Progress in Dust Modelling, Global Dust Budgets, and Soil Organic Carbon Dynamics

Dust emission is an important corollary of the soil degradation process in arid and semi-arid areas worldwide. Soil organic carbon (SOC) is the main terrestrial pool in the carbon cycle, and dust emission redistributes SOC within terrestrial ecosystems and to the atmosphere and oceans. This redistribution plays an important role in the global carbon cycle. Herein, we present a systematic review of dust modelling, global dust budgets, and the effects of dust emission on SOC dynamics. Focusing on selected dust models developed in the past five decades at different spatio-temporal scales, we discuss the global dust sources, sinks, and budgets identified by these models and the effect of dust emissions on SOC dynamics. We obtain the following conclusions: (1) dust models have made considerable progress, but there are still some uncertainties; (2) a set of parameters should be developed for the use of dust models in different regions, and direct anthropogenic dust should be considered in dust emission estimations; and (3) the involvement of dust emission in the carbon cycle models is crucial for improving the accuracy of carbon assessment.


Introduction
The aeolian and fluvial processes play a fundamental role in earth systems and have important environmental and ecological effects at both local and global scales [1]. Wind erosion is a natural geological process involving the detachment, transport, and deposition of soil particles by strong winds [2][3][4][5], and it is a key soil degradation process in arid and semi-arid areas worldwide [6][7][8][9]. In contrast to water erosion, where the eroded material follows determined paths, wind-eroded material is widely dispersed over the landscape [10]. The mineral dust generated by soil particle emissions, in turn caused by wind erosion, is considered the most important source of atmospheric aerosols [11]. The global annual emission amount of mineral dust due to wind erosion is estimated to be around 1 to 5 billion (10 9 ) tons [11][12][13][14], which account for approximately 30-50% of the total aerosol introduced into the atmosphere [15]. Dust aerosols play important roles in regulating the Earth's radiation budget, climate, global biogeochemical cycles, terrestrial soil formation, air quality, and human health [16][17][18][19][20][21][22][23][24][25][26][27][28][29][30].
To assess the socio-economic and environmental effects of dust processes, it is essential to quantify the dust emission rates at different spatial and temporal scales. Dust emission involves complex interactions among soil properties, climate, vegetation, and land use regimes. The understanding of dust processes and the capability of dust emission models have improved considerably over the past five decades. Based on the measured physical properties of dust emission at the field scale, several approaches have been adopted to estimate dust emission rates, such as mathematical simulation using data on the relationships between meteorological records and interacting surface parameters [31], remote sensing [32,33], and using geographic information systems (GIS) [34][35][36][37]. Numerous dust models have been developed to quantify dust emission rates and soil losses in the field [38,39], regional [40,41], continental, and global scales [42,43].
Soil is the main terrestrial reservoir of organic carbon and contributes substantially to the global carbon cycle [1,44] . Small changes in the soil organic carbon (SOC) stock may result in large changes in atmospheric carbon dioxide (CO 2 ) concentration [4,45]. Dust emission is an essential component of the carbon budget; it removes carbon from vast areas and, if the wind is strong enough, readily transports carbon dust offshore [46,47]. Thus, soil redistribution through dust mobilisation is an important mechanism underlying carbon cycling in terrestrial ecosystems, the atmosphere, and oceans. The active component of SOC and the organic carbon combined with the fine fraction of the soil are easily removed from terrestrial ecosystems via dust emission [48]. Wind-driven mobilisation of carbon augments the net loss of carbon from terrestrial systems.
In this review, we discuss empirical and physical dust models at multiple spatial scales, developed worldwide over the past five decades; the effects of dust emission on global dust budgets and SOC dynamics; and the link between dust processes and the global carbon cycle.

Factors Influencing Dust Emissions at Multiple Scales
Dust emission is a dynamic natural process regulated by complex interactions among the climate, soil properties (grain size, aggregation, structure, moisture, and surface roughness), vegetation (cover, distribution, and height), and land use at different spatial and temporal scales [3,34,[49][50][51][52][53][54]. This process is recognised as a major source of uncertainty in climate models [55,56].
