Climate response to changes in atmospheric carbon dioxide and solar irradiance on the time scale of days to weeks

Recent studies show that fast climate response on time scales of less than a month can have important implications for long-term climate change. In this study, we investigate climate response on the time scale of days to weeks to a step-function quadrupling of atmospheric CO2 and contrast this with the response to a 4% increase in solar irradiance. Our simulations show that significant climate effects occur within days of a stepwise increase in both atmospheric CO2 content and solar irradiance. Over ocean, increased atmospheric CO2 warms the lower troposphere more than the surface, increasing atmospheric stability, moistening the boundary layer, and suppressing evaporation and precipitation. In contrast, over ocean, increased solar irradiance warms the lower troposphere to a much lesser extent, causing a much smaller change in evaporation and precipitation. Over land, both increased CO2 and increased solar irradiance cause rapid surface warming that tends to increase both evaporation and precipitation. However, the physiological effect of increased atmospheric CO2 on plant stomata reduces plant transpiration, drying the boundary layer and decreasing precipitation. This effect does not occur with increased solar irradiance. Therefore, differences in climatic effects from CO2 versus solar forcing are manifested within days after the forcing is imposed.


Introduction
A good understanding of the climate response to changes in external forcings, such as carbon dioxide and aerosols, Content from this work may be used under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
is the key to a reliable projection of future climate change.
Most studies have been focusing on the response of the climate system on the time scales from years to centuries (Meehl et al 2007). Recently, an increasing number of studies investigated climate response in the conceptual framework of 'fast climate adjustment' and 'slow climate feedbacks' (e.g., Gregory and Webb 2008, Andrews et al 2009, Doutriaux-Boucher et al 2009, Cao et al 2011. In these studies fast climate adjustment refers to changes in climate variables such as precipitation and cloud due to changes in atmospheric structure that occurs before surface temperature change, and slow climate feedbacks refer to climate response associated with gradual changes in surface temperature. Total climate response can be approximated by the sum of fast climate adjustment and slow climate feedbacks. In practice, fast climate adjustment is estimated under the idealized situation with either zero global mean surface temperature change (Gregory and Webb 2008, Andrews et al 2009, Doutriaux-Boucher et al 2009, Cao et al 2011 or with fixed sea surface temperature (SST) (Gregory and Webb 2008.
Several studies have shown the importance of fast climate adjustment for the long-term climate response. For example, Gregory and Webb (2008) found that in response to a doubling of atmospheric CO 2 some of the model spread in equilibrium climate sensitivity can be attributed to the modeled differences in the fast adjustment of cloud forcing. Bala et al (2010) found that in response to a doubling of atmospheric CO 2 the modeled fast response for a few climate variables such as precipitation and evaporation is almost 40% of the total response. Furthermore, it is found that for different forcing agents model-estimated slow climate feedbacks is similar, but fast climate response is different (e.g., Andrews et al 2009, Cao et al 2011. In particular, a number of studies found that for the same amount of increase in global mean surface temperature modeled increase in global mean precipitation is greater in response to changes in solar irradiance than that in response to changes in atmospheric CO 2 content: the difference in precipitation response is mainly a result of different fast climate adjustment in response to CO 2 versus solar forcing (Andrews et al 2009, Cao et al 2011.
Despite the acknowledgment of the importance of fast climate response, an understanding of the development of the fast climate adjustment and the associated physical mechanisms has remained elusive. Doutriaux-Boucher et al (2009) investigated the adjustment of radiative forcing to a doubling and quadrupling of atmospheric CO 2 on a monthly resolution. Dong et al (2009), using an atmosphere general circulation model with fixed SST, studied daily evolution of climate change in response to a doubling of atmospheric CO 2 and imposed changes in SST over a period of one month. As an extension of previous studies, here we investigate daily climate response to CO 2 and solar forcing on the time scale of days to weeks using a coupled atmosphere-ocean climate model. To our knowledge, this is the first study that examines daily evolution of climate response to both CO 2 and solar forcing in a coupled atmosphere-ocean system. The main purpose of this study is to provide a mechanistic understanding of the development of fast climate response, and investigate how the development of fast climate adjustment differs in response to CO 2 versus solar forcing, and between land and ocean. While previous studies on fast climate response focused on the adjustment of cloud and radiative forcing (e.g., Gregory and Webb 2008, Andrews et al 2009, here we focus on the response of temperature and the hydrological cycle on the times scale of days to weeks.

