, and Hegglin and Shepherd  estimated changes in erythematic UV based on the stratospheric column ozone simulated by a CCM. These previous studies mainly focused on long-term changes in surface UV radiation in response to stratospheric ozone changes. As these papers mention, however, future changes in surface UV radiation would depend not only on: (1) the stratospheric ozone, but also on (2) tropospheric ozone, (3) absorption and scattering by aerosols in the troposphere, (4) cloudiness and optical properties of clouds, and (5) surface albedo that may change with sea ice and snow cover as well as surface vegetation. A decrease in surface reflection of UV causes a reduction of downward diffusive fluxes of UV by air molecules, aerosols and clouds. In this respect, a long-term future projection of surface UV radiation requires a new comprehensive Earth System Model (ESM) that includes the abovementioned key processes, and self-consistently simulates time evolution of the Earth's climate.
 The present paper describes results for direct simulations of surface UV-B through 1960–2100 performed with MIROC-ESM-CHEM [Watanabe et al., 2011]. The simulations are based on two future socioeconomic scenarios for the RCPs (Representative Concentration Pathways), which specify different evolution of GHG concentrations, emissions of tropospheric ozone precursors, emissions of aerosol precursors, and land use [Moss et al., 2010; Lamarque et al., 2010]. The main focus of this paper is on the long-term behavior of surface UV-B and its association with tropospheric and stratospheric ozone, aerosols, clouds, and surface albedo. In addition, the magnitude of interannual variability of surface UV-B radiation associated with the 11 year solar cycle and local processes will be discussed. Detailed investigations for the amount and optical properties of aerosols and clouds are beyond the scope of the present study, and qualitative effects of their increase/decrease on surface UV-B radiation are considered.
2. Model and Experiment
 MIROC-ESM-CHEM is based on a global climate model MIROC consisting of coupled atmosphere, ocean, sea ice, river, and land surface models [Watanabe et al., 2011]. The MIROC atmosphere model includes an online aerosol model (SPRINTARS), and considers the direct and indirect effects of aerosols [Takemura et al., 2005]. MIROC-ESM-CHEM further includes an atmospheric chemistry component (CHASER), an ocean ecosystem component, and a terrestrial ecosystem component addressing dynamic vegetation [Watanabe et al., 2011]. The CHASER has been developed as a tropospheric CCM, and includes major species and reactions for the simulation of tropospheric chemistry [Sudo et al., 2002a, 2002b]. The system of tropospheric chemistry (Ox-HOx-NOx-CO-VOCs: Volatile Organic Compounds) is similar to the original one except that the present version considers oxidation of methane. In addition, the CHASER component in MIROC-ESM-CHEM predicts major chlorine and bromine compounds (Cly and Bry) that are important for the simulation of stratospheric ozone. Parameterizations for liquid and solid particles in the stratosphere are included to calculate heterogeneous reactions on liquid sulfate aerosols and PSCs [e.g., Carslaw et al., 1995; Hanson and Mauersberger, 1988]. Surface area densities of PSCs are diagnosed as a function of the predicted temperature and mixing ratio of H2O and HNO3.
 The atmosphere model of MIROC-ESM-CHEM has a T42 horizontal resolution (corresponding to a horizontal grid spacing of about 300 km), and contains 80 vertical layers from the surface to a height of about 85 km. The model employs orographic and nonorographic gravity wave parameterizations, and self-consistently simulates the seasonal progression of the general circulation in the troposphere, stratosphere, and mesosphere, as well as the quasi-biennial oscillation in the equatorial lower stratosphere [e.g., Watanabe et al., 2008; Watanabe, 2008; Watanabe et al., 2011].
 Following a historical simulation through 1850–2005, future projection simulations through 2006–2100 were performed. The simulation setup was based on the CMIP5 (Coupled Model Intercomparison Project phase-5) web site (http://cmip-pcmdi.llnl.gov/cmip5/). Here we present results based on two future socioeconomic scenarios of RCPs [Moss et al., 2010; Lamarque et al., 2010]. RCP4.5 is a scenario in which: (1) anthropogenic increases of GHG concentrations decline with time, (2) anthropogenic emissions of aerosol precursors, e.g., SO2 from fossil fuel burning and organic/black carbon from biomass burning, decrease with time, and (3) anthropogenic emissions of ozone precursors in the troposphere, e.g., NOx from fossil fuel burning and CO from biomass burning, decrease with time. RCP8.5 is regarded as a “business as usual” scenario, in which the abovementioned climate changing agents are not strongly regulated. In each simulation, common time evolving concentrations of major ODSs are prescribed at the surface based on the WMO (World Meteorological Organization) A1 scenario [e.g., Daniel et al., 2007]. Kawase et al.  used a troposphere version of the CHASER to calculate future changes in tropospheric ozone based on RCPs, and more detailed descriptions for the GHG concentrations and emissions used in this study are given in their paper.