Dust emission is essentially a flow process in which soil is detached from an erodible surface and transported in various ways (surface creep, saltation, and suspension) in response to wind shear stress ( Figure 1) [57]. Dust transport mechanisms redistribute soil and associated nutrients and organic materials at different spatial scales ( Figure 1). The mechanism by and the distance to which soil particles are transported are determined by their size. Large (>500 µm) and medium-sized (100-500 µm) particles are more likely to be transported via surface creep and saltation, respectively, over relatively short distances; smaller particles (<100 µm) can be transported via suspension over longer distances, across regions, continents, and the world [32,[57][58][59][60][61]. The development of dust models requires an understanding of the factors affecting dust emission at different spatial scales. At the grain scale (<10 −2 m), dust emission is controlled by wind shear speed and the structure, texture (particle size distribution), moisture content, mineral composition, electrostatic forces, chemistry, and microbiota composition of the soil ( Figure 1) [57,[62][63][64][65][66][67][68][69][70]. Together, these factors determine the weight, drag, and interparticle cohesion of soil aggregates and threshold friction velocity (u*t) [57]. The u*t, which controls both frequency and intensity of erosion events, is the minimum friction velocity required to initiate the movement of soil particles, representing the strength of forces among the soil particles and the capacity of an aeolian surface to resist wind erosion [60,[71][72][73]. This crucial parameter controls the frequency and intensity of dust emission. Soil erodibility is defined as the susceptibility of soils to detach and transport by erosive agents, namely water or wind. Soil erodibility is also dependent on the intrinsic properties of soils (include texture, mineralogy, chemistry, and organic matter content) and the combined influence of temporal soil properties, namely moisture, aggregation, surface crusting, and the availability of loose erodible material [57,66,[68][69][74][75][76][77]. At the field scale, the grain-scale conditions of soil texture, soil moisture, and inter-particle bonding control soil aggregation and crusting, and thus, influence soil particle movement and the potential for dust emission [74,75]. Aggregation and crusting affect soil surface roughness, u*t, and the availability of loose erodible soil particles. The latter parameters affect soil erodibility at the landscape (10 3 m) scale [62].
At the landscape scale, dust emission is determined by soil type, vegetation cover, cultivation practices, soil surface roughness, u*t, and the availability of loose erodible material [57]. However, at the regional to global scales (>10 4 m), the transport, transformations, and deposition of dust particles, and their chemical reaction with air pollutants are affected by soil type, landforms, climate, ecoregional environmental conditions, and practices of land use and land management [76]. Together, these factors determine the relative influence of soil moisture, aggregation, and crusting on the soil surface, as well as the spatial and temporal variations in soil erodibility at the field scale. The regional The development of dust models requires an understanding of the factors affecting dust emission at different spatial scales. At the grain scale (<10 −2 m), dust emission is controlled by wind shear speed and the structure, texture (particle size distribution), moisture content, mineral composition, electrostatic forces, chemistry, and microbiota composition of the soil ( Figure 1) [57,[62][63][64][65][66][67][68][69][70]. Together, these factors determine the weight, drag, and interparticle cohesion of soil aggregates and threshold friction velocity (u*t) [57]. The u*t, which controls both frequency and intensity of erosion events, is the minimum friction velocity required to initiate the movement of soil particles, representing the strength of forces among the soil particles and the capacity of an aeolian surface to resist wind erosion [60,[71][72][73]. This crucial parameter controls the frequency and intensity of dust emission. Soil erodibility is defined as the susceptibility of soils to detach and transport by erosive agents, namely water or wind. Soil erodibility is also dependent on the intrinsic properties of soils (include texture, mineralogy, chemistry, and organic matter content) and the combined influence of temporal soil properties, namely moisture, aggregation, surface crusting, and the availability of loose erodible material [57,66,68,69,[74][75][76][77]. At the field scale, the grain-scale conditions of soil texture, soil moisture, and inter-particle bonding control soil aggregation and crusting, and thus, influence soil particle movement and the potential for dust emission [74,75]. Aggregation and crusting affect soil surface roughness, u*t, and the availability of loose erodible soil particles. The latter parameters affect soil erodibility at the landscape (10 3 m) scale [62].
