Impacts of Land–Atmosphere Interactions on Boundary Layer Variables: A Classification Perspective from Modeling Approaches

Abstract

Previously, the types of impacts of land–atmosphere interactions have scarcely been clarified systematically. In this article, we present a classification of these impacts based on modeling boundary layer variables/parameters, which is grouped into local, regional, and remote impacts. In the narrow sense, land surface processes (LSPs) influence the atmospheric state via vertical land–atmosphere coupling at local scales, which is referred to as local LSP impacts. However, local LSP impacts can lead to the advection effect due to the horizontal heterogeneity in the parameters over a region, which can be defined as regional LSP impacts. Furthermore, remote LSP impacts on the regional atmospheric state are induced by some land/sea surface variables/parameters over remote key areas of the Earth’s surface, which are conventionally taken as strong signals of climate variation. Of the three impacts, local impacts are the most important essential, as the other two types of impacts are derived from these impacts. We describe the quantification of local impacts based on our previous studies from the perspective of modeling approaches, and we discuss some issues related to these impacts. Previous investigations showed that local LSP impacts are mostly stronger than regional LSP impacts, e.g., the diabatic process is dominant in the physical processes responsible for daily maximum temperatures, and two first-order physical processes including vertical diffusion largely induce changes in surface wind speed in China. Finally, some aspects for future research are noted. This study provides insights into the research on land–atmosphre interactions at different scales.

1. Introduction

Under the background of global warming, various extreme events have occurred frequently in recent decades, such as high temperatures, heavy precipitation, and compound drought and heatwave events [1,2,3,4,5,6]. These extreme events have a strong impact not only on the natural environment but also on the economy and human life. Therefore, quantifying and explaining the processes that influence these extremes can help with predicting the events and effectively preventing the associated damage.

One of the most effective methods of predicting these extremes is based on numerical weather or earth system models, which use the fluid dynamics approach with a set of primitive equations to describe the state of the atmosphere [7]. The full terms of these equations include various physical, chemical, and biogeochemical processes, some of which must be parameterized because these processes cannot be resolved by the prescribed model’s spatial scales of grids or spectra.

Among the parameterizations of regional and global models, land–atmosphere interactions are of great importance [8,9]. It has been found that natural and anthropogenic factors are two types of causes of global warming over the long term (https://www.epa.gov/climatechange-science/causes-climate-change; accessed on 28 May 2024), and, over a much shorter period, land surface processes (LSPs) can significantly affect weather or climate variability [8,9].

Broadly, land–atmosphere interactions can be defined as feedback between the land and the atmosphere that affects the transport of the momentum, energy, and mass of various chemical species on the Earth’s surface [8,9]. This feedback has a significant influence on the near-surface atmospheric state at different spatiotemporal scales. In atmospheric modeling over meteorological and climate time scales, LSPs are parameterized. These parameterizations are referred to as land surface schemes (LSSs) or models, e.g., there are seven LSSs in the Weather Research and Forecasting Model (WRF), version 4.5, such as the SLAB, Noah, RUC, Noah-MP, and CLM4 LSSs (https://www2.mmm.ucar.edu/wrf/users/wrf_users_guide/build/html/physics.html#surface-physics; accessed on 28 May 2024). Through land–atmosphere coupled model simulations and projected climate changes, a great number of investigations have been conducted, e.g., on the effects of soil moisture on climate through soil moisture–temperature, soil moisture–evapotranspiration, and soil moisture–precipitation feedback, as well as vegetation feedback in various regions around the globe, according to the climate regimes for the region under study [10,11,12,13,14,15,16,17,18,19,20,21].