 The reconstructed historical solar spectral irradiance data was provided by SOLARIS (Solar Influence for Stratospheric Processes and their Role in Climate (SPARC)) (http://www.geo.fu-berlin.de/en/met/ag/strat/forschung/SOLARIS/Input_data/CMIP5_solar_irradiance.html), which was binned into spectral bands for the radiative transfer code (mstrnX) of MIROC-ESM-CHEM [Sekiguchi and Nakajima, 2008]. Following a recommendation on the SOLARIS web site, the last solar cycle (cycle 23 spanned 12.2 years from 1996.4 to 2008.6) was recursively used through 2009–2100. The UV-B radiation considered in this study is a combination of the three spectral bands of mstrnX: 278–290 nm, 290–303 nm, and 303–317 nm. The monthly averaged daily integral of downward (direct and diffusive) UV-B fluxes at the surface are output for all-sky and clear-sky conditions, respectively. For simplicity, we refer them to all-sky/clear-sky UV-B radiation hereafter. Although the seasonal variation of UV-B radiation due to changes in the solar zenith angle is predominant in the monthly time series at each location, an annual or seasonal average is considered in this study in order to focus on long-term changes.
3.1. Long-Term Tendency
Figures 1 and 2 compare the long-term evolution of zonal mean column ozone and all-sky UV-B radiation, which are averaged over several latitudinal bands. In order to focus on long-term tendencies, interannual variability is eliminated by recursively applying a 1:2:1 smoothing to each time series. Following the previous multiCCM study [Eyring et al., 2010b], column ozone anomalies from the 1980 baseline are shown in DU. In addition to total column ozone, dotted curves in Figure 1 show the stratospheric column ozone above 200 hPa, which can be compared to earlier multiCCM simulations [e.g., Eyring et al., 2010b]. Three major points are noted here: (1) simulated long-term evolution of the stratospheric column ozone generally agrees with the published results of multiCCM simulations, which is mainly caused by the increase and subsequent decline of ODSs, and by cooling of the upper stratosphere and changes in the stratospheric Brewer-Dobson circulation associated with increases in GHG concentrations, (2) the simulated long-term evolution of all-sky UV-B radiation does not always follow column ozone changes, implying the importance of changes in aerosols, clouds, and surface albedo, and (3) the long-term evolution of all-sky UV-B radiation strongly depends on the RCP scenario.
 In the tropics, a slight reduction of column ozone is simulated after 1980, which is mainly due to global destruction of ozone in the upper stratosphere. The column ozone in RCP4.5 recovers to the 1980 baseline near 2040, and then decreases to −5 DU by 2080. The column ozone in RCP8.5 shows an overrecovery in the middle of the 21st century. An increase in column ozone in 2000–2040 is mainly attributable to the reduction of ODSs. An additional increase in the total column ozone in 2020–2040 of RCP8.5 is caused by an increase in tropospheric ozone concentration [Kawase et al., 2011]. The decrease in column ozone after 2040 is attributable to a reduction of the ozone concentration in the lower stratosphere, which is dynamically caused by strengthened upward advection of low-O3 concentration air due to acceleration of the Brewer-Dobson circulation [e.g., Eyring et al., 2010a, 2010b; Butchart et al., 2010]. All-sky UV-B radiation in the tropics generally follows the changes in total column ozone, showing a clear anticorrelation. It increases after 2040 to reach more than +2% of the 1980 baseline after 2056 (2085) in RCP4.5 (RCP8.5). Climatological UV-B radiation is most abundant in the tropics due to the low solar zenith angle, and it is projected to increase toward an unprecedented level in the latter half of the 21st century.