At the landscape scale, dust emission is determined by soil type, vegetation cover, cultivation practices, soil surface roughness, u*t, and the availability of loose erodible material [57]. However, at the regional to global scales (>10 4 m), the transport, transformations, and deposition of dust particles, and their chemical reaction with air pollutants are affected by soil type, landforms, climate, ecoregional environmental conditions, and practices of land use and land management [76]. Together, these factors determine the relative influence of soil moisture, aggregation, and crusting on the soil surface, as well as the spatial and temporal variations in soil erodibility at the field scale. The regional climate, other ecoregional conditions, and land use practices may, in turn, be affected by dust transport and deposition. This interdependence generates feedback that affects soil erodibility at various scales, from the landscape to the microscopic [57].

Dust Models at Multiple Spatial and Temporal Scales
To understand the role of dust in the earth system, numerous models that simulate dust emission at various spatial and temporal scales have been developed since the 1960s [38,[78][79][80][81]. Most of these models are used to predict dust emission rates. Dust models can be divided into empirical and physical types [7]. Empirical models are based on functions derived from field or wind tunnel experiments under a wide variety of soil types and soil surface roughness conditions. Physical dust models focus largely on the physical mechanisms of dust movement and predict patterns of dust emission, transport, and deposition driven by climate, land use, and/or the land management measures being employed. The evolution of the dust models reviewed in this study is illustrated in Figure 2. We systematically reviewed a representative selection of 18 dust models developed over the past 60 years.
Dust models usually concentrate on smaller (<100 µm in diameter) soil particle emissions, as such particles can be suspended in the atmosphere and transported over long distances [61]. One such model is the wind erosion equation (WEQ) developed by Woodruff and Siddoway [38] from empirical functions that describe the effects of environmental factors on the rate of soil loss. Physical models were developed mainly in the 1980s ( Figure 2). As shown in Figure 2, before 2000, dust models mainly constituted dust emission modules at the field scale (e.g., WEQ, wind erosion prediction system (WEPS), Texas erosion analysis model (TEAM), revised wind erosion equation (RWEQ), wind erosion assessment model (WEAM), and wind erosion stochastic simulator (WESS)) [38,39,[82][83][84][85]. With increasing awareness of the role of dust at regional scales, regional dust emission models were developed and forced by climate datasets (e.g., wind erosion on European light soils (WEELS) and Australian land erodibility model (AUSLEM)) or dust transport models were developed by integrating dust emission modules with regional-to global-scale climate models (e.g., integrated wind  To account for the complex interaction between the physical processes and anthropogenic factors of wind erosion, dust models are drawn from field and laboratory measurements. Owing to the differences in model complexity, required inputs, and outputs [61], these models provide variable dust process simulations at specific spatial and temporal scales. The spatio-temporal scales, input parameters, and outputs of the dust models reviewed in this study are summarised in Table 1. The early dust models are mainly focused on the development of dust emission models at a field scale. However, dust models after 2000 have mainly concentrated on dust transport models at the regional and global scales. Field-scale dust models can be used to assess soil losses due to wind erosion under different land management regimes. Physical models require more detailed inputs and are difficult to implement owing to the lack of soil and land-surface parameters. In addition, these emission models are mainly applied at a field scale and usually cannot estimate spatial variations of dust emissions for a region. Regional and global-scale dust models integrated into dust emission modules and climate models  [53]; DEAD: Dust En-trainment and Deposition [42]; DENAPAP: Dust Emission Model, supported by the National Acid Precipitation Assessment Program [92]; DPM: Dust Production Model [93]; GMOD: Global Transport Model of Dust [43]; GOCART: Global Ozone Chemistry Aerosol Radiation and Transport [94]; GOCART-AFWA: GOCART-Air Force Weather Agency [88]; GOCART-UoC: GOCART-University of Cologne [11,89,90] To account for the complex interaction between the physical processes and anthropogenic factors of wind erosion, dust models are drawn from field and laboratory measurements. Owing to the differences in model complexity, required inputs, and outputs [61], these models provide variable dust process simulations at specific spatial and temporal scales. The spatio-temporal scales, input parameters, and outputs of the dust models reviewed in this study are summarised in Table 1. The early dust models are mainly focused on the development of dust emission models at a field scale. However, dust