As a key interface of the components of the climate system (i.e., atmosphere, biosphere (life), cryosphere (ice), hydrosphere (water), land, and humans https://scied.ucar.edu/learning-zone/earth-system/climate-system; accessed on 28 May 2024), the land surface has significant impacts on various variables/parameters at the land surface, e.g., moisture and temperature of surface soil and air, over local, regional, and continental spatial scales. However, LSSs are generally vertically coupled with atmospheric modules in one dimension [8,9,22]. Thus, the direct effect of land–atmosphere interactions has been simulated at local scales, i.e., a relatively homogeneous computational unit such as a grid cell in a model domain [8,9,22]. In the context of atmospheric theories, more effects can be predicted over much larger areas; currently, the classification of impacts of land–atmosphere interactions has rarely, if at all, been addressed systematically. Hence, the purpose of this short opinion article is to present some of our views on the classification of the impacts of land–atmosphere interactions on boundary layer variables/parameters from the perspective of modeling approaches and on some issues related to quantifying the specific types of impacts.

2. Methodology

In this section, we address approaches for quantifying the impacts of land–atmosphere interactions on surface air temperature (SAT), wind speed, and geopotential height (GPH) in the narrow sense of reflecting turbulent vertical transport at local scales, through which local LSP impacts are introduced.

2.1. Local Impact on Surface Air Temperature

In primitive equations, the SAT tendency can be derived from the first law of thermodynamics, which is expressed as [5,23]

$$\frac{\partial T}{\partial t}=-V\cdot \nabla T-w({\gamma }_d-\gamma )+H_t$$

(1)

where V is the horizontal wind vector; $w$ is the vertical velocity; ${\gamma }_{\mathrm{d}}$ and $\gamma$ represent the dry adiabatic lapse rate and the temperature lapse rate of the atmospheric layer, respectively; and $H_t$ is the diabatic heating. This equation indicates that the temporal variation in SAT is determined by three terms, i.e., the advection ($-V\cdot \nabla T$), convection (or adiabatic; $-w({\gamma }_{\mathrm{d}}-\gamma )$), and diabatic ($H_t$) terms, with the last term containing the effects induced by radiation, as well as turbulence energy, and phase-change energy transfer.

The impact of land–atmosphere interactions on SAT is attributed to the vertical transport of energy and mass into the surface air column, as induced by various surface fluxes, i.e., the fluxes of radiation, sensible and latent heat, and moisture, which are included in the diabatic term $H_t$ in (1) and involve turbulent processes [5].

It is well-known that in fluid dynamics, computing the terms of turbulence processes is difficult because of the closure problem of turbulence [7]. In atmospheric modeling, parameterizations for planetary boundary layers have been applied [7]. However, in current atmospheric models such as global models (e.g., the Earth System Model; https://soccom.princeton.edu/modeling/what-earth-system-model-esm; accessed on 28 May 2024) or regional models (e.g., the Weather Research and Forecasting (WRF)-Advanced Research WRF (ARW) model, www.wrf-model.org, accessed on 6 April 2024), $H_t$ in (1) is very complexly parameterized with associated physical processes and is not a term with a direct model output, while the other three terms in (1) (i.e., SAT tendency and the advective and convective terms) can be calculated using model outputs. Therefore, Zeng et al. [5,23,24] directly calculated the three terms; the contribution of $H_t$ to changes in SAT can be obtained by considering Equation (1) as a residue. In other words, the influence of the feedback between the land and atmosphere on the SAT at local scales (i.e., local LSP impact) is then quantified. Because all the terms of the contributions to variations in SAT can be quantified, the relative importance of each of the physical processes in SAT variations is clarified.

2.2. Local Impact on Surface Wind Speed

Similarly, derived from the momentum balance equation [7], the equation of the integral terms in the $\sigma$ coordinate system for local wind speed change (V_t), advection (ADV), pressure gradient force (PRE), convection (CON), and turbulent diffusion (DFN) can be written as [25]

$$V_t=\mathrm{A}\mathrm{D}\mathrm{V}+\mathrm{P}\mathrm{R}\mathrm{E}+\mathrm{C}\mathrm{O}\mathrm{N}+\mathrm{D}\mathrm{F}\mathrm{N}$$

(2)

In (2), the 5 terms are as follows:

$$V_t={\displaystyle \underset{t}{\int }\frac{\partial V}{\partial t}}dt,$$

(3)

$$\mathrm{ADV}=-{\displaystyle \underset{t}{\int }(u\frac{\partial V}{\partial x}+v\frac{\partial V}{\partial y})}dt,$$

