1 3

DOI 10.1007/s00382-014-2371-6Clim Dyn

Quantitative decomposition of radiative and non‑radiative contributions to temperature anomalies related to siberian high variability

Tae‑Won Park · Jee‑Hoon Jeong · Yi Deng · Renjun Zhou · Ming Cai

Received: 23 May 2014 / Accepted: 9 October 2014 © Springer-Verlag Berlin Heidelberg 2014

effects of cloud and water vapor account for the remaining cold anomalies of −1.00 and −0.60 K, respectively, while surface dynamics (0.71 K) and latent heat flux (0.26 K) help to mitigate the cold temperature anomalies. Influences of ozone and albedo processes are found to be relatively weak.

Keywords Siberian High-related temerature change · Temperature decomposition · CFRAM · Radiative and non-radiative processess

1 Introduction

The Siberian High (SH) is a conspicuous surface high-pressure system observed over the Eurasian continent dur-ing boreal winter. It is recognized as a semi-permanent or quasi-stationary atmospheric center of action associated with the coldest and densest air masses found in the North-ern Hemisphere (Lydolf 1977). The SH is climatologically located over northern Mongolia but often amplifies and expands to a large part of East Asia, causing enormous influences on weather and climate in Northern Eurasia, East Asia, and even further in South Asia (Cohen et al. 2001). The influences of the SH activity range from syn-optic to interannual, and even to interdecadal time-scales. The amplification and southeastward expansion of the SH on synoptic to intraseasonal time-scales is a key factor in causing severe cold surge outbreak in East Asia (Boyle 1986; Zhang et al. 1997). The intraseasonal to interannual variability of the SH is closely connected with the strengths of East Asian winter monsoon (Jhun and Lee 2004), Arctic Oscillation (Jeong and Ho 2005; Park et al. 2011), strato-spheric circulation (Jeong et al. 2006), and El-Nino South-ern Oscillation (Chen et al. 2004). In addition, interdecadal variation of the SH is accompanied by long-term changes

Abstract In this study, we carried out an attribution analy-sis that quantitatively assessed relative contributions to the observed temperature anomalies associated with strong and weak Siberian High (SH). Relative contributions of radia-tive and non-radiative processes to the variation of surface temperature, in terms of both amplitude and spatial pattern, were analyzed. The strong SH activity leads to the continen-tal-scale cold temperature anomalies covering eastern Sibe-ria, Mongolia, East China, and Korea (i.e., SH domain). The decomposition of the observed temperature anomalies asso-ciated with the SH variation was achieved with the Coupled atmosphere–surface climate Feedback-Responses Analysis Method, in which the energy balance in the atmosphere–sur-face column and linearization of radiative energy perturbation are formulated. For the mean amplitude of −3.13 K of cold temperature anomaly over the SH domain, sensible heat flux is tightly connected with a cooling of −1.26 K. Atmospheric dynamics adds another −1.13 K through the large-scale cold advection originated from the high latitudes. The longwave

T. Park (*) · Y. Deng School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, GA, USAe-mail: park2760@gmail.com; taewon.park@eas.gatech.edu

J. Jeong Department of Oceanography, Chonnam National University, Gwangju, South Korea

R. Zhou CAS Key Laboratory of Geospace Environment, School of Earth and Space Sciences, University of Science and Technology of China, Hefei, Anhui, China

M. Cai Department of Earth, Ocean, and Atmospheric Science, Florida State University, Tallahassee, FL, USA

T.-W. Park et al.

1 3

in temperatures and large-scale atmospheric circulations in East Asia (Panagiotopoulos et al. 2005; Jeong et al. 2011).

Fundamental characteristics of the SH are very cold temperatures near surface and associated with strong sta-bility in the lower troposphere, which are sustained by cold and dry continental air masses accumulated in the SH domain. For the causes of cold temperatures, previous lit-eratures reported the radiative and non-radiative dynamical processes associated with the SH (Ding 1990). A strong radiative cooling above the snow-covered surface over Eurasian continent is most responsible for the formation of a shallow cold-core pressure system confined to the lower troposphere over northern Mongolia in early winter (Pana-giotopoulos et al. 2005). Ding and Krishnamurti (1987) emphasized the role of thermal advection of cold air in the amplification of the cold high, where dynamical process resulting from tropospheric horizontal convergence is pri-marily responsible for the initial pressure rise. Outside the SH domain, the cloud feedback produces extremely cold and dry conditions over large parts of central and eastern Siberia through cloud cover reduction and the subsequent longwave radiation loss (Lydolf 1977).

Despite efforts in evaluating the linkage of processes with SH-related cold temperatures, a complete and quan-titative account of the relative importance of various radia-tive and dynamical processes is still missing in terms of their contributions to the cold temperature. Recent studies suggested that the intraseasonal to interdecadal variation of SH intensity and accompanying cold surge outbreaks over East Asia are strongly influenced by radiative processes associated with surface conditions, e.g., snow cover (Jeong et al. 2011) and large-scale atmospheric conditions like the Arctic Oscillation (Gong et al. 2001; Jeong and Ho 2005) and Madden-Julian Oscillation (Jeong et al. 2005; Wang et al. 2012). It becomes increasingly more important to quantify the relative contributions of these processes on SH amplification for accurate weather prediction and interpre-tation of long-term climate change over the eastern Eurasia.