models after 2000 have mainly concentrated on dust transport models at the regional and global scales. Field-scale dust models can be used to assess soil losses due to wind erosion under different land management regimes. Physical models require more detailed inputs and are difficult to implement owing to the lack of soil and land-surface parameters. In addition, these emission models are mainly applied at a field scale and usually cannot estimate spatial variations of dust emissions for a region. Regional and global-scale dust models integrated into dust emission modules and climate models should be employed to predict spatial and temporal variations of dust processes, such as dust emission, transport, and deposition. Evidently, u*t is the key factor affecting dust emission simulations. Dust emission will occur when the wind friction velocity over the land surface (u*) exceeds u*t. Generally, there are two approaches for representing the factors that influence the soil's susceptibility to wind erosion in dust models: (1) constructing empirical relationships between soil surface conditions, soil moisture, and vegetation cover to predict rates of soil loss (e.g., models WEQ and RWEQ); and (2) integrating physical processes and theoretical relationships among soil properties, land surface conditions, and u*t (e.g., models CFD-WEM, GMOD, TEAM, and WEAM). Empirical models can account for dynamic variations in soil erodibility [96], but they largely depend on field measurements, which are not available at large spatial scales [38]. Physical models enable the inclusion of large-scale spatial inputs and are not restricted to specific environments [40,42]. The complexity of the description of u*t and dust emission has increased as the development of dust models progressed over the past five decades ( Table 1). The definitions of u*t and the soil's susceptibility to wind erosion differ between these models; therefore, calculations and predictions obtained from different models are not directly comparable. Thus, there is a need to integrate the two types of models to reduce the uncertainty of dust models overall.
Precisely modelling the spatial and temporal variability of dust emissions is a prerequisite to estimate and forecast atmospheric dust concentrations and their effects. Current dust emission models mostly include main physical processes of dust production, which can reproduce the spatial and temporal variability of dust emissions if the model inputs are accurately described. Studies have confirmed that the accuracy of ground surface condition data is the key determinant of spatial and temporal variability accuracy of dust emissions models [9,77,97]. In addition, the accuracy of temporal variability of dust emission models is also determined by specific model parameter values, such as the Kawamura coefficient value in a dust scheme [90] and the roughness correction factor to u*t in the dust schemes of the Community Earth System Model (CESM) [77]. However, it is difficult to define or customise these values owing to the spatial heterogeneity of ground surface conditions and the dearth of dust observations. To evaluate and improve the performance of temporal variations of dust emission models, it is essential to improve the accuracy of surface parameters in dust emission models and strengthen the collection of dust observation globally.
Dust models are important tools to account for the complex interaction between the physical processes and anthropogenic factors of wind erosion. However, there are no universally accepted parameters for these models in different regions/countries. Therefore, most of these models have to be parameterised before they can be applied to other regions. For example, models IWEMS and CFD-WEM have been successfully applied to the simulation of dust emissions in Asia after calibration of their parameters [9,96,[98][99][100][101]. Model parameterisation is essential to ensure the accuracy of the estimation results. Dust models can be evaluated using in-situ measurements of dust and other required inputs. To ensure the accuracy of the simulation, before the application of a dust model in a region, the model's empirical variables can be adjusted by comparing the model's predictions with field measurements [101]. However, this comparison is challenging because of the difficulty in obtaining dust data. Several studies have attempted to validate dust model predictions against measured and observed data. The performance of some of the selected dust models in Europe, Australia, and China (Table 2) have shown considerable differences in the accuracy among different dust models and among estimations in different regions using the same model. This also proved that the localisation of model parameters is important for the simulation accuracy of dust models.

Dust Sources and Sinks
Global dust source regions have been identified using different approaches, such as information gathering from dust weather records [99], remote sensing [32,100], dust monitoring networks [106], and dust models [42,55,86,107]. The seven main dust source regions of the world are North Africa, Middle East/Central Asia, East Asia, North America, South America, South Africa, and Australia ( Figure 3).