(4)

$$\mathrm{PRE}=-{\displaystyle \underset{t}{\int }\frac{1}{V}(\alpha u\frac{\partial p}{\partial x}+\alpha v\frac{\partial p}{\partial y}+u\frac{\partial \Phi }{\partial x}+v\frac{\partial \Phi }{\partial y})dt},$$

(5)

$$\mathrm{CON}=-{\displaystyle \int \stackrel{·}{\sigma }\frac{\partial V}{\partial \sigma }dt},$$

(6)

$$\mathrm{DFN}={\displaystyle \underset{t}{\int }\frac{\mathit{V}\cdot \mathit{F}}{V}}dt,$$

(7)

where $u$ and $v$ are the zonal and meridional wind speeds, respectively; V is the full wind speed ($V=\sqrt{u^2+v^2}$); p is pressure; $\dot{\sigma }$ is the vertical velocity of the $\sigma$ coordinate system; $\alpha$ is the specific volume; $\mathsf{\Phi }$ is the geopotential; and F is the friction.

Because the DFN cannot be computed with the outputs from the model [25], it is also obtained by calculating Equation (2) after directly calculating the four terms other than the DFN. Hence, the relative contribution of DFN to wind speed changes can be quantified, the influence of land–atmosphere feedback on wind speed at local scales is then simulated.

2.3. Local Impact on Geopotential Height

Geopotential height (GPH) is one of the most fundamental elements and is conventionally used to describe atmospheric patterns in meteorological/climatological research and applications. Using the hydrostatic equation and the equation of the state for air, the finite differential forms of GPH changes at 850 and 500 hPa can be, respectively, rewritten as follows [26]:

$$\delta H_{850}\approx H_{850}\frac{\delta \overline{T}}{\overline{T}}+6.15H_{850}\frac{\delta p_s}{p_s}$$

(8)

and

$$\delta H_{500}\approx H_{500}\frac{\delta \overline{T}}{\overline{T}}+1.44H_{500}\frac{\delta p_s}{p_s}$$

(9)

where $H_{850}$ and $H_{500}$ are 850 and 500 hPa GPH, respectively; $\overline{T}$ is the averaged air temperature of the reference air column; and $p_{\mathrm{s}}$ stands for surface air pressure.

In Equations (8) and (9), the influence of LSP on the GPH at local scales is determined by diabatic processes (e.g., turbulence and radiation) that act on $p_{\mathrm{s}}$ and $\overline{T}$; i.e., land surface thermal and dynamical perturbations induce changes in $p_{\mathrm{s}}$ and $\overline{T}$. Therefore, by computing the changes in $p_{\mathrm{s}}$ and $\overline{T}$, the change in GPH can be obtained.

Using WRF ensembles for the 2003 summer simulation, the results indicated that the sensitivity of the simulated GPH in the boundary layer (i.e., at 850 hPa,) to different LSSs was found to be higher than that in the mid-troposphere GPH (i.e., 500 hPa) because the LSS choice largely influences the simulated GPH fields directly by modifying surface fluxes in the lower atmosphere rather than in the upper atmosphere, and the land perturbations could substantially influence regional circulation patterns [26].

Overall, we suggest that, in a narrow sense, the above-mentioned impacts of land–atmosphere interactions via turbulent vertical transport at local scales could be defined as the local LSP impact, with the support of numerical experiments in the associated investigations [5,23,24,25,26]. Meanwhile, in the broad sense, local LSP impacts can induce larger LSP impacts in larger areas, as discussed in the following section.

3. Discussion

3.1. Regional and Remote Impacts

Broadly, land–atmosphere interactions can exert much more influence on the atmospheric state through various processes in much larger areas, which is theoretically demonstrated below.

