SEARCH

SEARCH BY CITATION

Keywords:

  • polar warming amplification;
  • climate feedbacks

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Method and Data
  5. 3. Results
  6. 4. Summary
  7. Acknowledgments
  8. References

[1] The IPCC AR4 global warming climate simulations reveal a pronounced seasonality of polar warming amplification with maximum warming amplification in winter and minimum in summer. In this paper, we study the relative importance of surface albedo feedback (SAF), changes in cloud radiative forcing (CRF), changes in surface sensible and latent heat fluxes, changes in heat storage, and changes in the clear-sky downward infrared radiation in causing the strong seasonality of polar warming amplification by calculating partial temperature changes due to each of these processes using the surface energy budget equation. The main thermodynamic factor for a small polar warming amplification in summer is that the positive SAF is largely cancelled out by the negative surface CRF feedback in summer. The positive SAF is relatively much weaker in winter compared to its amplitude in summer, therefore does not contribute to the pronounced polar warming amplification in winter. The seasonal cycle of polar surface warming amplification, in terms of both spatial patterns and temporal amplitude, closely follows the seasonal cycle of the warming due to changes in clear-sky downward longwave radiation alone, indicating the importance of the atmospheric processes, such as water vapor feedback and dynamical feedbacks associated with the enhancement of poleward moist static energy transport, in causing the pronounced seasonality of polar warming amplification.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Method and Data
  5. 3. Results
  6. 4. Summary
  7. Acknowledgments
  8. References

[2] Polar amplification of surface warming is found in both observations and model simulations. The surface albedo feedback (SAF), cloud feedback, and poleward heat transport feedback contribute to the polar warming amplification [e.g., Holland and Bitz, 2003; Vavrus, 2004; Cai, 2005; Winton, 2006]. Because multi-processes are involved in the polar warming amplification, it is important to choose a suitable method to perform the climate feedback analysis [Winton, 2006]. The methods based on the radiation balance at the top of the atmosphere (TOA) have limitations in quantifying polar warming amplification, because the net change in radiation at the TOA, in response to external forcing, is not in balance locally, but has to be balanced by the atmospheric and oceanic poleward energy transport.

[3] The seasonality of polar warming has long been found in climate simulations [Rind, 1987; Holland and Bitz, 2003], but less emphasis was put on it because the climate sensitivity analysis mainly concerns with the change in annual mean temperature. As shown in Figure 1, the polar surface warming is largest in winter and smallest in summer in both hemispheres. Furthermore the inter-model spread of climate sensitivity over high latitudes closely follows the seasonality of polar warming itself. These features and the seasonal difference in dynamic and thermodynamic processes highlights the need to understand the seasonality of polar warming amplification.

image

Figure 1. Zonal-averaged surface air temperature change (K) due to the doubling of CO2: The ensemble mean (contours) and standard variation (shadings) of the 11 CMIP3 model simulations.

Download figure to PowerPoint

[4] In this paper, we attempt to compare quantitatively the relative importance of SAF, changes in cloud radiative forcing (CRF), changes in surface sensible and latent heat fluxes, changes in heat storage, and changes in the clear-sky downward infrared radiation in the seasonality of polar surface warming amplification by calculating partial temperature changes (PTCs) due to each of these processes using the surface energy budget equation.

2. Method and Data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Method and Data
  5. 3. Results
  6. 4. Summary
  7. Acknowledgments
  8. References

[5] The monthly mean of surface heat budget equation can be written as:

  • equation image

where Q is the heat storage term for both land surface and ocean and it also includes oceanic energy transport in oceans. S[DOWNWARDS ARROW] and S[UPWARDS ARROW] are surface downward and upward short-wave radiations, α is the surface albedo, F[DOWNWARDS ARROW] and F[UPWARDS ARROW] are surface downward and upward long-wave (LW) radiations with F[UPWARDS ARROW]σTs4, in which the surface emissivity has been assumed to be one at all wavelengths, H and LE are surface sensible and latent heat fluxes. Then the perturbation of equation (1) due to the doubling of CO2 is, after rearranging the Q and F[UPWARDS ARROW] terms,