The purposes of this study are to fill the knowledge gap and to provide a comprehensive assessment utilizing a recently developed feedback analysis method, called by the Coupled atmosphere–surface climate Feedback Responses Analysis Method (CFRAM) (Cai and Lu 2009; Lu and Cai 2009). Deng et al. (2012) adopted the CFRAM to under-stand the relative roles of radiative and dynamical pro-cesses in contributing to warming sea surface temperature (SST) anomalies over the eastern equatorial Pacific associ-ated with the El Ninõ–Southern Oscillation (ENSO). Most recently, the CFRAM was used to quantify contributions of climate processes to the structure of zonal-mean tempera-ture changes driven by increased atmospheric CO2 concen-tration (Taylor et al. 2013) and to the annual-mean surface temperature biases (Park et al. 2014) in model simulations.

Here, we use the CFRAM to assess quantitative contribu-tions of individual radiative and dynamical processes (e.g., albedo, cloud, water vapor, sensible/latent heat flux, surface and atmospheric dynamics) to the SH-related temperature changes that are defined as differences between strong and weak SH winters.

This paper is organized as follows. Section 2 describes the application of CFRAM to a decomposition analysis of the SH-related cold temperature, including description of the data used, the mathematical formulation of CFRAM, and detailed procedures of decomposition. The main results are provided in Sect. 3, which contains two subsections. The first subsection reports the partial temperature anoma-lies due to radiative and dynamical processes, and the sec-ond subsection discusses the possible causes. The final sec-tion includes an overview of the decomposition results and some discussion.

2 Methodology

For input variables, the CFRAM analysis requires solar insolation at the top of the atmosphere, atmospheric/surface temperature, specific humidity, ozone mixing ratio, cloud amount, cloud liquid/ice water content, and surface albedo. Solar insolation at the top of the atmosphere (I) is calcu-lated by the following equation:

where S0 is solar constant of which values is 1,360 W m−2 approximately according to measurements from the Total Irradiance Monitor (TIM) on NASA’s Solar Radiation and Climate Experiment (SORCE) and a series of new radio-metric laboratory tests (Kopp and Lean 2011). ϕ is lati-tude, and δ is solar declination angle. RE and R0 indicate the distance of the Earth from the sun and its annual mean, respectively; for simplicity, we assume that they are same. H is length of day, which is defined as

Only the solar insolation is calculated from the equa-tion, other variables are taken from the European Cen-tre for Medium-range Weather Forecasts (ECMWF) Re-Analysis Interim (ERA-Interim) data set (Dee et al. 2011). ERA-Interim is the latest ECMWF global atmos-pheric reanalysis, which covers the period 1979 to pre-sent and has a horizontal resolution of 1.5° × 1.5° with 37 isobaric levels. Though it is used as an observation in the present study, the ERA-Interim is a data assimila-tion product where real observations blend with a model. Thus, variables in ERA-Interim have biases with respect

(1)I =R2

0

R2E

S0

π[H sin ϕ sin δ + cos ϕ cos δ cos H]

(2)H = cos−1

[− tan ϕ tan δ].

Quantitative decomposition of radiative and non-radiative contributions

1 3

to actual observations, particularly cloud-related vari-ables and surface heat flux estimates. The accuracy of the ERA-Interim has been discussed in a lot of recent litera-tures. For example, in Decker et al. (2012), the fidelity of the ERA-Interim surface fluxes is evaluated with global flux tower observations. Jakobson et al. (2012) examined standard variables in the Arctic from multi-reanalysis products and showed that the best overall performance came from the ERA-Interim. Through a comparison to satellite retrievals, the reliability of the ERA-Interim is validated for the Arctic (Zygmuntowska et al. 2012), the Antarctic Ocean (Tastula et al. 2013), and Antarctica (Bracegirdle and Marshall 2012; Rodrigo et al. 2013). In Wang and Zeng (2012) and Bao and Zhang (2013), biases of the ERA-Interim over the Tibetan Plateau were shown to be smaller compared to other reanalysis datasets. The ERA-Interim is also highly correlated with ground-based observations in Ireland (Mooney et al. 2011) and France (Szczypta et al. 2011). Jeong et al. (2011) showed that the ERA-Interim well reproduces the recent change of the Siberian High intensity found from observational datasets. Here we adopt the ERA-Interim as the observation given its reasonable quality and comprehensive list of available variables for the CFRAM diagnostics.

The continental-scale temperature variation related to the SH is derived from a composited difference of tem-perature anomalies between strong and weak SH winters. The SH intensity (SHI) is defined as the winter (Decem-ber–January–February, or DJF, when the SH tends to be the most active) mean sea level pressure (SLP) averaged over 80°–120°E and 40°–65°N where the center of the SH in DJF-mean climatological SLP is observed (Fig. 1a), fol-lowing Jeong et al. (2011). For 31 winters over the period 1979/1980 to 2009/2010, the SHI is calculated and its nor-malized value is plotted in Fig. 1b. In constructing the com-posites, we identify the strong (weak) SH winters when the normalized SHI is greater (less) than +1.0 (−1.0). Given the above definitions and procedures, a total of four strong

and five weak SH winters are selected and used in the com-posite analysis.