Thermal wind ($V_T$) is expressed as

$$V_T=-\frac{R}{f}\ln \frac{p_1}{p_2}{\nabla }_p\overline{T}\times k$$

(10)

where $\overline{T}$ is the average air temperature between the $p_1$ and $p_2$ pressure levels, $R$ is the specific gas constant, and f is the geotropic constant. This clearly indicates that when SAT variability is induced by LSPs (cf. Equation (1)), thermal wind is generated, and wind is further added to the well-known geotropic wind ($V_g$):

$$V_g=-\frac{1}{f\rho }\nabla hp\times k$$

(11)

where $\rho$ and $p$ are the air density and pressure, respectively. In other words, pressure gradient force is induced, which can influence the wind and then the advection terms in Equations (1) and (2).

In short, local-scale (e.g., a grid cell) LSPs can induce horizontal variabilities as reflected by the effects of advection that are generally represented by the difference between the adjacent grid cells in numerical models. i.e., single grid cells over a region. This type of impact of LSPs can be grouped into regional impacts instead of local impacts.

Here, regional LSP impacts are defined by their relatively close relationship with the advection effect of the corresponding parameters/variables, in particular in the regions that have a spatial circulation pattern clearly favoring advection, e.g., heterogeneity in the local LSP impacts of solar radiation can cause SAT heterogeneity, with the latter further inducing temperature advection over a region, i.e., local impacts can induce regional impacts.

In contrast with local and regional impacts [5,23,24,25,26,27,28,29,30,31,32,33,34], there might be a remote correlation between a strong signal and the climate state in a region [35,36,37,38,39,40,41,42]; e.g., strong correlations between predicted and observed NIN03 SST anomalies were found with a lead time over 10 months [35]. The Western Pacific Warm Pool Area index was found to be a major indicator of summer precipitation in the upper catchment of the Three Gorges Dam [40]. The snow cover on the Tibetan Plateau and Lake Baikal can intensify the North Atlantic Oscillation in winter [36], and resonance was observed between the projected future Tibetan Plateau surface darkening and the Arctic climate [41]. These remote correlations can be explained by the propagation of Rossby waves [41,42], in which no advection effect can be attributed because the mechanism is initiated by waves rather than direct downstream air flows. In this context, this type of LSP impact is referred to as a remote impact, which could strongly influence the climate in a region generally far away from the source of the vital signals.

Overall, we propose grouping LSP impacts into three types, i.e., local, regional and remote impacts, where remote LSP impacts serve as external forcings, while the regional and local LSP impacts could be two of the key factors responsible for regional and local climate/weather, respectively.

3.2. Some Issues Associated with Local LSP Impacts

Strictly speaking, in numerical modeling, local LSP impacts can be quantified only by “offline” experiments [8]. In “online” or coupled model simulations, the advection effect is inevitable, which induces regional LSP impacts. In some cases where the advection effect is very weak, local LSP impacts may be very pronounced, while the regional impacts of the LSP may be negligible [43].

Because of the importance of LSP impacts, the terms in the equations for variable changes (e.g., $H_t$ in Equation (1) and DFN in Equation (2)) are two of the most significant terms in balancing the other physical terms (e.g., in the high-temperature weather event that occurred in eastern China); the local LSP impact (i.e., the diabatic term) is the dominant process responsible for the daily maximum temperatures and is balanced by the adiabatic term that is associated with vertical movement [5]; regarding the summer 10 m wind speed in mainland China, its change is largely dominated by two first-order physical processes, i.e., by the terms associated with the pressure gradient and diffusion [25].

Notably, the distinction between local and regional LSP impacts is unclear in some cases. For example, a dozen climate modeling groups found hot spots (i.e., regions) of the impact of soil moisture on precipitation around the globe, in which the hot spots can be taken as the result of regional LSP impacts. However, because many groups identified these hot spots using the same highly controlled numerical experiment, in which the advection effect is assumed to be basically removed, the hot spots predominantly exhibited soil moisture–precipitation coupling at local scales. Similarly, a case was described by Zhao et al. [34], which indicated the effects of Lake Nam Co and the surrounding terrain on the extreme precipitation on the Tibetan Plateau. These results suggest that local LSP effects present common or integrated characteristics in some regions.

