Sensitivity of Atmospheric Energetics to Optically Thin Ice Clouds During the Arctic Polar Night

Abstract

Cloud feedback is a major source of uncertainty in climate projections. In particular, Arctic clouds, arguably one of the most poorly understood aspects of the climate system, strongly modulate radiative energy fluxes from the Earth’s surface to the top of the atmosphere. In situ and satellite observations reveal the existence of ubiquitous optically thin ice clouds (TICs) in the Arctic during polar nights, whose influence on atmospheric energy is still poorly understood. This study quantifies the effect of TICs on the atmospheric energy budget during polar winter. A reanalysis-driven simulation based on the Canadian Regional Climate Model version 6 (CRCM6) was used with the Predicted Particle Properties (P3) scheme (2016) to produce an ensemble of 3 km mesh simulations. This set is composed of three simulations: CRCM6 (reference, the original dynamically coupled cloud formation), CRCM6 (nocld) (clear-sky) and CRCM6 (100%cld) (overcast, 100% cloud cover as a forcing perturbation). Using the regional energetic equations (Nikiema and Laprise), we compare the three cases to assess TIC forcing. The results show that TICs cool the atmosphere, with the difference between two simulations (cloud/no clouds) reaching up to −2 K/day, leading to a decrease in temperature on the order of ~−4 KMonth⁻¹. The energetics cycle indicates that the time-mean enthalpy generation term G_M and baroclinic conversion dominate Arctic circulation. The G_M acting on the available enthalpy reservoir (A_M) increased by a maximum value of ~5 W·m⁻² (58% on average) due to the effects of TICs, increasing in energy conversion. TICs also lead to average changes of 9% in time-mean available enthalpy and −5.9% in time-mean kinetic energy. Our work offers valuable insights into the Arctic winter atmosphere and provides the means to characterize clouds for radiative transfer calculations, to design measurement instruments, and to understand their climate feedback.

1. Introduction

Due to their influence on the radiative balance and hydrological cycle, clouds are an important component of the atmosphere, through which they affect long- (terrestrial) and shortwave (solar) radiative transfers; clouds also affect atmospheric dynamics through latent heat release and absorption during water phase changes and condensate mass loading. Precipitation that reaches the Earth’s surface also affects the land surface energy budget and the salinity of oceans. Despite their importance, the representation of clouds and cloud processes remains one of the largest sources of uncertainty in climate and numerical weather prediction models [1,2,3,4,5].

Arctic clouds, arguably one of the most poorly understood aspects of the Arctic climate system, strongly modulate radiative energy fluxes from the Earth’s surface to the top of the atmosphere [5,6,7,8,9,10,11]. As such, they have the potential to influence climate variability and change both in the Arctic and globally. For instance, the presence of clouds over sea ice in winter can be the difference between a −40 W·m⁻² imbalance in surface radiative energy and a balanced surface radiation budget, influencing surface temperature and sea ice growth [12,13,14].

The energy balance and climate of the Arctic are highly sensitive to disturbances in the microphysical properties of clouds [15,16], sea-ice cover, and greenhouse gases, especially water vapor, at low concentrations. This is particularly true for semi-transparent tenuous or optically thin ice clouds (TICs), which are characterized by a small crystal size radius (<100 μm) and low optical depth (<3), as their spectral emissivity and cooling rates are known to vary drastically as a function of crystal size and ice water content through Jacobian transmittance amplitude [17]. Some modeling studies have evaluated the influence of ice clouds on Arctic radiation and climate during winter [18,19,20]; however, the influence of TICs on Arctic atmospheric energetics has not yet been quantified. This can be achieved by performing an energy budget calculation based on the field equations that are used in numerical forecast and climate models [21] and analyzing the contributions of the generation and conversion terms.

A pioneer in our understanding of global atmospheric energy was Lorenz [22], who introduced the concept of available potential energy (APE), which can be converted into kinetic energy (KE). To better understand the role of transient weather systems in atmospheric energetics, Lorenz further decomposed the energy fields into two components: one associated with the zonal mean atmospheric state and the other with departures thereof, termed eddies. Much of our current understanding of global atmospheric energetics derives from Lorenz’s seminal work (e.g., [22,23,24,25,26,27,28,29] to name just a few). Following an alternative approach, Marquet [30,31,32] proposed a formalism based on available enthalpy (AE) instead of APE. Inspired by these previous works, Nikiéma and Laprise [21,33,34] established an approximate energy cycle applicable to a limited-area domain, considering both transient-eddy (or time) variability (TV) and internal (or inter-member) variability (IV). TV reflects the passage of weather events (e.g., storms, cyclones, and floods), while IV is specific to models and represents a measure of dispersion in many simulations. Clément et al. used TV to study a particularly intense storm observed over North America [35], decomposing atmospheric variables into their time-mean and time-variability states (perturbations) rather than into the zonal mean and deviations, as is usual when studying global energetics [22,36].

The aim of this current study is to quantify the effect of optically thin polar clouds (TICs) on atmospheric energetics during polar winter, analyzing the relationships between atmospheric dynamics and the diabatic processes of clouds during this period. In this study, we applied the regional energetic equations formulated by Nikiema and Laprise [21] and adapted by Clément et al. [35] to three reanalysis-driven simulations for January 2007 using the Canadian Regional Climate Model (CRCM6), including the Predicted Particle Properties (P3) cloud microphysical scheme proposed by Morrison and Milbrandt [37]. The physical properties of the energy budget cycle were analyzed while quantifying the effect of TICs. This paper is organized as follows: Section 2 expands on the used energy budget equations and describes the experimental design; Section 3 presents our results and analysis of the characteristics identified in the studied climate and period, and the impact of TICs on the monthly mean energetics; and finally, Section 4 provides a summary of our findings and our conclusions.