Lu and Cai (2009) developed the CFRAM based on the difference of the total energy balance in a multi-layer atmosphere–surface column between two time-mean cli-mate states written as

where ∆∂−→E

∂t is the difference in energy storage, and ∆

−→S

and ∆−→R are the differences in the divergence of the short-

wave radiation flux and the convergence of longwave radiation flux, respectively. ∆

−→Q non−radiative represents

the difference in the convergence of total energy flux due to non-radiative, dynamical processes. If the two climate states are steady, we can ignore the difference in energy storage. Assuming negligible impacts due to interactions among various radiative and non-radiative processes, we can then linearize the right hand side (RHS) of Eq. (3) and express them as the sum of partial energy differences due to changes in individual radiative and non-radiative processes as follows:

where superscripts α, w, c, and O3 stand for surface albedo, water vapor, cloud, and ozone, respectively. ∆

−→T is the

total temperature difference. (

∂−→R

∂−→T

)

is the Planck feedback

matrix. In Eq. (6), ∆−→Q (SH) and ∆

−→Q (LH), indicating energy

(3)∆∂−→E

∂t= ∆

−→S − ∆

−→R + ∆

−→Q non−radiative

,

(4)∆−→S ≈ ∆

−→S (α)

+ ∆−→S (w)

+ ∆−→S (c)

+ ∆−→S (O3),

(5)∆−→R ≈ ∆

−→R (w)

+ ∆−→R (c)

+ ∆−→R (O3) +

∂−→R

∂−→T

∆−→T ,

(6)∆

−→Q non−radiative

= ∆−→Q (SH)

+ ∆−→Q (LH)

+ ∆−→Q (sfc_dyn)

+ ∆−→Q (atmos_dyn)

,

(a) (b)

Fig. 1 a Climatological sea level pressure for December–January–February (units: hPa) and b normalized Siberian High Index. The box in (a) indicates the domain used for the Siberian High Index

T.-W. Park et al.

1 3

differences due to changes in sensible heat flux and latent heat flux, respectively, have non-zero values in all layers except at the surface and the lowest atmospheric layer. At the surface, ∆

−→Q (SH) and ∆

−→Q (LH) are changes in sensible

heat flux and latent heat flux, respectively, leaving from the surface to the atmosphere. In the lowest atmospheric layer, ∆

−→Q (SH) and ∆

−→Q (LH) are the changes in energy conver-

gence into the lowest-atmospheric layer from the surface due to surface latent and sensible heat fluxes, respectively. ∆

−→Q (sfc_dyn) is energy difference due to changes in sur-

face dynamics change. ∆−→Q (sfc_dyn) has non-zero values in

the surface layer and zero values in the atmospheric lay-ers above. ∆

−→Q (atmos_dyn) represents energy differences

driven by turbulent, convective and large-scale atmospheric motions; it has zero values in the surface layer and non-zero values in all the atmospheric layers above.

Substituting Eqs. (4)–(6) into Eq. (3), rearranging terms, and multiplying both sides of the resultant equation by (

∂−→R

∂−→T

)−1

, we obtain

Equation (7) states that local temperature differences between two climate states (∆

−→T) can be decomposed into

eight partial temperature differences due to (from left to right) changes in surface albedo, water vapor, cloud, ozone, sensible heat flux, latent heat flux, surface dynamics, and atmospheric dynamics. For more detail of the CFRAM formulation, read-ers are referred to Lu and Cai (2009) and Park et al. (2012).

To quantify the relative contributions of the eight pro-cesses on the RHS of Eq. (7) to continental-scale cold temperature anomalies related to the SHI, the two climate states being considered are the strong and weak SH win-ters. The radiative energy differences (i.e., the first four terms in the bracket on the RHS of Eq. (7)) and the Planck feedback matrix are derived by off-line radiative transfer calculations using the Fu-Liou radiative transfer model (Fu and Liou 1992, 1993). The radiative transfer code is applied to the composited values of all the input variables for the strong and weak SH winters. When computing cloud-induced radiative energy differences, the use of time and domain mean clouds can produce significant errors in radiative fluxes due to the unresolved sub-grid variability of cloudy atmosphere (Pincus and Klein 2000). To reduce such errors, Taylor et al. (2011, 2013) suggested the Monte Carlo Independent Column Approximation (MCICA) tech-nique utilizing a maximum-random overlap cloud genera-tor (Raisanen et al. 2004). In the MCICA method, each

(7)

∆−→T ≈

(

∂−→R

∂−→T

)−1

{

∆−→S (α)

+ ∆(−→S −

−→R )(w)

+ ∆(−→S −

−→R )(c)

+∆(−→S −

−→R )(O3

)+ ∆

−→Q (SH)

+ ∆−→Q (LH)

+∆−→Q (sfc_dyn)

+ ∆−→Q (atmos_dyn)

}

grid box is subdivided into 100 sub-grid boxes, and cloud profiles for each sub-grid box are stochastically generated following the maximum-random overlap rule (Geleyn and Hollingsworth 1979). In our study, we essentially followed the same approach to deal with cloud-related uncertainties in radiative flux calculations. The total energy differences due to changes in surface sensible and latent heat fluxes are estimated directly using the surface heat flux data in the ERA-Interim data set. The last two terms in the brackets on the RHS of Eq. (7), i.e., the energy differences due to sur-face and atmospheric dynamics, are estimated as the resid-uals following Eq. (3), with changes in the heat storage neglected, because the land and snow surface does not have a large specific heat capacity or a deep layer over which the energy can be stored in contrast to ocean.

3 Results

3.1 Decomposition of siberian high-related temperature

The composite surface air temperature differences between strong and weak SH winters, including SH-related cold temperatures, are shown in Fig. 2a. In general, the influ-ences of the SH over land are much more than over the ocean. To focus on the temperature differences over land, we set the temperature values over the ocean as “missing values.” Based on the standard two-sample t test, statisti-cally significant regions at the 99 % confidence level are observed in the SH domain including eastern Siberia, Mon-golia, East China, and Korea, indicating that the cold tem-perature anomalies are closely associated with the interan-nual variability of the SH over much of Eurasian continent. Interestingly, another statistically significant warm temper-ature change occurs over the Mediterranean-Sahara under the influence of strong SH activity. Hong et al. (2008) reported that the atmospheric teleconnection pattern associ-ated with the North Atlantic Oscillation (NAO) represents the dipole of SLP anomalies between eastern Siberia and the Mediterranean-Sahara. This dipole pattern seems to be associated with the cold temperature around the SH region and with the warm temperature over the Mediterranean-Sahara. In addition, cold temperature changes are found over Europe though they are much weaker and insignificant compared to the above-mentioned signals. Figure 2b shows the corresponding distribution of the temperature changes (sum of the eight partial temperature changes) obtained through the CFRAM analysis. In the CFRAM frame-work, since the partial temperature changes are obtained through linearizing radiative energy perturbations, to a certain degree the accuracy of the CFRAM can be evalu-ated by comparing the temperature differences directly calculated from the original data with the sum of the