Methods
Here we briefly discuss the climate simulations performed in this study. A detailed description of the climate model used and simulations conducted is given in supplemental online materials (SOM).
We used the UK Met Office Hadley Centre global climate model, HadCM3L (Cox et al 2000), to investigate daily climate change in response to CO 2 and solar forcing. To achieve a clear mechanistic understanding of the daily climate response, we conducted large ensemble simulations (altogether 180 simulations) involving a step-function quadrupling of atmospheric CO 2 concentration and 4% increase in solar irradiance, together with a set of control simulation with a default atmospheric CO 2 concentration of 280 ppm and solar constant of 1365 W m −2 . Each experiment simulated one month of elapsed time and we saved daily-mean output. Following the experiment design of Doutriaux- Boucher et al (2009), for each set of simulation under a specific forcing, the effect of seasonal cycle is removed by averaging the results of twelve individual runs starting from each month of the year.
Increasing atmospheric CO 2 exerts a forcing on the climate system not only through its greenhouse effect on atmospheric radiation, but also through its physiological effect on the opening of plant stomata (Sellers et al 1996, Betts et al 2007, Doutriaux-Boucher et al 2009, Cao et al 2010, Andrews et al 2011. To investigate the individual and combined effect of CO 2 -radiative and physiological forcing, we conduct a series of simulations to separate the climate effect resulting from the change of atmospheric radiation and/or plant stomata to increasing atmospheric CO 2 content. In the simulations with increased solar irradiance, a 4% increase in model solar constant is chosen based on the fact that this amount of increase in solar irradiance yields similar long-term change in global mean surface temperature to that in response to a quadrupling of atmospheric CO 2 : averaged over the last 100 years of the 1000 yr simulations: a quadrupling of atmospheric CO 2 and 4% increase in solar irradiance causes a surface temperature change of 5.71 and 5.70 K, respectively.
In addition to the one-month simulations with daily mean output as described above, we also analyze annual mean climate change in response to a step-function quadrupling of atmospheric CO 2 and 4% increase in solar irradiance at both the monthly and yearly resolutions. In the following discussions, all results are presented in terms of the change in perturbation simulations relative to the control simulation.

Fast climate adjustment in the context of long-term climate change
The simulated change in global mean precipitation against global mean surface temperature at the daily, monthly and Figure 1. (a) HadCM3L-simulated change in global mean precipitation against change in global mean surface temperature for the simulation that involves a quadrupling of atmospheric CO 2 (4 × CO 2 , black) and simulation that involves a 4% increase in solar irradiance (4% solar, red). Results are shown at the daily (square, day 1 to day 30), monthly (circle, month 1 to month 60), and yearly (plus sign, year 1 to year 15) resolutions. Results at the monthly resolution are obtained by averaging twelve 5 yr runs starting from different months of the year, and results at the yearly resolution are obtained from the first 15 years of the 1000 yr simulations. Black and red lines are linear regression to the data at monthly and yearly resolutions as described in the SOM. (b) Daily evolution of global mean precipitation change in response to a quadrupling of atmospheric CO 2 (thick black line) and 4% increase in solar irradiance (thick red line) plotted with regressed fast precipitation response (thin black line for 4 × CO 2 and thin red line for 4% increase in solar irradiance) as determined from the vertical-axis intercept in the linear regression shown in panel (a). Shaded bands represent the 95% confidence interval for the regressed fast precipitation response. A version with land-mean and ocean-mean results is shown in figure S1 (available at stacks.iop. org/ERL/7/034015/mmedia). yearly resolutions is shown in figure 1 (corresponding land-mean and ocean-mean results are given in figure S1 available at stacks.iop.org/ERL/7/034015/mmedia). The distinct behavior of precipitation change on different time scales is clear. During the first month there is a general decrease in global mean precipitation as surface temperature increases with the decrease being much greater for CO 2 forcing than for solar forcing. On the time scale beyond one month, in response to both CO 2 and solar forcing global mean precipitation increases with the increase in surface temperature.
We performed linear regression of precipitation change against changes in global mean surface temperature using annual mean data at monthly resolution for the first 5 yr and yearly data after 5 yr (figure 1, figure S1 available at stacks.iop.org/ERL/7/034015/mmedia). Linear regressions show that in the limit of zero global mean surface temperature change, there is a significant reduction in global mean precipitation in response to CO 2 forcing, whereas increased solar irradiance causes a much smaller reduction in precipitation (figure 1, table S1 available at stacks.iop. org/ERL/7/034015/mmedia). This finding is consistent with previous modeling studies on the fast adjustment of climate system (e.g., Andrews et al 2009, Andrews et al 2011. However, as demonstrated in figure 1(b), the regressed fast precipitation response neglects climate processes occurring on time scales shorter than the first data point used for the regression (here shorter than one month). In reality, immediately after a forcing is imposed, surface temperature starts to respond, and so does precipitation. In the following we examine the daily climate evolution and development of fast climate adjustment during the first month in response to CO 2 and solar forcing.