2. Diagnostic Equations and Data

2.1. Energy Budget Equations Associated with Climate Studies

In atmospheric systems, which are expressed in pressure coordinates, the basic forms of energy used are the specific enthalpy H = CpT and the specific kinetic energy K = V² (the term “specific” will henceforth be omitted to lighten the terminology). The most widely used approach in studying regional energetics is the concept of available enthalpy [21,30]. This approach has the advantage of representing a much smaller positive definite quantity than enthalpy.

Available enthalpy can be divided into its pressure-dependent component B and temperature-dependent component A [21,30]. Component A can be further divided into components associated with the time-mean A_M ∝ 〈T〉² and time-variability A_TV ∝ T′² atmospheric states. Kinetic energy can also be decomposed into a component associated with the time-mean K_M ∝ 〈V〉² and time-variability K_TV ∝ V′² atmospheric states, which can be achieved as follows:

Any atmospheric variable defined by $\mathsf{\Psi }ϵ\left\{\overrightarrow{F},Q,\overrightarrow{V}\left(u,v\right),\alpha ,\omega ,\mathsf{\Phi }\right\}$ can be decomposed into $\mathsf{\Psi }=〈\mathsf{\Psi }〉+{\mathsf{\Psi }}^{’}$ (i), where $〈\mathsf{\Psi }〉={\displaystyle \frac{1}{\tau }}\sum _{t=1}^{\tau }{\mathsf{\Psi }}_t$ is the time average, $\tau$ is the number of time steps, and ${\mathsf{\Psi }}^{’}=\mathsf{\Psi }-〈\mathsf{\Psi }〉$ is deviations from the mean, corresponding to temporal variability in our study.

In this study, the energy cycle equations formulated by [21] and adapted in [35] will be used.

The time evolution equation for the climatological mean available enthalpy associated with the time-mean temperature is given as

$${\displaystyle \frac{\partial A_M}{\partial t}}=G_M+I_{AB}-C_M-C_A-F_{AM}-H_{AM}$$

(1)

where

$$A_M={\displaystyle \frac{C_p}{2T_r}}{⟨T-T_r⟩}^2$$

$$G_M=⟨\left({\displaystyle \frac{T}{T_r}}-1\right)⟩⟨Q⟩$$

$$I_{AB}=-⟨\omega ⟩{\alpha }_r$$

$$C_M=-⟨\omega ⟩⟨\alpha ⟩$$

$$C_A=-{\displaystyle \frac{C_p}{T_r}}\left(⟨T^{\mathrm{’}}\overrightarrow{V^{\mathrm{’}}}⟩\cdot \overrightarrow{\mathit{\nabla }}\right)⟨T⟩$$

$$F_{AM}=\overrightarrow{\nabla }\cdot \left(⟨\overrightarrow{V}⟩A_M\right)$$

$$H_{AM}=\overrightarrow{\nabla }\cdot ⟨\overrightarrow{V\mathrm{’}}\left({\displaystyle \frac{C_p}{T_r}}⟨T-T_r⟩T\mathrm{’}\right)⟩$$

The time evolution equation for the time-mean available enthalpy associated with the time-variability temperature is given by

$${\displaystyle \frac{\partial A_{TV}}{\partial t}}=G_{TV}-C_{TV}+C_A-F_{ATV}-H_{ATV}$$

(2)

where

$$A_{TV}={\displaystyle \frac{C_p}{2T_r}}⟨T{\mathrm{’}}^2⟩$$

$$G_{TV}=⟨{\displaystyle \frac{T\mathrm{’}Q\mathrm{’}}{T_r}}⟩$$

$$C_{TV}=-⟨{\omega }^{\mathrm{’}}{\alpha }^{\mathrm{’}}⟩$$

$$C_A=-{\displaystyle \frac{C_p}{T_r}}\left(⟨T^{\mathrm{’}}\overrightarrow{V^{\mathrm{’}}}⟩\cdot \overrightarrow{\nabla }\right)⟨T⟩$$

$$F_{ATV}=\overrightarrow{\nabla }\cdot \left(⟨\overrightarrow{V}⟩A_{TV}\right)$$

$$H_{ATV}=\overrightarrow{\nabla }\cdot ⟨\overrightarrow{V\mathrm{’}}\left({\displaystyle \frac{C_p}{2T_r}}T{\mathrm{’}}^2\right)⟩$$

The time evolution equation for the climatological mean kinetic energy associated with the time-mean wind speed is given as

$${\displaystyle \frac{\partial K_M}{\partial t}}=C_M-C_K-D_M-F_{KM}-H_{KM}$$

(3)

where

$$K_M={\displaystyle \frac{1}{2}}⟨\overrightarrow{V_h}⟩\cdot ⟨\overrightarrow{V_h}⟩$$

$$C_M=-⟨\omega ⟩⟨\alpha ⟩$$

$$C_K=-⟨\overrightarrow{V_h\mathrm{’}}\cdot \left(\overrightarrow{V^{\mathrm{’}}}\cdot \overrightarrow{\nabla }\right)\cdot ⟨\overrightarrow{V_h}⟩⟩$$

$$D_M=-⟨\overrightarrow{V_h}⟩\cdot ⟨\overrightarrow{F_h}⟩$$

$$F_{KM}=\overrightarrow{\nabla }\cdot \left\{⟨\overrightarrow{V}⟩\left(K_M+⟨\varphi ⟩\right)\right\}$$

$$H_{KM}=\overrightarrow{\nabla }\cdot ⟨\overrightarrow{V\mathrm{’}}\left(\overrightarrow{V_h\mathrm{’}}\cdot ⟨\overrightarrow{V_h}⟩\right)⟩$$

The time evolution equation for the time-mean kinetic energy associated with the time-variability wind speed is given as

$${\displaystyle \frac{\partial K_{TV}}{\partial t}}=C_{TV}+C_K-D_{TV}-F_{KTV}-H_{KTV}$$