  • equation image

Before we discuss the perturbation surface energy budget equation further, let us pause for a while on the concept of CRF. The CRF, defined as the difference between the net total-sky and clear-sky radiations at the top of the atmosphere (TOA) or at the surface, has long been used to represent the effects of clouds on climate [Cess and Potter, 1988; Ramanathan et al., 1989]. Because part of SAF and other feedbacks are also included in the change of the CRF, Soden et al. [2004] highlighted the need to interpret the change in the CRF properly. To exclude the surface albedo feedback in the change of surface CRF (ΔCRFs), we define

  • equation image

where (·)cld = (·) − (·)clr, (·) denotes radiations for total sky condition, and (·)clr for clear sky condition, and equation image is the surface albedo of the unperturbed mean climate state. In equation (3), we have utilized the fact that at surface, ΔF[UPWARDS ARROW],cld = 0 because the mathematical expression for the radiation emitted from the surface is not directly related to clouds. Note that the definition of ΔCRFs is different from the one used by Vavrus [2006] since we retain the mean surface albedo term in ΔCRFs. Using equation (3), we can rewrite equation (2) as:

  • equation image

[6] In equation (4), the overbar denotes the unperturbed mean climate state. The terms on the right-hand side (RHS) of equation (4) represent SAF, the change in surface CRF, the non-SAF-induced change in clear-sky SW radiation, the change in downward clear-sky LW radiation fluxes, the change in heat storage, and the changes in surface sensible/latent fluxes, respectively. The change in downward clear-sky LW radiation fluxes represents the sum of downward LW radiation changes at surface due the doubling of CO2, due to changes in atmospheric water vapor (water vapor feedback), and changes in both vertical and horizontal moist static energy transport by atmospheric motions (dynamical feedbacks).

[7] Each term in the RHS of equation (4), except for the heat storage term and the surface albedo, can be directly obtained from the CMIP3 model outputs. The surface albedo can be derived from the ratio of S[UPWARDS ARROW] to S[DOWNWARDS ARROW] at surface. The heat storage (Q) can be obtained by applying equation (1) in both the unperturbed and perturbed climate states, and then the ΔQ term can be obtained from the difference. Divided by 4σequation images3, the coefficient of ΔTs in equation (4), each term on the RHS of equation (4) constitutes the partial temperature change (PTC) due to the corresponding surface energy perturbation term. The sum of the PTCs equals, in the linear sense, to the actual (or total) surface warming. The calculation is made at each model grid, but only the zonal average will be presented to focus on the meridional pattern of global warming.

[8] The data used in the analysis come from the outputs of the control (1 × CO2) and 2 × CO2 equilibrium experiments of 11 CMIP3 coupled slab ocean-atmospheric general circulation climate models. The data for the last 20 years of each simulation are utilized. Except in Figure 1, only the simulations by GFDL CM2.0 will be presented here. The same procedure has been applied to other model simulations as long as they provide requisite data for equation (4). The main results in the paper remain robust though the details may be different among different models.

3. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Method and Data
  5. 3. Results
  6. 4. Summary
  7. Acknowledgments
  8. References

[9] Figure 2 shows the meridional-vertical profiles of the zonal-averaged changes in atmospheric temperature for the four seasons, derived from the GFDL CM2.0 global warming experiments. The ratio of the zonal-averaged temperature change to the global-averaged temperature change at each atmosphere level (shadings) depicts a rich seasonal cycle signal in the meridional-vertical profile of the warming amplification in the troposphere. It is seen except for summer seasons (JJA for Northern Hemisphere and DJF for Southern Hemisphere), there exists a polar warming amplification in both hemispheres and the polar warming amplification is strongest at the surface and gradually diminishes with height. The Antarctic warming amplification, which exists mainly near the surface, is much weaker than the Arctic warming amplification. In JJA, the warming amplification over Arctic exists in mid-troposphere, between 400 hPa and 600 hPa, instead of at the ground. During the austral summer (DJF), there is no Antarctic warming amplification both near the surface and in the troposphere, although the mid-tropospheric warming in Antarctic area is larger than that near the surface. Figure 2c and the results from other CMIP3 models (not shown) suggest that the vertical structure of Arctic warming in summer in climate simulations is qualitatively consistent with the results obtained from the reanalysis data [Graversen et al., 2008].

image

Figure 2. Zonal-averaged atmosphere temperature changes (contours, K), and the ratio (shadings) of the zonal-averaged temperature change to the global-averaged temperature change at each level, of (a) DJF; (b) MAM; (c) JJA; (d) SON.