Quantitative decomposition of radiative and non-radiative contributions

1 3

partial temperature changes (Lu and Cai 2009). Comparing Fig. 2b with Fig. 2a, the CFRAM does a reasonably good job in reproducing the SH-related anomalous temperature field directly derived from the ERA-interim data. It dem-onstrates that the linearization of radiative energy perturba-tion in the CFRAM calculation is a sound approximation.

Figure 3 shows the partial temperature changes due to the eight radiative and dynamical processes obtained through the CFRAM. Over the SH domain, cold tempera-ture changes related to the sensible heat flux contribute sig-nificantly to the total temperature change (Fig. 3a). Overall, the spatial distribution of the partial temperature changes due to sensible heat flux resembles that of the total temper-ature changes shown in Fig. 2, in particular, for cooling and warming over Europe and the Mediterranean-Sahara. This indicates the sensible heat flux is a primary driver of the SH-related temperature pattern. The contribution of sensi-ble heat flux to the cold temperature over the SH domain is counteracted partly by a negative contribution (i.e., warm-ing) from the latent heat flux (Fig. 3b). On the other hand, water vapor gives a positive contribution to enhance the total temperature changes (Fig. 3c). In Fig. 3d, the partial temperature changes due to cloud process are mainly over the region north of 50°N and account for the warming over Europe and cooling over Russia. The contributions of cloud process is confined mainly to significant cooling anomalies around the northern boundary of the SH domain. Figure 3e shows that atmospheric dynamics is another primary driver of cold temperatures over the SH domain whose contribu-tion is comparable with that of the sensible heat flux. This agrees with the previous studies that emphasized sensible heat transfer and cold advection as key factors to drive the cold temperatures (Ding and Krishnamurti 1987). The con-tribution of surface dynamics, from energy changes due to mostly horizontal heat diffusion in the soil and transport of energy by river run-off, provides localized temperature anomalies of warming and cooling over the SH domain

(Fig. 3f). The magnitude of contribution related to ozone change is minimal throughout the whole domain (Fig. 3g). Contribution of the albedo process is also weak, except over isolated regions with intra-winter variability of snow cover, a key factor to control albedo (Fig. 3h).

To quantify the relative contributions of individual pro-cesses to both the spatial pattern and mean amplitude of the total temperature changes at a given region, beyond the grid-by-grid attribution as done in Fig. 3, we calculate a pattern-amplitude projection (PAP) coefficient following Park et al. (2014). The PAP coefficient is defined as

where ϕ and λ are latitude and longitude, respectively; a is the mean radius of the Earth, and A is the area of the region under consideration. ΔT and ΔTi are respectively the total and par-tial temperature changes associated with the ith radiative/dynamical process at an individual grid-point. Instead of spa-tial average, the PAP is used in the present study in order to better estimate area-averaged contribution from each process to the total temperature changes when temperature changes of opposite signs co-exist and cancel each other over the regions under consideration. For example, if the partial temperature changes due to a specific process have the mostly uniform sign over the region under consideration (e.g., sensible heat flux, water vapor, and atmospheric dynamics), the quantifica-tion to use weighted spatial average would be fine. However, in a case that partial temperature changes have both nega-tive and positive signs (e.g., surface dynamics), area average will lead to significant cancellation of the signals. To avoid this problem, the PAP is essentially defined as the weighted area average of the observed temperature changes multiplied by a “pattern projection coefficient”. The sum of the pattern projection coefficients calculated for all radiative/dynamical

(8)PAPi = A−1

∫

A

a2∆T cos ϕd�dϕ ·

A−1

∫

A

a2∆Ti∆T cos ϕd�dϕ

A−1∫

A

a2(∆T)2 cos ϕd�dϕ

(a) (b)

Fig. 2 a Surface temperature difference (units: K) between strong and weak Siberian High winters. Dotted regions indicate the statisti-cally significant grid cells at 99 % confidence level. b Sum of partial

temperature changes (units: K) derived from the CFRAM according to Eq. (7). The box indicates the domain used for the Siberian High Index

T.-W. Park et al.

1 3

processes gives 1 exactly. This ensures that the sum of all the PAPs due to various processes for a given region equals to the area average of the total temperature changes over this region. The PAP thus provides a balanced measure of relative con-tributions from individual radiative and dynamical processes when considering both the spatial pattern and amplitude of temperature differences.

Here, we define the domain of SH influence as a given region where statistically significant cold temperature differences are found over the Eurasian domain shown in Fig. 2a. The estimated PAP coefficients for the eight

individual processes are shown in Fig. 4. Sensible heat flux, atmospheric dynamics and cloud process are found to be the major contributors to cold temperatures over the SH domain, whereas surface dynamics and latent heat flux partly cancel out the cold temperature. As expected from Fig. 3, the impacts of ozone and albedo process are weak. Dynamical forcing due to sensible heat flux changes contributes −1.26 K to the regional mean amplitude of −3.13 K. Atmospheric dynamics and cloud process add −1.13 and −1.00 K, respectively, to the total cooling. Providing the negative contribution to the above cooling,

(a)

(c)

(e)

(g) (h)

(f)

(d)

(b)

Fig. 3 Partial temperature changes (units: K) related to the Siberian High activity due to (a) sensible heat flux change, (b) latent heat flux change, (c) water vapor change, (d) cloud change, (e) atmospheric

dynamics, (f) surface dynamics, (g) ozone feedback, and (h) surface albedo change. Dotted regions are same as those in Fig. 2a

Quantitative decomposition of radiative and non-radiative contributions

1 3

surface dynamics produces a warming of 0.71 K. Latent heat flux also gives a warming of 0.26 K, which weakens the total cold temperature change.