(4)

where

$$K_{TV}={\displaystyle \frac{1}{2}}⟨\overrightarrow{V_h\mathrm{’}}\cdot \overrightarrow{V_h\mathrm{’}}⟩$$

$$C_{TV}=-⟨{\omega }^{\mathrm{’}}{\alpha }^{\mathrm{’}}⟩$$

$$C_K=-⟨\overrightarrow{V_h\mathrm{’}}\cdot \left(\overrightarrow{V^{\mathrm{’}}}\cdot \overrightarrow{\nabla }\right)\cdot ⟨\overrightarrow{V_h}⟩⟩$$

$$D_{TV}=-⟨\overrightarrow{V_h\mathrm{’}}\cdot \overrightarrow{F_h\mathrm{’}}⟩$$

$$F_{KTV}=\overrightarrow{\nabla }\cdot \left(⟨\overrightarrow{V}⟩K_E\right)$$

$$H_{KTV}=\overrightarrow{\nabla }\cdot ⟨\overrightarrow{V\mathrm{’}}\left({\displaystyle \frac{\overrightarrow{V_h\mathrm{’}}\cdot \overrightarrow{V_h\mathrm{’}}}{2}}\right)⟩$$

Here, T is the temperature of air;$T_r$ is a reference temperature (here, set to the maximum temperature on the Arctic ocean subdomain, $T_r=257 K$for this study); ${\overrightarrow{V}}_h$(u,v) is the horizontal wind vector; $\overrightarrow{V}$(u,v,ω) is the three-dimensional wind vector, with ω = dp/dt being the coordinate for vertical motion in pressure; Φ is the geopotential; ${\overrightarrow{F}}_h$ is the external horizontal force; Cp is the specific heat at constant pressure; α is the specific volume; and Q is the total diabatic heating rate. In the equations $\overrightarrow{{\nabla }_h}=\left({\displaystyle \frac{\partial }{\partial x}},{\displaystyle \frac{\partial }{\partial y}}\right)$ and $\overrightarrow{\nabla }=\left({\displaystyle \frac{\partial }{\partial x}},{\displaystyle \frac{\partial }{\partial y}},{\displaystyle \frac{\partial }{\partial p}}\right)$.

$A_M$and $K_M$are the climatological available enthalpy and kinetic energy, while $A_{TV}$ and $K_{TV}$ are the time averages of time-variability available enthalpy and kinetic energy. The terms G_M and G_TV represent the diabatic generation of climatological available enthalpy and the mean of time-variability available enthalpy, respectively. The terms D_M and D_TV relate to the destruction of kinetic energy through dissipation processes. I_AB is the term for converting pressure-dependent available enthalpy into climatological temperature-dependent available enthalpy, and C_M is the term for converting A_M to K_M. C_A and C_K are terms used to convert the climatological state (A_M and K_M) to perturbation (time-variability) states A_TV and K_TV, respectively. The term C_TV expresses the baroclinic energy conversions from A_TV to K_TV due to perturbation. The other terms (F_E and H_E) $Eϵ\left\{A_M,K_M,A_{TV},K_{TV}\right\}$ are boundary flux terms of energy E, exchanged between the regional domain of interest and the external environment; on a global domain, these terms would vanish, unlike on a regional domain. The positive/negative terms are sources/sinks of available enthalpy and kinetic energy.

These equations are used to establish the budget of the energy cycle and to investigate the physical and dynamic processes, in order to quantify the role of TICs in the disturbance of the energy budget of the Arctic atmosphere during the cold season.

2.2. Data Simulations

This study was conducted using three simulations of CRCM6, with and Morrison and Milbrandt’s P3 scheme [37] (see [38] for details about the model’s version), driven by an ERA5 reanalysis. The study domain is centered over the North Pole (Figure 1A,B) and includes complex topography such as the Greenland Ice Sheet, which exceeds 3 km in height and extends as far south as 50° N latitude; the domain includes parts of northern Canada, Alaska, the North Atlantic, northern Russia, and the entire Arctic Ocean.

The simulations were performed on an 1840-by-1840-point 0.03° (~3 km) grid mesh (excluding a halo zone of 10 grid points and a buffer zone of another 10 grid points). The model incorporates 62 levels in the vertical, with higher resolutions in the lower troposphere to better represent exchanges with the Earth’s surface, and a top level at 2 hPa. The initial and lateral boundary conditions (LBCs) of the atmospheric fields, together with the sea surface conditions, were provided by ERA5 Reanalysis on a 0.25° (~28 km) resolution grid. The lateral boundary conditions were updated every hour during the simulations.

The used radiation scheme was that of Li and Barker [39]. It employs a correlated k-distribution method for gaseous transmission, with nine frequency intervals for longwave and four for shortwave radiation fluxes. The prognostic total water content was used in the radiative transfer scheme. Radiation interacts with meteorological variables through the diabatic heating rate in the thermodynamic equation.

The parametric schemes used for the physical processes at the subgrid scales were the deep convection scheme of Kain and Fritsch [40], and the Kuo shallow convection scheme [41,42]. The used cloud microphysics scheme was Morrison and Milbrandt’s Predicted Particle Property (P3) scheme [37,43], with boundary-layer cloud contributions from the MoisTKE boundary layer scheme [42]. The land–surface processes were treated using the Canadian Land Surface Scheme CLASS3.5 [44,45].

The year 2007 was selected as part of the International Polar Year (IPY), since during that period, extensive studies and data analyses were conducted [46]. It also corresponds to the recent launch of two active instruments used in this study, the CloudSat radar and the CALIOP lidar as part of NASA’s A-Train. January was used as a representative month of the polar night conditions. Three simulations were performed over the Arctic regional domain over the January 2007 period; the difference between simulations was only cloud conditions for radiation calculations: one simulation without changes to the clouds (CRCM6); one with clear sky conditions (CRCM6(nocld)); and one where cloud properties such as cloud fractions, ice water content (IWC), and ice effective radius were fixed to TICs’ mean properties from observations (IWC and re) and to the CRCM6 cloud fraction (which equals 100% at this model resolution) for all atmosphere layers (see the values in

Table 1. Optical properties of the ice clouds used in each simulation.