 In the Northern Hemisphere (NH) midlatitudes, ozone recovery to the 1980 baseline is expected to occur by 2020 in RCP8.5 and by 2030 in RCP4.5. In RCP8.5, column ozone is projected to increase further to reach +28 DU in 2100. The overrecovery in column ozone in RCP8.5 is caused by a combination of increases in the ozone concentration in the troposphere and the lower stratosphere. The former is due to an increase in CH4 concentration, a major ozone precursor in the troposphere, and enhanced downward transport of stratospheric ozone [Kawase et al., 2011]. The latter is associated with the acceleration of the Brewer-Dobson circulation [e.g., Eyring et al., 2010a, 2010b; Butchart et al., 2010]. The column ozone in the RCP4.5 simulation holds close to the 2040 value through 2040–2100. Lower GHG concentrations compared to RCP8.5 weaken the acceleration of the Brewer-Dobson circulation, and an increase in stratospheric column ozone is compensated by a decrease in tropospheric column ozone due to fewer ozone precursors than the present level [Kawase et al., 2011].
 The long-term evolutions of all-sky UV-B radiation in the NH midlatitudes simulated by the two RCP scenarios are also qualitatively different from each other. In RCP8.5, UV-B radiation decreases after 2000 following increasing total column ozone. In contrast, UV-B radiation in RCP4.5 is projected to stay at the 2000 level until 2040, and increase afterward to reach +6% in 2100. This increase in UV-B radiation is unexpected from the nearly constant total column ozone in this period, implying the importance of reductions in aerosols and clouds (discussed in section 3.2). In summary, future all-sky UV-B radiation in the NH midlatitudes depends most strongly on the choice of future socioeconomic scenario.
 In the Southern Hemisphere (SH) midlatitudes, a decrease (an increase) in column ozone (all-sky UV-B) of about −15 DU (+7%) is simulated through 1980–2000. The column ozone and UV-B radiation are projected to return to their 1980 baseline values around 2030 (2020) in RCP4.5 (RCP8.5), continuing overrecovery thereafter. The column ozone in 2100 is projected to reach +24 DU and +13 DU in RCP8.5 and RCP4.5, respectively. The projected long-term evolution of stratospheric column ozone in the SH midlatitudes is less sensitive to the choice of future RCP scenario than the NH midlatitudes, which qualitatively agrees with results reported by Eyring et al. [2010b]. Such an interhemispheric difference in column ozone response to RCP scenarios (GHG concentrations) is likely attributable to interhemispheric differences in ozone transport due to the Brewer-Dobson circulation, and needs further investigations. The all-sky UV-B radiation in 2100 is −8% and −5% in the RCP8.5 and RCP4.5 simulations, respectively. In contrast to the NH midlatitudes, the long-term behavior of all-sky UV-B radiation qualitatively follows the changes in total column ozone. Note that these time series show averages for a broad latitudinal band (30–60°S), and that UV-B radiation changes in the SH midlatitudes have strong latitudinal gradient associated with the Antarctic ozone hole (Figure 3).
 In polar regions, averages in warm seasons are presented in order to avoid polar night. In the Arctic for both RCP scenarios, all-sky UV-B radiation is projected to rapidly decrease from 0% in 2000 to −10% in 2030 in spite of a relatively small ozone increase. This decrease in all-sky UV-B radiation is mostly attributable to an increase in cloudiness and a reduction of surface albedo, resulting from the disappearance of sea ice associated with climate change (Figure 3). The column ozone (UV-B radiation) in RCP8.5 continues to increase (decrease) after 2050, while both remain constant in RCP4.5. The most pronounced ozone reduction associated with the Antarctic ozone hole recovers to the 1980 baseline around 2045 in both RCP simulations. The column ozone (UV-B radiation) continues to increase (decrease) after this time, which occurs more strongly in RCP8.5 associated with higher GHG concentrations.
3.2. Global Maps
Figure 3 shows relative changes in the average column ozone, clear-sky UV-B radiation, and all-sky UV-B radiation between the 2090s and 2000s. Averages in the warm seasons in both hemispheres are shown separately, because UV-B radiation in the cold season is much less than in the warm season. The simulated changes in clear-sky UV-B radiation can be explained by changes in column ozone, aerosols, and surface albedo. The projected increase in column ozone in the extratropics of both hemispheres decreases the amount of clear-sky UV-B reaching the surface. The recovery of the Antarctic ozone hole leads to a substantial decrease in clear-sky UV-B radiation over the Antarctic and the southern tip of South America. A significant reduction of clear-sky UV-B radiation is projected in the Arctic, where decreased surface albedo is expected due to decreases in sea ice and snow cover associated with climate change (Figure 4). An increase in reflection of UV-B due to increasing sea-salt aerosols in the boundary layer is also expected in the Arctic related to the disappearance of sea ice (Figure 5a). On the other hand, an increase in clear-sky UV-B radiation is projected in the tropics, where a decrease in column ozone is expected.