3.2 Possible causes of the partial temperature changes

Next, we examine the possible causes of the SH-related partial temperature changes reported above. Note that within the scope of the current analysis, an explicit assess-ment of detailed energy perturbations by surface and atmospheric dynamics is not feasible due to the ‘‘off-line’’ approach where the energy perturbations related to dynam-ics are treated as the residuals. Instead, the energy pertur-bation by atmospheric dynamics is estimated implicitly by diagnosing horizontal temperature advection by large-scale atmospheric motions.

The energy perturbation by atmospheric dynamics is mostly linked to large-scale atmospheric motions. To assess this impact on the SH-induced temperature changes implicitly, we investigate the differences of temperature advection (Fig. 5a) and wind vector (Fig. 5b) at 850 hPa between strong and weak SH winters. Over the SH domain, lower-tropospheric cold advection originated from high latitudes, in particular the Chukchi Sea through eastern Russia (Fig. 5b), accounts for the cooling associated with the SH by the atmospheric dynamics (Fig. 3e). Figure 5c indicates that strong cold advection prevails throughout the entire troposphere with two maxima near the surface and in the upper troposphere. Takaya and Nakamura (2005a, b) pointed out that near-surface cold advection by anomalous wind across climatological temperature gradient modulates the amplification of the cold SH. The upper-tropospheric cold advection is also a favorable condition for the amplifi-cation of cold SH (Ding and Krishnamurti 1987).

Figure 6 shows the differences of surface sensible and latent heat fluxes between strong and weak SH winters directly calculated from the ERA-Interim data. In Fig. 6a,

the sign of the differences is opposite to the partial temper-ature change due to sensible heat flux (shown in Fig. 3a), because positive (negative) sensible heat flux change means more (less) conductive heat release from the surface to the atmosphere, resulting in cooling (warming) at the surface. As shown in Fig. 5c, for the SH domain, cold advection by atmospheric dynamics is stronger in the lower troposphere around 900 hPa than at the surface. This induces a sharp negative temperature gradient with altitude which leads to enhanced sensible heat flux from the surface to the lower atmosphere. Thus, a strong SHI is associated with exces-sive sensible heat flux over the SH domain, indicating a critical role of heat release from the surface in determin-ing SH-induced cold temperatures at the surface (Ding and Krishnamurti 1987). The cooling over the northern Europe is also contributed by the excessive sensible heat flux, whereas the strong SHI through reduced sensible heat flux is responsible for the warming over the Mediterranean-Sahara. Figure 6b plots the differences of latent heat flux

Fig. 4 Pattern-amplitude projections (PAPs) of the eight partial tem-perature changes (units: K) onto statistically significant cold tempera-ture regions over the SH domain shown in Fig. 2a

(a)

(b)

(c)

Fig. 5 Differences of (a) 850-hPa temperature advection (units: K m s−1), (b) 850-hPa wind, (c) vertical profile of temperature advection averaged over the Siberian High domain (units: K m s−1) between strong and weak Siberian High winters

T.-W. Park et al.

1 3

between strong and weak SH winters. Following a similar argument, the change of latent heat flux has opposite signs to the corresponding partial temperature changes shown in Fig. 3b. Latent heat flux is the heat flux from the surface to the atmosphere that is associated with evaporation or tran-spiration of water at the land surface. Therefore, an increase (decrease) in latent heat flux causes a decrease (increase) in temperature at the surface through more (less) evaporation or transpiration. In boreal winter, the ground over the SH domain is frozen and covered mostly with snow, which sig-nificantly limits the water availability and latent heat flux at the surface. Under a colder condition during a strong SH winter, the ground may be frozen more deeply and covered with more snow, leading to the reduction of latent heat flux

at the surface. As a result, reduced latent heat flux over the SH domain represents less heat being extracted from the surface, which tends to slightly mitigate the surface cooling by further sensible heat flux.

The differences of near-surface specific humidity between strong and weak SH winters are shown Fig. 7a, and its spatial distribution well matches that of the partial temperature change due to water vapor (Fig. 3c). There is also a general agreement between the surface humid-ity change and the surface radiative heating rate change (Fig. 7b) in that a positive (negative) humidity change leads to positive (negative) surface heating change. This confirms that surface temperature change related to water vapor is largely dominated by the longwave (i.e., greenhouse) effect

(a) (b)

Fig. 6 Differences of (a) surface sensible heat flux (units: W m−2) and (b) surface latent heat flux (units: W m−2) between strong and weak Siberian High winters

(a) (b)

(d)(c)

Fig. 7 Differences of (a) surface specific humidity (units: kg/kg), (b) surface heating rate (units: W m−2), (c) surface shortwave heating rate (units: W m−2), and (d) surface longwave heating rate (units: W m−2) between strong and weak Siberian High winters

Quantitative decomposition of radiative and non-radiative contributions

1 3

of water vapor (Fig. 7d) while the shortwave effect of water vapor (Fig. 7c) tends to partially counteract the longwave effect. Over mid- and high-latitudes in winter, the incident shortwave radiation is limited and the effect of water vapor on shortwave heating rate is minimal. The total heating rate is controlled mostly by the longwave effect. Because the SH during the wintertime is usually covered by snow, which has a very large emissivity, the longwave radiative cooling is very effective. As a result, reduced water vapor in the SH domain and over Europe is responsible for par-tial cold temperature change shown in Fig. 3c, indicating a positive feedback to the total cold temperature change. The partial warm temperature change over the Mediterranean-Sahara is related to the positive water vapor change.