Simulations	Cloud Fraction	Cloud Ice Effective Radius (Microns)	IWC (g/kg)	IWP(g/m2)
CRCM6(ori)	CRCM6 + P3	CRCM6 + P3	CRCM6 + P3	CRCM6 + P3
CRCM6(100%cld)	1	55	0.55	50
CRCM6(nocld)	0	0	0	0

). For diagnostic purposes, the variables were archived every 10 min and interpolated over 52 pressure levels. The energy terms were calculated using three hourly archives on 26 pressure levels.

3. Results and Analysis

We first analyzed the climatology of CRCM6’s thermodynamics variables and physical tendencies over the Arctic domain, shown in Figure 1 for the polar-night month of January 2007. We then investigated the impact of TICs on atmosphere energetic terms in the Arctic Ocean subdomain.

3.1. Arctic Thermodynamics Variables—January 2007 Means

Figure 2A,B show CRCM6-simulated mean sea-level pressure (MSLP) (Figure 2A) and surface air temperature at 2 m (Figure 2B) over the Arctic domain shown in Figure 1B for the month of January 2007. The transient-eddy standard deviations (std) are represented by white contour lines on the maps. We note that, as a general statement, the mean sea-level circulation, representative of winter in the Arctic, is dominated by the Siberian high pressure over east-central Asia and the Icelandic low pressure over the southeast coast of Greenland. The standard deviation ranges from approximately 7 to 19 hPa. We note high std values in the contour lines near the pole. For the surface temperature (Figure 2B), a strong north–south temperature gradient exists across the domain, specifically along the North Atlantic. We note dense std over land areas and sparse std over the Arctic Ocean. The std amplitudes are generally <6 °C over the Arctic Ocean and vary rapidly near the coasts of Greenland and the Canadian Archipelago due to surface process uncertainties such as those based on orography and sea ice.

Figure 2C,D show the corresponding synoptic patterns at 500 hPa and 850 hPa. The limits of the polar vortex are clearly visible at 500 hPa. The temperature advection area is clearly visible on the 850 hPa map over the Canadian Archipelago, for example. Regions where the atmosphere is baroclinic are also visible on both maps.

Figure 2E,F display the vertically integrated time-mean temperature and wind patterns, respectively, as simulated for January 2007. A strong north–south temperature gradient is seen across the domain. The wind pattern is well explained by the thermal–wind relation, with the strongest winds being located over the largest horizontal temperature gradient. January Arctic wind patterns are typically characterized by strong, persistent winds that blow around the pole. These winds are caused by the high-pressure system that typically forms over Siberia during the winter months. The basic thermodynamic variables are given in the following sections based on their influence, for easy interpretation of the energetics terms.

3.2. Contribution of CRCM6 Physical Tendencies in Total Diabatic Heating

Figure 3A–F show how the terms for physical tendencies contribute to the diabatic heating rates, as simulated through CRCM6 for the various Arctic subdomains. The results show that heating or cooling in most Arctic subdomains (Arctic, North Atlantic, Eurasia, and Greenland) is due to a strong contribution of vertical diffusion, condensation, and radiation. Note that for the Greenland subdomain (Figure 3C), the profiles below 700 hPa (in gray) are strongly influenced by the topography, which can reach 3000 m (Figure 1B). This considerably reduces the grid points used in spatial averaging. For the North Atlantic subdomain (Figure 3D), the physical tendency values can reach 10 Kday⁻¹ near the surface, explained by the influence of mid-latitude storm debris and the relatively warm temperatures in this subdomain (see Figure 2C). In the Arctic Ocean and Canadian Archipelago subdomains, radiative cooling dominates the diabatic contributions. Given the complex topography of the Canadian Archipelago subdomain, in order to study the role of clouds, we chose the Arctic Ocean subdomain, where radiation is the dominant process during the polar night, with major cloud contributions.

The vertical profile of time averages and time series at 900 hPa of diabatic heating rate terms from CRCM6 are displayed in Figure 4A and Figure 4B, respectively, for the Arctic Ocean subdomain. The total heating rate is dominated by radiative cooling. To confirm this, the time series at 900 hPa shows that even if vertical diffusion and condensation contribute mainly near the surface, these two contributions cancel each other out and radiation remains the dominant contribution to the total diabatic heating rates.

3.3. TICs’ Effect on the Temperature and Heating Rates

Figure 5 displays the time-mean, vertically integrated maps of temperature deviation (T-Tr) for our three simulations and its associated differences (Figure 5A and Figure 5B, respectively). For all simulations, due to our choice of reference temperature, the temperature deviation map (Figure 5A) is negative across the entire study domain, with smaller negative values near the right-hand side and bottom corners and decreasing values closer to the center. The trends for the three simulations are similar, but colder in the case with cloud cover (CRCM6(100%cld)) and warmer for the CRCM6 simulation. The differences between simulations (Figure 5B) show that CRCM6 is warmer, especially in the lower part of the domain, and that the difference between CRCM6 and CRCM6(nocld) is both positive (at the bottom and to the right of the domain) and negative (at the top of the domain). The difference between CRCM6(100%cld) and CRCM6(nocld), which showcases the effect of TICs on temperature, are shown in the bottom and right-hand panels of Figure 5B and is negative, with a maximum value of −4 K located north of Greenland. These low difference values between simulations CRCM6(100%) and CRCM6 show that the local change in clouds does not have a strong effect on the temperature variation, which is probably governed by large-scale dynamics.