Some studies simulated the global dust emission, deposition, and budgets over the past three decades ( Table 3). The map of global dust emission and deposition in different regions ( Figure 3) generated based on the data from previous studies represented in Table 4 shows that North Africa is the largest dust source region in the world. Because of the Sahara, the world's largest desert, North Africa accounts for approximately 60% of the global dust emissions and approximately 65% of the global atmospheric dust load [55,108]. The second largest dust source region is Asia, comprising Arabia, Central Asia, and East Asia. Dust emissions and atmospheric dust loads in Asia account for approximately 30% of the global values [42,43,55]. Specifically, the dust emission and atmospheric dust load in East Asia are approximately 214 and 1.1 Tg yr −1 , respectively [55]. Australia is the largest contributor to dust emissions in the Southern Hemisphere, accounting for approximately 6% of the global dust emissions [33,42,55,[108][109][110] and 5% of the global atmospheric dust load [43,55]. The smallest dust source regions are North and South America, accounting for 0.3% and 2.5% of the global dust emissions, respectively [33,42,55,109]. The amount of dust deposition over land is around three orders of that deposition over oceans [52]. Although dust deposition measurements are relatively scarce and incomplete worldwide, existing dust deposition rates records show large variations on land and oceans [52]. The estimates of dust deposition on the ocean shown in Figure 3 illustrate a considerable discrepancy among different studies. Nevertheless, according to most estimates, the region of maximum dust deposition is the North Atlantic due to the Saharan dust, which accounts for nearly 43% of the total dust deposited worldwide [18]. The second largest deposition centre is the Indian Ocean, receiving dust from North Africa, Arabia, Central Asia, and Australia, accounting for approximately 25% of the total dust deposition worldwide. Dust deposition in the North Pacific, South Pacific, South Ocean, and South Atlantic is 15%, 6%, 6%, and 4% of the global total, respectively.

Dust Budgets
The estimated global dust emission ranges from 895 to 8079 Tg yr −1 , and the global atmospheric dust load is estimated to be between 8 and 41.65 Tg yr −1 (Figure 4). Similarly, there is uncertainty regarding the lifetime of the global atmospheric dust load and the ratio of dry to wet deposition. Evidently, there are large discrepancies among the dust models. These discrepancies can be attributed to the following: differences in the description of dust processes in different dust models; different particle size ranges utilised in each model (particle size is a fundamental parameter for simulating soil particle processes and estimating the effect of dust particles on radiation and cloud processes); and different meteorological/climatic data that form a part of the model input.  Figure 3b denote annual dust emission (from land regions) and deposition (in oceans), respectively, estimated by different studies. The particle size ranges (r) of dust emissions are Werner et al. [109], 0.1 ≤ r ≤ 219 μm; Luo et al. [108], 0.1 ≤ r ≤ 10 μm; Zender et al. [42], 0.1 ≤ r ≤ 10 μm; Ginoux et al. [33], 0.1 ≤ r ≤ 6 μm; Miller et al. [110], r < 10 μm; Tanaka and Chiba [55], 0.2 ≤ r ≤ 20 μm; Kok et al. [80], 0.2 ≤ r ≤ 20 μm; Checa-Garcia et al. [117] and Aryal et al. [116], multi-model with different maximum dust particle size. dust budgets at different regions, estimated by several dust models. Grey and black arrows in (a) denote dust emission and deposition (percent of the total dust deposition worldwide), respectively. Horizontal and vertical bars in (b) denote annual dust emission (from land regions) and deposition (in oceans), respectively, estimated by different studies. The particle size ranges (r) of dust emissions are Werner et al. [109], 0.1 ≤ r ≤ 219 µm; Luo et al. [108], 0.1 ≤ r ≤ 10 µm; Zender et al. [42], 0.1 ≤ r ≤ 10 µm; Ginoux et al. [33], 0.1 ≤ r ≤ 6 µm; Miller et al. [110], r < 10 µm; Tanaka and Chiba [55], 0.2 ≤ r ≤ 20 µm; Kok et al. [80], 0.2 ≤ r ≤ 20 µm; Checa-Garcia et al. [117] and Aryal et al. [116], multi-model with different maximum dust particle size. In Africa, the estimated rates of dust emission and deposition are 1112 and 685 Tg yr −1 , respectively (Figure 3). In Asia, 736 Tg of dust are suspended in the atmosphere, and 611 Tg of dust are deposited on the land surface annually. In Australia, the largest dust source in the Southern Hemisphere, the dust emission and deposition rates are 73 and 46 Tg yr −1 , respectively. However, in North and South America, the dust emission rate is considerably low (approximately 14 Tg yr −1 ), whereas the dust deposition rates are 27 (in North America) and 11 Tg yr −1 (in South America). Europe has the smallest rate of dust emission (~1 Tg yr −1 ), but the dust deposition rate is approximately 50 Tg yr −1 . It can be concluded that Africa, Asia, and Australia are net dust sources, whereas North and South America and Europe are net dust sinks. Approximately 25% of the global dust emissions are deposited into the open ocean [118]. The deposition of dust is a primary source of micronutrients, such as nitrate, phosphates, and iron, to the sea surface. This has important implications for the CO2 budget; by increasing the iron concentration of the global ocean, dust deposition can increase plankton productivity and thus decrease CO2 concentration in the atmosphere [4,45].