The partial temperature change related to the cloud pro-cess is attributable to changes in cloud water content and the subsequent cloud forcing. In Fig. 8a, we present the dif-ference of cloud water content vertically integrated for the whole atmosphere. The widely spreading minimal clouds associated with the SH are observed over the central and eastern Siberian, which has a good agreement with the finding by Lydolf (1977). Figure 8b shows a change in sur-face radiative heating associated with cloud water content change (i.e., cloud forcing change), whose spatial distri-bution closely resembles that of the cloud-induced partial temperature changes shown in Fig. 3d. Comparing Fig. 8a and b, however, it is found that the cloud forcing change in low and mid latitudes (south of 45°N) is very small though there are considerable changes in cloud water content

associated with the SH. This indicates the impact of short-wave cloud forcing (SWCF), blocking effect of solar inso-lation reaching the surface, is mostly offset by greenhouse effect due to longwave cloud forcing (LWCF) over these regions (Fig. 8c, d). Over high-latitude regions (north of 45°N), on the other hand, negative (positive) cloud water content changes lead to negative (positive) cloud forc-ing at the surface, consistent with the presence of strong LWCF (Fig. 8d) and the absence of SWCF (Fig. 8c). Due to the dominant role of LWCF seen in Fig. 8d over cen-tral and eastern Siberia, decreases in cloud water content are mostly responsible for the cloud-related partial cool-ing through longwave radiation loss (Lydolf 1977). Warm-ing over Western Europe is, on the contrary, contributed by increased cloud water content and associated warming effect by the longwave radiation.

4 Summary and discussion

This study applies a new framework of the CFRAM to iso-late quantitative contributions from individual radiative and dynamical processes to the observed surface temperature changes associated with the SH in boreal winter. Through the CFRAM technique, the temperature change associated with the SH is decomposed into eight partial temperature changes associated with changes in surface albedo, water vapor, cloud, ozone, sensible heat flux, latent heat flux, surface dynamics, and atmospheric dynamics. It is found

(a) (b)

(d)(c)

Fig. 8 Differences of (a) vertically integrated cloud water content (units: kg/kg), (b) surface cloud forcing (units: W m−2), (c) surface shortwave cloud forcing (units: W m−2), and (d) surface longwave cloud forcing (units: W m−2) between strong and weak Siberian High winters

T.-W. Park et al.

1 3

that over the SH domain the sensible heat flux provides the largest cooling effect on the surface temperature anomalies associated with the strong SH. Contributions of atmos-pheric dynamics, cloud change and water vapor change are the second, third and fourth important contributors to the SH-related cold temperature change, respectively. On the other hand, surface dynamics and latent heat flux act to reduce the cold temperature anomalies. Possible causes of individual processes to the SH-related cold temperature change are investigated. During strong SH winters, there is an excessive release of sensible heat flux from the surface to the atmosphere that leads to the majority of cold tem-perature anomalies. The atmospheric dynamics supports the cooling over the SH domain by a strong cold advection associated with large-scale northerly wind. The enhanced vertical temperature gradient in the lower troposphere sup-ports the sensible heat flux as well. On the other hand, more frozen and snow-covered ground condition related to the strong SH winters reduces the latent heat flux from the surface that counteracts the cold temperature anomalies. The negative specific humidity at the surface contributes to the cold surface temperature changes in the SH domain through the longwave (i.e., greenhouse) effect of water vapor. The cloud water content also gives cold surface tem-perature changes through the dominant LWCF, in particu-lar, over central and eastern Siberia.

In this study, we assess separately the isolated contribu-tions of individual processes to the total SH-related cold temperature change. However, various processes, in par-ticular, the four dominant processes (i.e., sensible heat flux, atmospheric dynamics, water vapor, and cloud), which pro-vide positive contribution to the cold temperature change, are linked together. The amplified high pressure in strong SH winters brings anomalous northerly wind accompa-nying cold and dry air originated from high latitudes. Under condition of positive vertical temperature gradient associated with the lower atmospheric cold advection, an increased sensible heat flux over the snow-covered sur-face in Eurasia continent during boreal winter modulates the amplification of the cold SH (Ding and Krishnamurti 1987). Reduced specific humidity by the intrusion of dry air gives an additional cooling to the SH domain by a weakened greenhouse effect. The high-pressure system and dry air condition are related to the decrease of cloud cover and the subsequent negative LWCF anomaly, adding cloud-induced cooling to the total cold temperature change.

Despite the successful decomposition shown in Fig. 2, the limitation of CFRAM analysis should be noted care-fully. For instance, the diagnosis is highly dependent on the overall quality of the ERA-Interim data. Although the ERA-Interim is a blend of real observations with the most updated model physics/dynamics (Dee et al. 2011), it still has large discrepancies with respect to observations

(e.g., ground-based observations, flux tower data, and radi-osonde data), in particular in terms of cloud-related vari-ables and surface heat flux estimates. The discrepancies, particularly those related to clouds, have effects on the par-titioning of energy perturbations between cloud change and atmospheric–surface dynamics in the CFRAM.