Figure 6 presents time-mean and vertically integrated maps of the total diabatic heating rates (HRs) for the three simulations and their differences (Figure 6A and Figure 6B, respectively). The HR means (Figure 6A) are mostly negative, except for a very small area in the bottom of the domain, and the magnitude of HR values is larger in the case with cloud cover, CRCM6(100%cld). The trends for the CRCM6 and CRCM6(100%cld) simulations are similar, with uniform cooling rates near −2 K/day over the Arctic Ocean, consistent with those from Figure 4 but slightly colder in the case of CRCM6(100%cld). The CRCM6(nocld) simulation shows much lower cooling rates, with a relative warming effect with respect to the cloudy cases, as seen in the differences between simulations (Figure 6B). The presence of clouds cools most of the Arctic Ocean atmosphere, with the exception of the influence of heat exchange from the ice-free sectors of the North Atlantic, near Svalbard. The difference between CRCM6 and CRCM6(100%cld) is positive, showing that the more realistic and localized variable clouds of CRCM6, closely resembling North Atlantic storms, are more effective at cooling the atmosphere than the prescribed fixed, uniform cloud approximation. The difference between CRCM6(100%cld) and CRCM6(nocld) expresses a smoother average effect of TICs on the heating rates and is negative, with a maximum value of −1 K/day, which is comparable with that of the variable cloud case. These small differences between simulations CRCM6(100%) and CRCM6 show that for monthly time-averaged results, the cloud perturbation effect on the HR is more linear, mostly due to temperature advection and the adjustments made to fit the simulations into the ERA5 objective analysis.

Figure 7 shows comparisons of time-mean vertical profiles for CRCM6(nocld), CRCM6, CRCM6(100%cld), and CRE time-meanduring January 2007: (A) the temperature deviation and (B) the diabatic heating rates. Note that CRE (cloud radiative effect) is the difference between CRCM6(100%cld) and CRCM6(nocld). The temperature deviation profiles (Figure 7A) show that the CRCM6(100%cld) simulation is the coldest; CRCM6(nocld) is warmer; and the CRCM6 simulation lies between both profiles, below the 700 hPa pressure level. All temperature profiles are similar, with values ranging from −15 K to −5 K near the surface and reaching −46 K at 300 hPa for all simulations.

The profiles of the diabatic heating rates (Figure 7B) are generally negative throughout the vertical profile, with average values on the order of −2 Kday⁻¹, showing minimum values of −0.75 Kday⁻¹ for CRM6(nocld) and ~−2 Kday⁻¹ for CRCM6(100%cld). The CRCM6 profile lies between the other two profiles but is more similar to that of CRCM6(100%cld), implying a mean maximum CRE amplitude of about −1.25 Kday⁻¹. This cooling decreases temperatures on the order of ~−4 KMonth⁻¹. The diabatic cooling is partially compensated by the horizontal advection of temperature and adiabatic movements.

3.4. TICs’ Influence on Atmospheric Energetics

In this section, we study the energy cycle climatology on the Arctic Ocean domain to highlight TICs’ effects on dynamical and physical processes contributing to maintaining the Arctic polar night climate. Only contributions associated with the time-mean variables and time deviations are relevant.

Figure 8 illustrates the climatology energy cycle, where boxes represent various energy reservoirs in Jm² and arrows indicate energy conversions between the reservoirs, generation and dissipations, and transport terms across the domain’s lateral boundaries in W·m⁻². All terms are horizontally averaged over the domain shown in Figure 1C; vertically integrated from 950 hPa to 300 hPa; and time-averaged over the month of January 2007 for the three simulations—CRCM6(100%cld), CRCM6, and CRCM6(nocld). The vertical extension for the integration was chosen in order to focus on the weather-sensitive troposphere processes and eliminate the effects of stratospheric clouds, where the dynamics may be different from those of the troposphere.

The analysis of Figure 8 shows that the energy reservoir A_M is larger in magnitude than the other reservoirs due to the effect of the mean stratification of the Arctic atmosphere during winter polar nights. K_M is the smallest energy reservoir, thus indicating that the time-mean wind speeds in the Arctic are quite weak compared with those at mid-latitudes. Finally, the A_TV and K_TV energy reservoirs have the same magnitude, confirming the strong correlation between temperature variation and the generation of storms in the region.

Regarding the impact of clouds on energy reservoirs, clouds play a role in augmenting A_M while diminishing K_M. However, the influence of clouds on A_TV and K_TV reservoirs is not linear, and drawing definitive conclusions is challenging at this stage.

In the following sections, we analyze each reservoir and flux term’s TIC effects by focusing on the TIC effects on atmospheric energy cycle terms. The energy cycle terms can be grouped as follows:

• The four energy reservoir terms (A_M, K_M, A_TV, and K_TV);
• The term responsible for the generation of available enthalpy through diabatic processes in the time-mean state (G_M and G_TV) and for kinetic energy dissipation due to surface turbulence effects (D_KM and D_TV);
• The terms responsible for converting energy between reservoirs (C_M, C_A, C_TV, and C_K). The term I_AB, which converts available enthalpy between its pressure and temperature and will not be analyzed as it acts similarly to C_M;
• The terms responsible for transport due to limited area domains (F_AM, F_KM, F_ATV, F_KTV, H_AM, H_KM, H_ATV, and H_KTV).

3.4.1. Energy Reservoir Terms

Figure 9 presents the vertically integrated time-averaged maps of the four energy reservoirs, namely A_M, K_M, A_TV, and K_TV (note that the scale for A_M and K_M is different from that for A_TV and K_TV).

The available enthalpy of the time-averaged A_M (Figure 9A, left panel, CRCM6) shows similar trends and is proportional to the temperature deviation (Figure 5A, central panel, CRCM6). Values are largest in the upper-left sector of the study domain and in the case of the variable cloud simulation (CRCM6). The difference between CRCM6 and CRCM6(nocld) shows a negative deviation on the Northern Atlantic side (bottom of the figure) and positive trends over the Eastern Arctic (upper part of the figure), but with small amplitudes (~+4%). The difference between CRCM6 and cloudy cases (CRCM6(100%cld)) is also higher in the same region where the available enthalpy reaches its maximum, an absolute value of 14 × 10⁵ J·m⁻², with a minimum just below 2 × 10⁵ J·m⁻² located in the top right corner of the domain (Figure 9A, right-hand panel).