Daily climate change
In the following analysis we focus on land-mean and ocean-mean climate responses, which are presented in figures 2-5. Corresponding global mean results are presented in figures S2-6 (available at stacks.iop.org/ERL/7/034015/ mmedia). Spatial distribution of daily climate response and numerical values of daily climate changes are presented in figures S7-8 and tables S2-5 (available at stacks.iop.org/ERL/ 7/034015/mmedia).

Daily evolution of temperature change.
Increasing atmospheric CO 2 concentration perturbs the Earth energy balance primarily by absorbing and emitting longwave radiation, whereas increasing solar irradiance perturbs the Earth energy balance primarily by increasing the incoming shortwave radiation. Increasing atmospheric CO 2 imposes an additional form of climate forcing by suppressing plant transpiration through reduced opening of plant stomata, an effect known as CO 2 -physiological forcing (Sellers et al 1996, Betts et  Model-simulated daily evolution of temperature change over land and ocean is shown in figure 2 and the change in global mean values is given in figure S2 (available at stacks. iop.org/ERL/7/034015/mmedia). With a small effective heat capacity land warms rapidly in response to both CO 2 and solar forcing (figure 2(a)). As for CO 2 forcing, on one hand, increased CO 2 concentration warms the land surface through its radiative effect that increases downward longwave radiation. On the other hand, increased CO 2 concentration further warms the land surface through its physiological effect that reduces evaporative cooling and increases shortwave radiation reaching the surface as a result of decrease in low cloudiness (figure S3 available at stacks.iop.org/ERL/ 7/034015/mmedia). This mechanism of surface warming associated with CO 2 -physiological forcing has been discussed extensively in previous studies (e.g., Boucher et al 2009, Dong et al 2009, Cao et al 2010, Andrews et al 2011. In response to a quadrupling of atmospheric CO 2 , averaged on day one, land surface warms by 0.57 ± 0.03 K (uncertainty is represented by one standard error calculated from the ensemble simulations). On day five, land surface has warmed by 1.39 ± 0.03 K, which is 16% of the near-equilibrium warming (Here, near-equilibrium warming is estimated by the averaged surface warming over the last 100 years of the 1000 yr simulations.) ( figure 2(b)). The importance of CO 2 -physiological forcing on the short-term land warming Figure 2. HadCM3L-simulated daily changes in temperature for the simulation involving a quadrupling of atmospheric CO 2 (4 × CO 2 , thick black lines), a quadrupling of atmospheric CO 2 but with only CO 2 -radiative effect included (4 × CO 2 RAD, thin black lines), and 4% increase in solar irradiance (4% solar, thick red lines). Upper panels are land-mean results and lower panels are ocean-mean results. Shown are surface temperature change relative to the mean of the control simulation ((a), (d)), fractional surface temperature change relative to near-equilibrium warming (here near-equilibrium warming is estimated by the warming averaged over the last 100 years of the 1000 yr simulations) ((b), (e)), and differences between temperature change at 850 mb and surface (temperature change at 850 mb minus temperature change at the surface) ((c), (f)). The dashed lines represent inter-daily variability (2σ ) calculated from the standard deviation of the control simulation. A version with global mean results is shown in figure S2 (available at stacks.iop.org/ERL/7/034015/mmedia).