 The degree of contrast between the decreasing clear-sky UV-B radiation in the extratropics and increasing clear-sky UV-B radiation in the tropics differs between the two RCP simulations, especially in the NH. In the NH extratropics, a greater increase of column ozone is projected in RCP8.5 than in RCP4.5, which leads to a greater decrease in clear-sky UV-B radiation. The column ozone increases (decreases) north (south) of 40°N in RCP4.5, whereas this boundary is located near 20°N in RCP8.5. This difference is mainly associated with changes in ozone transport in the lower stratosphere. In RCP8.5, downward advection of high-ozone concentration air is enhanced in the subtropical lower stratosphere due to strengthening of the Brewer-Dobson circulation (not shown). In addition, a greater reduction in aerosols is expected in the NH midlatitudes in RCP4.5 than in RCP8.5, which increases clear-sky UV-B radiation reaching the surface (discussed later). Overall, clear-sky UV-B radiation in the NH midlatitudes increases in RCP4.5, and decreases in RCP8.5.
 Changes in clouds are also scenario-dependent, and introduce additional inhomogeneity to the global distribution of UV-B radiation changes. The model projects a general reduction of cloudiness in extrapolar regions associated with climate change, and the effects can be seen as the difference between all-sky and clear-sky UV-B radiation. Downstream of industrial regions and regions of active biomass burning, i.e., from East Asia to the North Pacific, from North America to the North Atlantic, Europe, India, Southeast Asia, and the Amazon basin, relatively large reductions in clouds are projected, increasing all-sky UV-B radiation by more than +10%. In the equatorial Pacific and South Indian Ocean, the decrease in clouds is greater in RCP8.5 than in RCP4.5, causing a greater increase in all-sky UV-B radiation in RCP8.5.
 At the end of this subsection, the projected changes in aerosols are briefly described. Figure 5 illustrates future changes in the aerosol mass column loading in RCP4.5 between the 2090s and 2000s. Here we can only partly and qualitatively attribute the projected changes in clear-sky UV-B (Figure 3a, middle) to those for the aerosol mass column loading, because clear-sky UV-B is also affected by changes in column ozone, and actual contaminating effects of aerosols on UV-B transmission depend on complicated optical properties and vertical distribution of aerosols. In the Arctic, the projected increase in the sea-salt aerosols due to the reduction of sea ice would decrease UV-B reaching the surface. Around industrial regions in the NH midlatitudes, the projected reductions of the sulfate and black carbon (BC) aerosols associated with presumed regulations of fossil fuel burning would increase UV-B. The projected reductions of BC and organic carbon (OC) aerosols over the tropical rain forest associated with presumed regulations of anthropogenic deforestation would increase UV-B. Among those regions, the reductions of aerosols near East Asia, the maritime continent, and the Amazon basin would have prominent effects on the projected increase in clear-sky UV-B.
3.3. Interannual Variability
 Long-term tendencies in all-sky and clear-sky UV-B radiation have been described in the previous subsections, where the roles of long-term changes in column ozone, aerosols, clouds, and surface albedo are highlighted. Here we focus on interannual-to-decadal variability of UV-B radiation, which has a magnitude comparable to the long-term variations.
3.3.1. The 11 Year Solar Cycle
Figure 6 shows time series for the annually averaged global mean total column ozone and clear-sky UV-B radiation in RCP4.5, along with the intensity of solar UV-B and UV-C (178–278 nm is considered in MIROC-ESM-CHEM) incident at the top of atmosphere (TOA). The relative change in each quantity is compared to the 1980 baseline (1975–1985 average). Embedded in the long-term changes, clear-sky UV-B radiation shows a decadal-scale variation associated with the 11 year solar cycle, which had minima in 1964, 1976, 1986, 1996, 2008, and every 12.2 years thereafter. Although TOA incidence of UV-B varies only 0.4% between the solar minimum and solar maximum, surface UV-B radiation changes by as much as 2%. Furthermore, surface UV-B radiation anticorrelates with the solar cycle in the TOA UV-B. These relationships can be explained through photochemistry of the stratospheric ozone [e.g., Brasseur and Solomon, 1986]. The TOA incidence of UV-C varies by about 2% between the solar minimum and solar maximum. At the solar maximum, an increase in UV-C intensity in the stratosphere increases photodissociation of molecular oxygen to produce more atomic oxygen, leading to an increase in production of stratospheric ozone. Thus, the increased stratospheric ozone concentration at the solar maximum causes a reduction of UV-B radiation reaching the surface. Therefore, the surface clear-sky UV-B radiation anticorrelates with the solar cycle in the TOA UV-C (r = −0.3) and column ozone (r = −0.54). These relationships also hold for all-sky UV-B and the TOA UV-C (r = −0.3), and column ozone (r = −0.4). Shorter-term variability in the global mean clear-sky UV-B radiation is likely associated with variations in column ozone, aerosols, and surface albedo, which cannot be investigated based on global averages.