This study demonstrates that the CFRAM is an effi-cient “off-line” diagnostic tool that aids our quantitative understanding of the relative contributions from various radiative and non-radiative/dynamical processes to the temperature anomalies associated with the SH variation. The results of quantitative attribution obtained through the CFRAM give diagnostic assessment for understanding which process is dominant or negligible to modulate the SH and its surface temperature variability. Furthermore, the diagnostic information can be applied to predict the surface temperature changes associated with the SH activ-ity through monitoring the individual processes. For the statistical prediction model, the present study helps us to determine processes to be weighted as predictors. In terms of dynamical prediction, the implication of this analy-sis is applicable to provide critical information for model developers to effectively trace origins of model defects and initiate targeted effort for model improvement. For an in-depth study beyond the interannual variability of the SH, our ongoing investigations explore the use of CFRAM analysis in evaluating daily evolution of SH activity, which includes a further separation of the atmospheric dynam-ics into components related to large-scale and convective-scale motions. We also plan to perform a diagnosis of the surface dynamics using “on-line” approach based on the model outputs.

Acknowledgments The ERA Interim data used in this study were provided by the European Centre for Medium-Range Weather Fore-casts (ECMWF). Yi Deng and Tae-Won Park were supported by the DOE’s Office of Science Regional and Global Climate Mod-eling (RGCM) Program (DE-SC0005596) and by NSF grants AGS-1147601 and AGS-1354402. Jee-Hoon Jeong was supported by the Korea Polar Research Institute (PE14010).

References

Bao XH, Zhang FQ (2013) Evaluation of NCEP-CFSR, NCEP-NCAR, ERA-Interim, and ERA-40 reanalysis datasets against independent sounding observations over the Tibetan Plateau. J Climate 26(1):206–214. doi:10.1175/JCLI-D-12-00056.1

Boyle JS (1986) Comparison of the synoptic conditions in midlati-tudes accompanying cold surges over Eastern Asia for the months of December 1974 and 1978. 1. Monthly mean fields and indi-vidual events. Mon Weather Rev 114(5):903–918

Bracegirdle TJ, Marshall GJ (2012) The reliability of antarctic tropo-spheric pressure and temperature in the latest global reanalyses. J Climate 25(20):7138–7146. doi:10.1175/JCLI-D-11-00685.1

Cai M, Lu JH (2009) A new framework for isolating individual feed-back processes in coupled general circulation climate models.

Quantitative decomposition of radiative and non-radiative contributions

1 3

Part II: method demonstrations and comparisons. Clim Dynam 32(6):887–900. doi:10.1007/s00382-008-0424-4

Chen TC, Huang WR, Yoon J (2004) Interannual variation of the east Asian cold surge activity. J Climate 17(2):401–413

Cohen J, Saito K, Entekhabi D (2001) The role of the Siberian high in Northern Hemisphere climate variability. Geophys Res Lett 28(2):299–302

Decker M, Brunke MA, Wang Z, Sakaguchi K, Zeng XB, Bosilovich MG (2012) Evaluation of the reanalysis products from GSFC, NCEP, and ECMWF using flux tower observations. J Climate 25(6):1916–1944. doi:10.1175/JCLI-D-11-00004.1

Dee DP, Uppala SM, Simmons AJ, Berrisford P, Poli P, Kobayashi S, Andrae U, Balmaseda MA, Balsamo G, Bauer P, Bechtold P, Bel-jaars ACM, van de Berg L, Bidlot J, Bormann N, Delsol C, Dra-gani R, Fuentes M, Geer AJ, Haimberger L, Healy SB, Hersbach H, Holm EV, Isaksen L, Kallberg P, Kohler M, Matricardi M, McNally AP, Monge-Sanz BM, Morcrette JJ, Park BK, Peubey C, de Rosnay P, Tavolato C, Thepaut JN, Vitart F (2011) The ERA-Interim reanal-ysis: configuration and performance of the data assimilation system. Q J Roy Meteor Soc 137(656):553–597. doi:10.1002/Qj.828

Deng Y, Park T-W, Cai M (2012) Process-based decomposition of the global surface temperature response to El Niño in Boreal Winter. J Atmos Sci 69(5):1706–1712. doi:10.1175/JAS-D-12-023.1

Ding YH (1990) Buildup, air-mass transformation and propagation of siberian high and its relations to cold surge in East-Asia. Mete-orol Atmos Phys 44(1–4):281–292

Ding Y, Krishnamurti TN (1987) Heat-budget of the siberian high and the winter monsoon. Mon Weather Rev 115(10):2428–2449

Fu Q, Liou KN (1992) On the correlated K-distribution method for radiative-transfer in nonhomogeneous atmospheres. J Atmos Sci 49(22):2139–2156

Fu Q, Liou KN (1993) Parameterization of the radiative properties of cirrus clouds. J Atmos Sci 50(13):2008–2025

Geleyn J-F, Hollingsworth A (1979) An economical, analytical method for the computation of the interaction between scattering and line absorption of radiation. Contribut Atmos Phys 52:1–16

Gong DY, Wang SW, Zhu JH (2001) East Asian winter monsoon and Arctic Oscillation. Geophys Res Lett 28(10):2073–2076

Hong CC, Hsu HH, Chia HH, Wu CY (2008) Decadal relationship between the North Atlantic Oscillation and cold surge frequency in Taiwan. Geophys Res Lett 35(24). doi:10.1029/2008gl034766

Jakobson E, Vihma T, Palo T, Jakobson L, Keernik H, Jaagus J (2012) Validation of atmospheric reanalyses over the central Arctic Ocean. Geophys Res Lett 39:L10802. doi:10.1029/2012GL051591

Jeong J-H, Ho C-H (2005) Changes in occurrence of cold surges over east Asia in association with Arctic Oscillation. Geophys Res Lett 32:L14704. doi:10.1029/2005GL023024