Figure 3.
HadCM3L-simulated changes in the vertical profile of temperature at day one, two, five and thirty for the simulation involving a quadrupling of atmospheric CO 2 (4 × CO 2 , thick black lines), a quadrupling of atmospheric CO 2 but with only CO 2 -radiative effect included (4 × CO 2 RAD, thin black lines), and 4% increase in solar irradiance (4% solar, thick red lines). All results are relative to the control simulation. Upper panels are land-mean results and lower panels are ocean-mean results. A version with global mean results is shown in figure S4 (available at stacks.iop.org/ERL/7/034015/mmedia). is evident: during the first month more than half of the CO 2 -induced warming is from the response of plant stomatal (figures 2(a) and (b)). Compared to the effect of CO 2 forcing, increased solar irradiance warms the land mainly through increased shortwave radiation reaching the surface. In response to a 4% increase in solar irradiance, land surface warms by 0.14 ± 0.03 and 0.46 ± 0.04 K averaged over day one and day five, respectively.
With a large effective heat capacity, ocean surface experiences little warming during the first month in response to both CO 2 and solar forcing ( figure 2(d)). Temperature change at ocean surface is nearly zero on day one, and on day five, ocean surface only warms by 0.05 ± 0.01 and 0.03 ± 0.01 K in response to CO 2 and solar forcing, respectively. Different characteristics of changes in vertical temperature profile are observed between land and ocean, and in response to CO 2 versus solar forcing (figures 2(c), (f) and 3).
Over land, as a result of the rapid surface warming, in response to both CO 2 and solar forcing surface warms more than the overlying atmosphere (figure 2(c)), indicating an increased lapse rate and decreased vertical stability in the lower atmosphere. Over ocean, surface warms less than the lower atmosphere in response to both CO 2 and solar forcing (figure 2(f)), indicating a decreased lapse rate and increased vertical stability in the lower atmosphere. Averaged over day five, over ocean, a quadrupling of atmospheric CO 2 causes a warming of 0.34 ± 0.02 K around 850 mb, compared to the warming of 0.05 ± 0.01 K at the surface. It is interesting to note that the effect of CO 2 -physiological forcing further increases the warming of the atmosphere around 850 mb over ocean (figures 2(f) and 3), indicating that additional land surface warming caused by CO 2 -physiological effect spreads to the lower atmosphere over the ocean. Compared to the effect of increased atmospheric CO 2 , increased solar . HadCM3L-simulated changes in boundary layer moisture flux export ((a), (e)), near-surface specific humidity ((b), (f)), near-surface relative humidity ((c) (g)), and near-surface specific humidity deficit (saturation specific humidity at surface minus near-surface specific humidity) ((d), (h)) for the simulation involving a quadrupling of atmospheric CO 2 (4 × CO 2 , thick black lines), a quadrupling of atmospheric CO 2 but with only CO 2 -radiative effect included (4 × CO 2 RAD, thin black lines), and 4% increase in solar irradiance (4× solar, thick red lines). All results are relative to the control simulation. Upper panels are land-mean results and lower panels are ocean-mean results. The dashed lines represent inter-daily variability (2σ ) calculated from the standard deviation of the control simulation. A version with global mean results is shown in figure S5 (available at stacks.iop.org/ERL/7/034015/mmedia).

Figure 5.
HadCM3L-simulated daily changes in the global water cycle for the simulation involving a quadrupling of atmospheric CO 2 (4 × CO 2 , thick black lines), a quadrupling of atmospheric CO 2 but with only CO 2 -radiative effect included (4 × CO 2 RAD, thin black lines), and 4% increase in solar irradiance (4% solar, thick red lines). All results are relative to the control simulation. Upper panels are land-mean results and lower panels are ocean-mean results. Shown are evaporation ((a), (d)), precipitation ((b), (e)), and precipitation minus evaporation ((c), (f)). The dashed lines represent inter-daily variability (2σ ) calculated from the standard deviation of the control simulation. A version with global mean results is shown in figure S6 (available at stacks.iop.org/ERL/7/034015/mmedia).
irradiance causes a much smaller vertical gradient between the warming at ocean surface and that at the lower atmosphere (figures 2(f) and 3), indicating a much smaller change in atmospheric stability. The different vertical gradients of temperature change resulting from CO 2 versus solar forcing is consistent with the fact that the atmosphere is more transparent to shortwave radiation than longwave radiation (Hansen et al 1997).