The time-mean kinetic energy (K_M) (Figure 9B) reflects the time-mean flow speed. The kinetic energy is higher on the right-hand side and in the bottom corners of the study domain and decreases toward to the center of the domain. The differences between the CRCM6 simulation and both the CRCM6(nocld) and CRCM6(100%cld) simulations show small positive values, reaching a maximum value of 1.6 × 10⁵ J·m⁻². However, the TIC effect on the absolute time-mean kinetic energy is weaker than that on the available enthalpy but on the same order as that of the domain-averaged values, percentage-wise (9.4% and 5.9%, respectively; see

Table 2. Vertically integrated time- and domain-averaged atmospheric energetic terms for January 2007. Energy reservoirs are shown in 10⁵ J·m⁻² and energy fluxes in W·m⁻².

Variables	CRCM6	CRCM6 − CRCM6(nocld)	(CRCM6 − CRCM6(100%cld))	CRE
AM (×105 J·m−2)	68.48	0.06	7.08	7.14	9.4%
KM (×105 J·m−2)	0.69	−0.03	−0.01	−0.04	−5.9%
ATV (×105 J·m−2)	1.80	−0.51	0.19	−0.32	−16.1%
KTV (×105 J·m−2)	4.77	0.1	−0.56	−0.46	−10.9%
GM (W·m−2)	6.19	3.33	0.62	3.95	58.0%
CM (W·m−2)	−2.77	−4.21	−17.51	−21.72	107.1%
CA (W·m−2)	1.49	0.81	−0.09	0.72	51.4%
CK (W·m−2)	−0.34	−0.36	0.37	0.01	33.3%
CTV (W·m−2)	2.42	1.92	−0.34	1.58	−76.0%

). Finally, the cloud radiative effect reduces the K_M reservoir.

The time-variability available enthalpy (A_TV) and kinetic energy (K_TV) are shown in Figure 9C and Figure 9D, respectively. First, A_TV is higher on the left-hand side and bottom corners of the study domain and decreases toward to the center. The differences between CRCM6 and the perturbed cases (center and right figures) show similar trends and have mainly negative values for A_TV. When both difference maps exhibit a consistent sign and similar magnitudes, discerning the precise influence of TICs on the variables becomes challenging. However, a more conclusive insight emerges when examining the domain averages, as illustrated in Table 2. It becomes evident that clouds exert a diminishing impact on the AT_V reservoir at a rate of ~−16%. Secondly, K_TV is higher in the center of the domain and decreases toward the corner. The differences show similar patterns and have both positive and negative values; thus, making conclusions about the effects of TICs on the K_TV reservoir becomes difficult. An important observation in this region is that K_TV is larger than K_M, contrary to what is typically observed at mid-latitudes, as noted in previous studies [21,35].

Figure 10 shows comparisons of the vertical profiles of time-mean available enthalpy (A_M) and kinetic energy (K_M), and time-variability available enthalpy energy (A_TV) and kinetic energy (K_TV) for CRCM6(100%cld), CRCM6, CRCM6(nocld), and CRE during January 2007. The A_M profiles (Figure 10A) indicate that CRCM6 and CRCM6(100%cld) have similar but slightly larger values than those of CRCM6(nocld), resulting in almost constant CRE profiles with relatively small values. These profiles follow the temperature deviation profiles (Figure 6A), with values that increase with altitude. The K_M profiles (Figure 10B) show that CRCM6(nocld) has the largest values close to the surface, resulting in negative CRE values at these altitudes. CRCM6 and CRCM6(100%cld) have very similar profiles. After the 550 hPa pressure level, the CRE sign becomes positive, but its values always range between −1 × 10³ and 1 × 10³ J·kg⁻¹.

The profiles for A_TV and K_TV (Figure 10C and Figure 10D, respectively) are similar, with larger values for CRCM6(nocld). The K_TV profiles increase with altitude, while the A_TV profiles decrease. The effect of clouds on time-variability terms is complex, as these terms are involved in the cyclogenesis of storms, during which clouds are formed. The CRE values are negative for both terms, ranging between −20 J·kg⁻¹ and 0 J·kg⁻¹. The combination of vertical profile analyses and the domain average values (see Table 2) allows us to see that TICs decrease the K_TV reservoir.

3.4.2. Energy Generation and Dissipation Terms

Figure 11 presents time-mean and vertically integrated maps of the time-mean available enthalpy generation term G_M (Figure 11A), time-variability available enthalpy G_TV (Figure 11B), and dissipation of kinetic energy D_M (Figure 11C) and D_TV (Figure 11D) for CRCM6, CRCM6-CRCM6(nocld), and CRCM6-CRCM6(100%cld).

G_M is the generation term acting on the available enthalpy reservoir A_M, is obtained from the covariance of HR (Figure 6 and Figure 7B) and temperature deviation (Figure 5 and Figure 7A), and represents the main diabatic heating component that contributes to the generation of available enthalpy. The reference temperature used in the calculation of G_M is equal to the maximum horizontal average temperature in the domain (257.49 K), so the generation term is positive across the atmospheric layer. G_M shows similar trends to HR and is higher at the top and on the left-hand side of the domain, where A_M is higher, with a value reaching 7 W·m⁻², leading to a difference between simulations with positive values and a maximum not exceeding 7 W·m⁻².

The vertical profiles of the available enthalpy energy generation term are shown in Figure 12A. The profiles for CRCM6(nocld) and CRCM6(100%cld) show smaller and higher values, respectively, with values ranging from 0 to 1.5 × 10⁻³ W·kg⁻¹. It is noteworthy that the CRE exhibits maximum values of 0.75 × 10⁻³ W·kg⁻¹, which are reached at the 450 hPa pressure level.