It should be noted that the magnitude of local LSP effects is not consistent with the sensitivity of the climate to land–atmosphere interactions. Strictly speaking, higher fluxes at the surface correspond to stronger land–atmosphere interactions. However, large surface fluxes generally do not necessarily lead to atmospheric states that are sensitive to the changes in land surface states; e.g., the SAT shows relatively low sensitivity to changes in soil moisture in regions with a humid climate [22]. Hence, strong land–atmosphere coupling was found over some transient climate zones [1].

Additionally, when computing the term for the local impact (e.g., $H_t$ in (1)), the magnitude of the term might not agree well with the land surface fluxes. For example, many investigations indicated that a higher sensible flux corresponded well to a higher SAT. However, this is not always the case; i.e., a higher sensible flux might correspond to a lower SAT [26,44]. This inconsistency is due to the fact that the term $H_t$ is computed for the unit mass within an atmospheric layer rather than at the land surface [5,26]; i.e., $H_t$ is in agreement with the convergence of sensible heat fluxes of the atmospheric layer rather than the only sensible heat flux at the land surface.

We developed a classification for the impacts of land–atmosphere interactions on boundary layer variables from the perspective of modeling approaches, i.e., mainly over mainland China in the context of atmospheric dynamics. In fact, there have been a large number of investigations concerning the land–atmosphere interactions in various regions around the globe, which have taken into account the coupling between soil moisture and temperature, evapotranspiration and precipitation, as well as climate–vegetation feedback either through remote sensing or numerical experiments. For example, Koster et al. [1] indicated the regions around the globe that experience strong coupling between soil moisture and precipitation using numerical experiments as part of a coordinated comparison project. Zeng et al. [5] investigated the soil moisture–temperature coupling for a hot weather event in east China using a series of forecast lead times from the short range to the medium range. Yu et al. [11] investigated the impact of climate–vegetation interactions on future climate changes over West Africa using a regional climate model with synchronous coupling between climate and natural vegetation. Wang et al. [38] quantified evapotranspiration’s contribution to summer precipitation over the mid-lower Yangtze River Basin, China, using reanalysis data. Zhang et al. [15] assessed land–atmosphere coupling using soil moisture from Global Land Data Assimilation System and observational precipitation, while Hirschi et al. [45] analyzed soil moisture–temperature coupling around the globe using remote-sensed soil moisture. Regional and remote LSP impacts have generally been included in the above investigations.

4. Conclusions and Future Directions

This paper presents a classification of the impacts of land–atmosphere interactions on boundary layer variables from the perspective of modeling approaches. The impacts of land–atmosphere interactions can be divided into three groups: local, regional, and remote impacts. Narrowly, the impact of land–atmosphere interactions are induced at local scales (e.g., for separate grid cells in computation), which is referred to local LSP impacts. Due to the heterogeneity induced by the local impacts within a region, the advection effect is then caused, which is referred to as regional LSP impacts. Additionally, remote LSP impacts are defined as the influence induced by patterns of the parameters/variables (e.g., SST and snow cover) at the Earth’s surface within certain vital areas of a target region far from the target region.

In the context of the driving mechanisms, local LSP impacts further lead to regional and remote LSP impacts through the advection process and Rossby wave propagation, respectively; remote LSP impacts could serve as external forcings, while regional and local LSP impacts could be two of the critical factors responsible for regional and local atmospheric states, respectively.

We described how to quantify local LSP impacts for SAT, surface wind speed, and GPH in the atmospheric boundary layer in the context of atmospheric dynamics. Based on the primitive equations and the theories of geotropic wind and thermal wind, we also explained why regional LSP impacts were induced, and we clarified some issues in LSP impacts; i.e., in some cases, the distinction between local and regional LSP impacts is unapparent, the local effect is inconsistent with the sensitivity of the climate to land–atmosphere interactions, and the magnitude of the term might not consistently agree well with land surface fluxes.

For future studies, we suggest that more attention be paid to the long-term examination of the impacts of LSP on more parameters/variables, the relationship between local and regional impacts, and how remote LSP impacts regional and local parameter/variable changes. Additionally, these LSP impacts on extreme events in the historical climate and future scenarios could be a meaningful topic in follow-up studies.