Daily evolution of changes in the hydrological cycle.
Different characteristics of changes in the hydrological cycle are observed between land and ocean, and in response to CO 2 versus solar forcing (figures 4 and 5, figures S5 and S6 available at stacks.iop.org/ERL/7/034015/mmedia). Over land, rapid surface warming decreases vertical stability of the lower atmosphere, and thus the radiative effect of both CO 2 and solar forcing tends to increase the export of moisture flux from the boundary layer (figure 4(a)), enhancing evaporation and precipitation (figures 5(a) and (b)). However, during the first month the CO 2 -induced change in land hydrology is dominated by the CO 2 -physiological effect. Reduced opening of plant stomata decreases the source of water vapor to the atmosphere through reduced plant transpiration, drying the boundary layer (figures 4(b) and (c)), and subsequently decreasing evaporation and precipitation (figures 5(a) and (b)). In response to both CO 2 and solar forcing, rapid land surface warming induces a monsoonal type of flow that carries additional water vapor from the ocean to land, causing a freshwater gain over land (figure 5(c)) and freshwater deficit over ocean (figure 5(f)). Figure 6. A schematic illustration of HadCM3L-simulated climate change over land and ocean at day five in response to a quadrupling of atmospheric CO 2 and a 4% increase in solar irradiance. Green arrows represent the direction of change in boundary layer moisture flux with upward arrows indicating an increase in moisture flux out of the boundary layer and downward arrows indicating a decrease in moisture flux out of the boundary layer. Blue arrows represent the transport of water vapor from ocean to land. A version of global mean response is given in figure S9 (available at stacks.iop.org/ERL/7/034015/mmedia).
Over ocean, increased vertical stability in response to CO 2 forcing tends to suppress convective activity and moisten the boundary layer, increasing near-surface specific humidity (figure 4(f)) and decreasing near-surface specific humidity deficit (saturation specific humidity at surface minus near-surface specific humidity) (figure 4(h)). As a result, evaporation, which is directly controlled by the specific humidity deficit at boundary layer, decreases over ocean ( figure 5(d)). Consequently, precipitation decreases over ocean too ( figure 5(e)). In addition to the radiative effect, the inclusion of CO 2 -physiological forcing further reduces precipitation over ocean (figure 5(e)) mainly as a result of enhanced water vapor transport from the ocean to land associated with the additional land warming. Compared to CO 2 forcing, increased solar irradiance causes a much smaller change in evaporation and precipitation over ocean (figures 5(d) and (e)) as a result of much smaller change in vertical temperature gradient between the surface and lower atmosphere (figure 2(f)). Averaged over day five, a quadrupling of atmospheric CO 2 decreases precipitation by −0.07 ± 0.001 m yr −1 , whereas a 4% increase in solar irradiance causes a change in precipitation of −0.01 ± 0.003 m yr −1 .

Discussion and conclusions
Using HadCM3L model simulations under idealized stepfunction increases in atmospheric CO 2 content and solar irradiance, we have examined climate response on the time scale of days to weeks. Previous studies investigating fast climate response under the assumption of either zero global mean surface temperature change or fixed sea surface temperature reported a reduction in global mean precipitation in response to increased atmospheric CO 2 concentrations (e.g., Bala et al 2010, Andrews et al 2009, Cao et al 2011. This study provides a mechanistic understanding of the development of the fast climate response. Over land, a decrease in precipitation results from the physiological effect of increased CO 2 on plant stomata that reduces a source of water vapor to the overlying atmosphere. Over ocean, a decrease in precipitation is a result of increased vertical stability at the lower atmosphere that reduces the moisture export from the boundary layer. A schematic illustration of different daily climate response between land and ocean and to CO 2 versus solar forcing is given in figures 6 and S9 (available at stacks.iop.org/ERL/7/ 034015/mmedia). We acknowledge that the conclusion drawn here is from simulations using a single model. More studies using different climate models are encouraged to test the robustness of the finding here and to further explore the associated climate processes.
An understanding of the climate response to the forcing of carbon dioxide and solar irradiance (and other types forcing agents) depends on clearly delineating direct effects on atmospheric structure from effects resulting from gradual planetary warming brought about by an energy imbalance at the top of atmosphere. Here we have shown that significant climate effects from a step-function increase in atmospheric CO 2 and solar irradiance is manifested within days. Our study focuses on step-function changes in radiative forcing, but any continuous time series of radiative forcing changes may be thought of as a convolution of infinitesimal step-function changes (Good et al 2011). Thus, the conclusions drawn here also apply to continuous changes in radiative forcing. A good understanding of climate response on the time scale of days is an important element towards a complete understanding of the climate response on time scales ranging from seconds to millions of years.