The terms G_TV (Figure 11B and Figure 12B), DM (Figure 11C and Figure 12C), and D_TV (Figure 11D and Figure 12D) exhibit remarkably low values within this domain, posing a challenge in assessing the impact of clouds on these terms. However, it is evident that the dissipation of kinetic energy is predominantly driven by D_TV in this region.

These results demonstrate that in the Arctic, TICs substantially contribute to increases in available enthalpy generation on a large scale; in this case, enthalpy increases by 58% compared with the case with no clouds because the presence of clouds can significantly modulate the loss of energy through outgoing longwave radiation, resulting in a general cooling effect in atmospheric temperatures (warming of the surface), which, in turn, leads to an increase in available enthalpy. For this reason, clouds play a significant role in driving the Arctic’s weather patterns and influencing the region’s climate. These findings confirm and quantitatively illustrate why the formation and dissipation of clouds are important factors modulating the Arctic’s energy budget, circulation, and ultimately, climate.

3.4.3. Energy Conversions Terms

Figure 13 presents the four main energy conversion terms, namely C_M, C_A, C_TV, and C_K, for CRCM6 and their differences from those of the CRCM6(nocld) CRCM6(100%cld) simulations in January 2007. The term C_M (Figure 13A) converts available enthalpy A_M into kinetic energy K_M. The term C_M = −〈ω〉〈α〉 is proportional to the vertical velocity and exhibits a large magnitude of up to 200 W·m⁻². C_M displays strongly negative values at the bottom and center-left of the domain, indicating time-mean upward motion of warm air and downward motion of cold air. However, positive values are observed at the center-right part of the domain. The maps of differences display both positive and negative values; thus, determining the direction of energy conversion and the effect of TICs on C_M is difficult.

The term C_A (Figure 13B) converts time-mean available enthalpy A_M into time-variability available enthalpy A_TV. C_A represents the effect of the covariance of wind and temperature perturbations interacting with the time-mean temperature gradient. The temperature and wind patterns of January 2007 (not shown transport warm air from the bottom and right corners to the center of the domain and cold air from the top and left-hand corners to the center of the study domain. Hence, the time-mean temperature gradient is reduced, thus resulting in the conversion term C_A contributing positively. The energy conversion is largest at the bottom of the domain, where the temperature gradient is the strongest, similar to that found at mid-latitudes [35]. The maps of differences show positive values and can reach 6 W·m⁻², indicating that A_M is converted into A_TV and clouds contribute to increasing energy conversion.

The term C_TV (Figure 13C) converts time-variability available enthalpy A_TV into time-variability kinetic energy K_TV, representing a baroclinic conversion due to the covariance of vertical velocity and temperature perturbations (see Figure 2C). This conversion is positive everywhere throughout the period, across the domain of interest, and for all simulations, indicating that warm (cold) anomalies are associated with upward (downward) vertical velocity anomalies. The C_TV term is largest at the bottom of the domain, where the temperature variation is the strongest. The difference maps also show positive values, suggesting that TIC presence generates large temperature and vertical velocity perturbations at all pressure levels (Figure 2C), so the effect of TICs on the conversion can reach 10 W·m⁻². If we compare the conversion terms C_TV and C_A, we note much similarity between the trends and intensity of the conversion terms for all simulations, indicating that most of the energy brought by C_A to time-variability available enthalpy A_TV is readily converted by C_TV into time-variability kinetic energy K_TV, and TICs’ presence increases this conversion.

The barotropic term C_K (Figure 13D) is the conversion of kinetic energy between its time-mean state K_M and its time-variability state K_TV and represents the effect of the variance and covariance of horizontal wind perturbations interacting with the time-mean horizontal wind gradient. The field of C_K consists mostly of negative values corresponding to the conversion from K_TV into K_M, indicating that wind perturbations contribute to reinforcing time-mean wind speeds in this case. However, in some small areas, C_K displays positive values, corresponding to the conversion from K_M into K_TV, indicating that wind perturbations contribute to reducing the time-mean wind speed there. Both difference maps show similar trends to those of CRCM6, so TIC presence also increases this conversion.

Figure 14 shows a comparison of the time-mean vertical profiles of the main energy conversion terms C_M, C_A, C_TV, and C_K for CRCM6(100%cld), CRCM6, CRCM6(nocld), and CRE during January 2007. The C_M profiles (Figure 14A) indicate that below the 800 hPa pressure level, all simulations have negative values. On the other hand, above this level, the profiles of CRCM6 and CRCM6(nocld) become positive, generating a profile with negative values (ranging from −1.75 × 10⁻² to −0.25 × 10⁻² W·kg⁻¹) for the contribution of the clouds, CRE. The profiles for C_A (Figure 14B) and C_TV (Figure 14C) are similar and have positive values over the entire atmospheric column and for all simulations. However, C_TV has slightly larger values. In both cases, TICs reinforce the conversions. Finally, the C_K profile (Figure 14D) shows positive values below 600 hPa, which are equal for all simulations, and negative values above for the CRCM6 and CRCM6(nocld) simulations but positive values for the CRCM6(100%cld) simulation. Therefore, the CRE is zero below 600 hPa and positive elsewhere.