Download figure to PowerPoint

[10] Next, we decompose the surface temperature change by utilizing equation (4). Figure 3a shows the annual cycle of surface warming at high latitudes of the Northern Hemisphere, derived from the GFDL CM2.0 global warming experiments. Figure 3b3g are the individual PTCs corresponding to the terms on RHS of equation (4), and Figure 3h is the sum of these PTCs. The annual mean difference between Figure 3h and Figure 3a, which measures the error due to the linearization of surface upward LW radiation adopted on the left hand side of equation (4), is about 0.3 K.

image

Figure 3. The zonal-averaged surface temperature change (K) in the doubling CO2 simulation by GFDL CM 2.0 over the high latitudes of the Northern Hemisphere: (a) the total Arctic warming; the partial temperature changes due to (b) surface albedo feedback, (c) the change in surface CRF, (d) the non-SAF-induced change in clear-sky SW radiation, (e) the change in net clear-sky LW radiation fluxes, (f) the change in heat storage, and (g) the changes in surface sensible/latent fluxes, respectively; (h) the sum of Figures 2b to 2g. The curves at the bottom of each panel are the annual means of the annual cycles shown on the top.

Download figure to PowerPoint

[11] The Arctic surface warming is strongest in winter, and weakest in summer (Figure 3a). SAF (Figure 3b) reaches its maximum in summer, with the PTC being about 5K, while SAF is understandably negligible in winter. It has been suggested in the literature that SAF can contribute to the polar warming amplification in winter, through the increased heat storage in summer, which damps summer warming potential and amplifies winter warming [Moritz et al., 2002; Bitz and Fu, 2008]. However, Figure 3c indicates that part of SAF is canceled out by the negative surface CRF feedback, which is mainly due to the increase in reflection of solar radiation by clouds in summer. As a result, only a fraction amount of increased solar absorption is available for the increase in ocean heat storage in summer and for releasing in winter as shown in Figure 3f. The spatial-temporal warming pattern associated with the heat storage (release) in summer (winter) seasons shows little resemblance with the annual cycle of the total Arctic surface warming (Figures 3a versus 3f). Moreover, the positive PTC in Figure 3f associated with the heat release in winter is mostly companied by the increase in the surface sensible and latent fluxes, which in turn acts to reduce the surface warming during cold seasons (Figure 3g). This suggests that the role of the change in inter-seasonal heat transfer is secondary in the Arctic warming asymmetry between summer and cold seasons.

[12] The reflection of solar radiation by clouds is negligible during cold seasons when the solar radiation is minimum, but Figure 3c indicates that changes in surface CRF during cold seasons is positive, mainly in the LW part, with the corresponding PTCs being about 1-4 K. Although the change in surface CRF is in phase with the annual cycle of polar surface warming, its contribution to the polar warming amplification during cold seasons is not dominating, compared with the winter warming in Figure 3a. It should be noted that the annual cycle of the change in surface CRF is in phase with the annual cycle of the climatological mean surface CRF (not shown here. See Intrieri et al. [2002] for the annual cycle of the climatological mean surface CRF over Arctic).

[13] Figure 3d shows the PTC due to (1 − equation imageS[DOWNWARDS ARROW],clr, the non-SAF-induced change in the net clear-sky downward SW radiation at the surface. The main contribution to this term comes from the increase in water vapor, which increases the atmospheric absorption of solar radiation and reduces solar radiation at the surface. Because of its ability to absorb more solar radiation for the atmosphere-surface system, the SW effect of the increase in water vapor forms a positive feedback from the TOA-based method [Dessler et al., 2008], but from the perspective of surface energy budget, the SW effect of the increase in water vapor is negative.

[14] The annual cycle of the PTC due to the change in clear-sky downward LW radiation (Figure 3e) exhibits a similar spatial pattern as the annual cycle of the (total) surface warming (Figure 3a) with comparable amplitude. This implies much of the annual cycle of the polar surface warming is associated with the annual cycle of the increase in clear-sky downward LW radiation. Furthermore, the annual mean Arctic surface warming amplification is also mainly determined by the change in clear-sky downward LW radiation, as evident from the comparisons between the amplitude and its spatial pattern of the annual means of the (total) surface warming and their counterparts of these individual PTCs (the curves in the bottom of Figures 3a3h). The increase in clear-sky downward LW radiation at the surface represents the net effect of the changes in the atmospheric CO2 concentration, water vapor, and temperature. We calculated the direct effect of the doubling of CO2 on surface radiation under clear sky condition with a radiation transfer model [Fu and Liou, 1993], and found that it is about 2-3 W/m2 over polar areas, yielding about 1 K warming. Therefore ΔF[DOWNWARDS ARROW],clr and its annual cycle in Figure 3e should be mainly attributed to the changes in atmospheric temperature and water vapor. This suggests the importance of understanding the role of the change in atmosphere processes in polar surface warming. Particularly, part of the increase in atmosphere temperature over polar region is related to the enhancement of atmospheric poleward sensible and latent heat transport.