Jeong J-H, Ho C-H, Kim B-M, Kwon W-T (2005) Influence of the Madden-Julian Oscillation on wintertime surface air temperature and cold surges in east Asia. J Geophys Res Atmos 110:D11104. doi:10.1029/2004jd005408

Jeong J-H, Kim B-M, Ho C-H, Chen D, Lim G-H (2006) Strato-spheric origin of cold surge occurrence in East Asia. Geophys Res Lett 33:L14710. doi:10.1029/2006GL026607

Jeong JH, Ou TH, Linderholm HW, Kim BM, Kim SJ, Kug JS, Chen DL (2011) Recent recovery of the Siberian High intensity. J Geo-phys Res Atmos 116. doi:10.1029/2011jd015904

Jhun JG, Lee EJ (2004) A new East Asian winter monsoon index and associated characteristics of the winter monsoon. J Climate 17(4):711–726

Kopp G, Lean JL (2011) A new, lower value of total solar irradi-ance: evidence and climate significance. Geophys Res Lett 38(1):L01706. doi:10.1029/2010gl045777

Lu JH, Cai M (2009) A new framework for isolating individual feed-back processes in coupled general circulation climate models.

Part I: formulation. Clim Dynam 32(6):873–885. doi:10.1007/S00382-008-0425-3

Lydolf PE (1977) Climates of the Soviet Union. Elsevier, AmsterdamMooney PA, Mulligan FJ, Fealy R (2011) Comparison of ERA-40,

ERA-Interim and NCEP/NCAR reanalysis data with observed surface air temperatures over Ireland. Int J Climatol 31(4):545–557. doi:10.1002/JOC.2098

Panagiotopoulos F, Shahgedanova M, Hannachi A, Stephenson DB (2005) Observed trends and teleconnections of the Siberian high: a recently declining center of action. J Climate 18(9):1411–1422

Park T-W, Ho C-H, Yang S (2011) Relationship between the Arctic Oscillation and cold surges over East Asia. J Climate 24(1):68–83. doi:10.1175/2010jcli3529.1

Park T-W, Deng Y, Cai M (2012) Feedback attribution of the El Niño-Southern Oscillation-related atmospheric and surface temperature anomalies. J Geophys Res 117(D23):D23101. doi:10.1029/2012JD018468

Park T-W, Deng Y, Cai M, Jeong J-H, Zhou R (2014) A dissection of the surface temperature biases in the Community Earth System Model. Clim Dynam 1–17. doi:10.1007/s00382-013-2029-9

Pincus R, Klein SA (2000) Unresolved spatial variability and micro-physical process rates in large-scale models. J Geophys Res-Atmos 105(D22):27059–27065. doi:10.1029/2000JD900504

Raisanen P, Barker HW, Khairoutdinov MF, Li JN, Randall DA (2004) Stochastic generation of subgrid-scale cloudy columns for large-scale models. Q J Roy Meteor Soc 130(601):2047–2067. doi:10.1256/QJ.03.99

Rodrigo JS, Buchlin JM, van Beeck J, Lenaerts JTM, van den Broeke MR (2013) Evaluation of the antarctic surface wind climate from ERA reanalyses and RACMO2/ANT simulations based on automatic weather stations. Clim Dynam 40(1–2):353–376. doi:10.1007/s00382-012-1396-y

Szczypta C, Calvet JC, Albergel C, Balsamo G, Boussetta S, Carrer D, Lafont S, Meurey C (2011) Verification of the new ECMWF ERA-Interim reanalysis over France. Hydrol Earth Syst Sci 15(2):647–666. doi:10.5194/hess-15-647-2011

Takaya K, Nakamura H (2005a) Mechanisms of intraseasonal ampli-fication of the cold Siberian high. J Atmos Sci 62(12):4423–4440

Takaya K, Nakamura H (2005b) Geographical dependence of upper-level blocking formation associated with intraseasonal amplifica-tion of the Siberian high. J Atmos Sci 62(12):4441–4449

Tastula EM, Vihma T, Andreas EL, Galperin B (2013) Validation of the diurnal cycles in atmospheric reanalyses over Antarctic sea ice. J Geophys Res-Atmos 118(10):4194–4204. doi:10.1002/JGRD.50336

Taylor PC, Ellingson RG, Cai M (2011) Geographical distribu-tion of climate feedbacks in the NCAR CCSM3.0. J Climate 24(11):2737–2753. doi:10.1175/2010JCLI3788.1

Taylor PC, Cai M, Hu AX, Meehl J, Washington W, Zhang GJ (2013) A decomposition of feedback contributions to polar warming amplification. J Climate 26(18):7023–7043. doi:10.1175/JCLI-D-12-00696.1

Wang AH, Zeng XB (2012) Evaluation of multireanalysis products with in situ observations over the Tibetan Plateau. J Geophys Res Atmos 117:D05102. doi:10.1029/2011JD016553

Wang L, Kodera K, Chen W (2012) Observed triggering of tropical convection by a cold surge: implications for MJO initiation. Q J Roy Meteor Soc 138(668):1740–1750. doi:10.1002/Qj.1905

Zhang Y, Sperber KR, Boyle JS (1997) Climatology and interannual variation of the East Asian winter monsoon: results from the 1979-95 NCEP/NCAR reanalysis. Mon Weather Rev 125(10):2605–2619

Zygmuntowska M, Mauritsen T, Quaas J, Kaleschke L (2012) Arc-tic clouds and surface radiation—a critical comparison of satel-lite retrievals and the ERA-Interim reanalysis. Atmos Chem Phys 12(14):6667–6677. doi:10.5194/ACP-12-6667-2012