3.4.4. Boundary Flux Transport Terms

Figure 15, Figure 16, Figure 17 and Figure 18 present the eight transport terms at lateral boundaries, namely F_AM, H_AM, F_KM, H_KM, F_ATV, H_ATV, F_KTV, and H_KTV, for CRCM6 and their differences from those of the CRCM6(nocld) CRCM6(100%cld) simulations in January 2007. In the CRCM6 simulation, the terms F_AM (Figure 15A), H_AM (Figure 17A), and H_KM (Figure 17B) present particularly high positive values, reaching 40 W·m⁻² for F_AM, especially in the lower part of the domain. These high values suggest strong advection of the time-mean enthalpy A_M and kinetic energy K_M outside the domain. The difference maps further illustrate positive values for F_AM and H_KM and a mixture of positive and negative values for H_AM over the entire domain, indicating intensified advection of A_M and K_M influenced by the imposed presence of TICs. An intriguing aspect lies in the complexity of interpreting the impact of TICs on these terms. The direction of the effect of TICs on F_AM, H_AM, and H_KM remains uncertain, which may be attributed to the lateral boundary conditions, a factor that plays a role in maintaining model equilibrium, thus adding a layer of complexity to our understanding of how TICs influence these boundary terms. The terms H_ATV (Figure 17C) and H_KTV (Figure 17D) present a mix of positive and negative values in the same order of magnitude as the conversion terms C_A and C_TV, but the alternating signs make them weak in terms of domain averages and TIC effects. All other boundary flux terms, namely F_KM (Figure 15B), F_ATV (Figure 15C), and F_KTV (Figure 15D), are smaller; the generations/conversions and processes resulting in their variations are masked by terms of greater orders of magnitude.

Figure 16 and Figure 18 provide a comprehensive comparison of time-averaged vertical profiles for the eight transport terms at lateral boundaries during January 2007 for CRCM6(100%cld), CRCM6, CRCM6(nocld), and CRE. Examining the F_AM profiles (Figure 16A) shows that above the 900 hPa pressure level, all simulations exhibit positive values. However, the CRE profile shows very small values, underscoring the considerable challenge in interpreting the effect of TICs on this particular term.

Regarding F_KM (Figure 16B), F_ATV (Figure 16C), and F_KTV (Figure 16D), these profiles present a discernibly disorganized nature, with a change in sign occurring around the 700 hPa level, complemented by a low value in the CRE profile. Similarly, the profiles for H_AM (Figure 18A), H_KM (Figure 18B), H_ATV (Figure 18C), and H_KTV (Figure 18D) also exhibit pronounced disorganization (probably due to the fact that different types of circulation enter and leave the domain), with a sign change occurring near the 700 hPa level. Notably, strong values are observed for H_KM and HKTV near the surface. The CRE profile closely follows that of CRCM6(100%cld).

4. Summary and Conclusions

The aim of this study was to quantify the influence of TICs on the Arctic atmospheric energy budget during the polar night and to enhance understanding of the mechanisms underlying their radiative and dynamic effects. The regional energetic equations formulated by Nikiema and Laprise [21] and adapted by Clément et al. [35] were applied to three reanalysis-driven simulations of the CRCM6 (named CRCM6, CRCM6(nocld), and CRCM6(100%)) for the well-documented International Polar Year (IPY-2007) January 2007 using the P3 cloud microphysical scheme. The atmospheric variables and energetics terms from these three simulations were compared. A physical interpretation of the energy budget cycle was analyzed while quantifying the effect of TICs.

Spatial patterns of MSLP, temperature, and 500 hPa geopotential from CRCM6 were analyzed to contextualize the circulation regimes and thermal structure over the study domain. The vertical profiles revealed the position of the polar front, baroclinic zones, and regions of warm and cold advection.

Radiative heating rate diagnostics show that TICs induce a net cooling of approximately −2 K day⁻¹, corresponding to monthly mean cooling of about −4 K. This diabatic cooling is partially offset by horizontal advection of temperature and adiabatic subsidence heating. The main contributors to diabatic processes vary spatially: vertical diffusion, condensation, and infrared radiation dominate over land and sea ice, whereas over the Arctic Ocean, infrared radiative cooling associated with cloud emission is the prevailing term. The enhanced condensation at cloud tops increases precipitation and reduces precipitable water, demonstrating the radiative–microphysical coupling induced by TICs.

From the point of view of atmospheric energetics in the Arctic domain, the climatological available enthalpy reservoir A_M dominates over other energy reservoirs. A_M is mostly imported by the time-mean boundary flux contribution I_AB and exported by the conversion term C_M to the time-mean kinetic energy reservoir K_M. The available enthalpy diabatic generation G_M also makes a substantial contribution to A_M. The largest A_M value is obtained from the CRCM6(100%cld) simulation. The positive covariance between radiative heating (HR) and temperature anomalies increases G_M by about 5 W·m⁻², showing that thin clouds act as a source of available enthalpy during the polar night.

The advection term C_A converts A_M into time-variability available enthalpy (A_TV), while diabatic generation of A_TV (G_TV) remains weak. Losses in A_TV are almost entirely transferred to time-variability kinetic energy (K_TV) via the baroclinic conversion term C_TV, with a secondary contribution from barotropic conversion (C_K) from K_M. Frictional dissipation (D_TV) acts mainly through cyclolysis, whereas mean kinetic energy dissipation (D_KM) remains small. Overall, TICs enhance energy conversion efficiency, particularly the transfer from enthalpy to kinetic energy.

TICs exert a significant influence on enthalpy and kinetic-energy variability during the Arctic winter by trapping longwave radiation near the surface and amplifying baroclinic activity. The four principal conversion terms—C_M, C_A, C_TV, and C_K—all indicate enhanced energy conversion in the presence of TICs. These findings suggest that thin ice clouds contribute meaningfully to the maintenance of the Arctic energy cycle and can modulate atmospheric dynamics during the polar night.

Our modeled cloud-radiative effects align with recent observational analyses [47,48], which show systematic increases in downwelling longwave radiation due to thin cirrus. This agreement strengthens the credibility of the simulated responses. In a warming Arctic, increased humidity could foster the formation of thicker and more persistent ice clouds, altering the radiative balance, while changes in atmospheric circulation may reduce TIC frequency through enhanced poleward heat transport. Future work should therefore couple energetic diagnostics with climate projections to assess how TIC-related feedbacks evolve under continued anthropogenic forcing.

Finally, following Veiga and Ambrizzi [49], a comparison between current and future polar-night energy budgets would provide insight into how available potential energy generation and baroclinic conversions may weaken or intensify under global warming.