[15] The main features of the seasonality of Arctic warming can also be identified over the Antarctic (Figure 4). These features include the largest warming in winter, the largest SAF in summer, the seasonality of surface CRF, and also the close resemblance between the spatial-temporal pattern of PTC due to ΔF[DOWNWARDS ARROW],clr (Figure 4e) and the total Antarctic surface warming (Figure 4a). Again, the annual mean Antarctic surface warming amplification is also mainly determined by ΔF[DOWNWARDS ARROW],clr (the curves in the bottom of Figures 4a4h). However, some unique features for Antarctic warming should be stressed. First, unlike its counterpart in the Arctic, winter warming maximum in Antarctic is located at about 70°S, not at the South Pole. Second, the positive SAF maximum in summer (Figure 4b) is located at about 10 degrees further south than the negative ΔCRFs (Figure 4c). As a result, their cancellation is relatively weak in comparison with the Arctic, responsible for a dipole pattern in the change in the heat storage (Figure 4f) in warm seasons. The heat storage increases at the polar side of the dipole due to SAF, and decreases at the low latitude side of the dipole due to the negative surface CRF. Consequently, the change in the heat release during cold seasons, i.e., March to September, also forms a dipole pattern, which again is accompanied with in the dipole pattern of the changes in surface sensible and latent heat fluxes, as shown in Figure 4g. Both of the patterns in Figure 4f and Figure 4g show little resemblance with the annual cycle of the total Antarctic surface warming (Figure 4a).

image

Figure 4. The same as Figure 3, but for the Southern Hemisphere.

Download figure to PowerPoint

4. Summary

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Method and Data
  5. 3. Results
  6. 4. Summary
  7. Acknowledgments
  8. References

[16] The larger surface warming and strong surface warming amplification during cold seasons, as well as a larger tropospheric warming and the absence of the surface warming amplification in summer, are robust features of polar warming in CMIP3 model simulations. The negative surface CRF largely (partly) cancels out the positive surface albedo feedback (SAF) in summer in the Arctic (Antarctic), and may help to explain why SAF is not a dominating factor in the simulated Arctic amplification [Winton, 2006]. The cancellation reduces the increase in surface heat storage in summer, and makes the inter-seasonal energy transfer only a secondary contributor to the seasonality of polar warming amplification. Furthermore, the winter-release of the increased summer heat-storage is accompanied with the increase surface sensible and latent fluxes, with their sum being secondary to the polar surface warming in winter.

[17] The similarity between the change in clear-sky downward long-wave radiation at the surface and the polar surface warming is unique in high latitudes. Because the direct effect of the doubling of CO2 is relatively small, the similarity can be mainly attributed to changes in atmospheric temperature and water vapor. This suggests the importance of the atmospheric processes, such as the heat transport and the vertical structure of water vapor feedback and cloud feedback, which determines the change in atmosphere temperatures, in the seasonality of polar warming amplification. The physical attribution of the atmosphere temperature change is beyond the ability of surface energy budget analysis, but may be obtained by the coupled surface-atmosphere climate feedback analysis method [Lu and Cai, 2008; Cai and Lu, 2008].

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Method and Data
  5. 3. Results
  6. 4. Summary
  7. Acknowledgments
  8. References

[18] The authors greatly appreciate insightful comments from two anonymous reviewers. We acknowledge the modeling groups for making their simulations available for analysis, the Program for Climate Model Diagnosis and Intercomparison (PCMDI) for collecting and archiving the CMIP3 model output, and the WCRP's Working Group on Coupled Modelling (WGCM) for organizing the model data analysis activity. The WCRP CMIP3 multi-model dataset is supported by the Office of Science, U.S. Department of Energy. This work is supported by grants from NOAA CPO (GC06-038) and NSF (ATM-0833001).

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Method and Data
  5. 3. Results
  6. 4. Summary
  7. Acknowledgments
  8. References