3.3.2. Local Variability
 Although the 11 year solar cycle has a strong impact on the globally averaged UV-B radiation, interannual variability of UV-B radiation at a certain location is likely more strongly affected by various other processes, e.g., dynamics and chemistry of the stratosphere affecting column ozone, aerosols and clouds in the troposphere, and surface albedo. In particular, interannual variations in clouds likely enhance the interannual variability of all-sky UV-B radiation compared to that for clear-sky UV-B.
Figure 7 shows the time evolution of all-sky and clear-sky UV-B radiation at several grid points near large cities and representative observation sites. Note that the present model has a coarse horizontal grid spacing of about 300 km, so the results should be considered as a regional average surrounding each location. Averages in the warm season, April–September in the NH and October–March in the SH are considered except for Mumbai and Singapore, where annual averages are considered. The relative change is shown for each quantity compared to the 2000 baseline (1995–2005 average). In low and midlatitudes, the interannual variability of clear-sky UV-B radiation is about 2%–4% in terms of the difference between the minimum and maximum. In higher latitudes (Oslo and Ushuaia), the interannual variability of clear-sky UV-B radiation is enhanced by a larger dynamical variability in the stratosphere and the tropopause region that affects column ozone. The range of variation is about 6%–8% at these locations. Syowa in Antarctica is located near the edge of the Antarctic ozone hole, where dynamical variability of the stratospheric polar vortex strongly affects column ozone. A range in variation of more than 20% is seen here through 1990–2030, decreasing afterward in association with ozone recovery.
 The interannual variability of all-sky UV-B radiation always exceeds that for clear-sky UV-B radiation, and has a range of about 10%. An exception is Syowa, where clouds are optically too thin to make a difference between all-sky and clear-sky UV-B radiation. In general, all-sky UV-B radiation in RCP4.5 is larger than in RCP8.5. In practical applications, the maximum envelope of all-sky UV-B time series would be useful to estimate the maximal potential exposure of people and agricultural products to UV-B. Oslo in RCP4.5 would receive similar levels of all-sky and clear-sky UV-B through 2000–2100, though all-sky UV-B radiation would frequently exceed +10% after 2040. Rome and New York in RCP4.5 would experience a gradual increase of clear-sky UV-B radiation. Rome would have a larger increase in all-sky UV-B radiation by +15%–20% after 2040. This difference in all-sky UV-B radiation results from variations in clouds projected at these locations.
 Shanghai in RCP4.5 is projected to have the potential to receive 15%–20% higher all-sky UV-B radiation than present after 2050. This increase in all-sky UV-B radiation accompanies an increase in clear-sky UV-B radiation, which would exceed +5% after 2050 and reach +10% near 2070. Industrial emissions of aerosol precursors in China have peaks near 2020 in both RCP scenarios, and a decline of emissions would return clear-sky UV-B radiation to a 1960 level by 2060. Mumbai in RCP4.5 would exceed the present level of all-sky and clear-sky UV-B radiation after 2050, and all-sky UV-B radiation would exceed +10% after 2080. Clear-sky UV-B radiation in Singapore of RCP4.5 would gradually increase from the present level to about +8% by 2100, and all-sky UV-B radiation would sometimes reach +15% after 2070. In Sao Paulo and Brisbane in RCP4.5, all-sky and clear-sky UV-B radiation would not change dramatically from the level in the recent past. Wellington and Ushuaia in RCP4.5 would experience gradual decreases of all-sky and clear-sky UV-B radiation through 2000–2060, while a reduction in all-sky UV-B radiation would continue afterward in Ushuaia.
4. Concluding Remarks
 The long-term behavior of all-sky and clear-sky UV-B radiation through 1960–2100 was simulated using MIROC-ESM-CHEM. The present study is an extension of previous studies considering only the effect of stratospheric ozone [Tourpali et al., 2009; Hegglin and Shepherd, 2009], and considers for the first time effects of tropospheric ozone, aerosols, clouds, and surface albedo on the long-term changes in surface UV-B radiation.
 The simulated long-term changes in the stratospheric column ozone generally agree with those reported in the previous multiCCM simulations [e.g., Austin et al., 2010; Eyring et al., 2010a, 2010b; Butchart et al., 2010]. The ozone recovery to the 1980 baseline was expected to occur in the first half of the 21st century due to the reduction of ODSs. After the ozone recovery, the acceleration of the Brewer-Dobson circulation associated with increasing GHG concentrations would cause overrecovery of stratospheric column ozone in the extratropics and a reduction in the tropics. The dynamically caused ozone changes depended on the GHG concentrations as was pointed out by Eyring et al. [2010b]. In addition, the changes in tropospheric ozone considered in the present study added further long-term variations to the total column ozone, depending on the future RCP scenarios.
 The simulated long-term changes in clear-sky UV-B radiation were not only affected by column ozone but also by changes in aerosols and surface albedo. The simulated reduction of sea ice in the Arctic decreased the surface albedo and enhanced emissions of sea-salt aerosols from the open ocean, decreasing clear-sky UV-B radiation in the Arctic. The future clear-sky UV-B radiation in the NH midlatitudes would be most sensitive to the choice of future socioeconomic scenario of all latitude bands. In the RCP4.5 simulation, the reduction of aerosols allowed greater transmission of UV-B than the present level, whereas in the RCP8.5 simulation, the overrecovery of total column ozone dominated other effects and decreased clear-sky UV-B radiation. The clear-sky UV-B radiation in the tropics generally followed the evolution of column ozone, and reached an unprecedented level in the latter half of the 21st century. A reduction of clear-sky UV-B radiation compared to the 1980 level were projected in the SH midlatitudes and the Antarctic, following the recovery and overrecovery of total column ozone.
 The changes in clouds added additional inhomogeneity to all-sky UV-B radiation changes. MIROC-ESM-CHEM projected a general decrease (increase) in clouds in the extrapolar (polar) regions, increasing (decreasing) all-sky UV-B radiation. All-sky UV-B radiation information is required in practical applications estimating future potential impacts of increasing UV-B radiation on animals and plants. However, the response of clouds to the increasing GHG concentrations simulated by climate models is known to contain large uncertainty [e.g., Intergovernmental Panel on Climate Change, 2007]. MIROC-ESM-CHEM has shown a tendency for clouds (cloud radiative forcing on short-wave radiation) in the extrapolar regions to decrease with increasing GHG concentrations, while other climate models have shown the opposite response [Webb et al., 2006; Yokohata et al., 2010]. In this respect, the increasing trend of all-sky UV-B radiation in the tropics and NH midlatitudes projected in this study could be regarded as an upper limit.
 In addition to the long-term tendencies, the interannual variability of all-sky and clear-sky UV-B radiation was investigated. In the global average, clear-sky UV-B radiation showed a clear anticorrelation to the 11 year solar cycle, which was explained by the photochemistry of the stratospheric ozone. In the investigations of local variability, the interannual variations in clouds caused a 10% change in the range of all-sky UV-B radiation, which was much larger than the change in range of clear-sky UV-B radiation of about 2%–4% in the extrapolar regions. Focusing on the maximal envelope of all-sky UV-B radiation time series, several locations in the NH midlatitudes would receive +15%–20% higher UV-B in the latter half of the 21st century compared to the present level.
 As the future change in local clouds is one of the most uncertain processes in this class of climate model, practical applications should rely on multiESM ensemble assessments, which could be available in the next few years. Another issue to be addressed in the future works would be more detailed spectral changes in the future UV radiation. The present study is based on the broadband radiative transfer scheme in MIROC-ESM-CHEM and focuses on the integrated UV-B (290–317 nm), whereas practical applications may require more detailed information of spectral changes in the UV or erythematic UV, which is equivalent to UV index. Spectral dependence of UV responses to future potential changes in column ozone, aerosols, clouds, and surface albedo might be an interesting study. The present paper has presented the first step toward comprehensive long-term simulations of all-sky and clear-sky UV-B radiation, and uncovered several new features not reported in previous studies.
 The authors are grateful to anonymous two reviewers for useful comments and suggestions. Discussions with Hideharu Akiyoshi were helpful to improve the model and original manuscript. The authors thank Team-MIROC for their support and encouragement throughout the project. This study was supported by the Innovative Program of Climate Change Projection for the 21st Century, MEXT, Japan. The numerical simulations in this study were performed using the Earth Simulator, and all figures were drawn using a combination of GTOOL and the GFD-DENNOU Library.