Loss of SOC Due to Dust Emission
Generally, SOC storage represents the net long-term balance between photosynthesis and respiration in terrestrial ecosystems [29,119]. The global SOC storage is estimated to be approximately 1550 Pg of carbon; this accounts for nearly 54% of the terrestrial carbon pool and is twice the magnitude of the atmospheric carbon pool (760 Pg) [48]. Soil erosion by wind, and the transport and deposition of the eroded material, redistribute SOC across landscapes and regions [20,120]. These physical processes substantially affect the biological mediation of carbon mineralisation in the soil system. Erosion and mobilisation of mineralised carbon could result in a net release of carbon from the soil system to the atmosphere, which may offset carbon sinks in vegetation [120][121][122][123][124][125][126][127]. The fraction of soil carbonates in SOC entering the atmosphere may reduce the intensity of terrestrial carbon sequestration and further increase the CO2 concentration in the atmosphere, which has a positive feedback effect on climate warming [123,124,128].  [113]; B: Ginoux et al. [94]; C: Chin et al. [114]; D: Tegen et al. [115]; E: Werner et al. [109]; F: Zender et al. [42]; G: Luo et al. [108]; H: Miller et al. [110]; I: Tanaka and Chiba [55]; J: Yue et al. [43]; K-M: Huneeus et al. [13]; N-S: Wu et al. [14]; T: Zhao et al. [107].
In Africa, the estimated rates of dust emission and deposition are 1112 and 685 Tg yr −1 , respectively ( Figure 3). In Asia, 736 Tg of dust are suspended in the atmosphere, and 611 Tg of dust are deposited on the land surface annually. In Australia, the largest dust source in the Southern Hemisphere, the dust emission and deposition rates are 73 and 46 Tg yr −1 , respectively. However, in North and South America, the dust emission rate is considerably low (approximately 14 Tg yr −1 ), whereas the dust deposition rates are 27 (in North America) and 11 Tg yr −1 (in South America). Europe has the smallest rate of dust emission (~1 Tg yr −1 ), but the dust deposition rate is approximately 50 Tg yr −1 . It can be concluded that Africa, Asia, and Australia are net dust sources, whereas North and South America and Europe are net dust sinks. Approximately 25% of the global dust emissions are deposited into the open ocean [118]. The deposition of dust is a primary source of micronutrients, such as nitrate, phosphates, and iron, to the sea surface. This has important implications for the CO 2 budget; by increasing the iron concentration of the global ocean, dust deposition can increase plankton productivity and thus decrease CO 2 concentration in the atmosphere [4,45].

Loss of SOC Due to Dust Emission
Generally, SOC storage represents the net long-term balance between photosynthesis and respiration in terrestrial ecosystems [29,119]. The global SOC storage is estimated to be approximately 1550 Pg of carbon; this accounts for nearly 54% of the terrestrial carbon pool and is twice the magnitude of the atmospheric carbon pool (760 Pg) [48]. Soil erosion by wind, and the transport and deposition of the eroded material, redistribute SOC across landscapes and regions [20,120]. These physical processes substantially affect the biological mediation of carbon mineralisation in the soil system. Erosion and mobilisation of mineralised carbon could result in a net release of carbon from the soil system to the atmosphere, which may offset carbon sinks in vegetation [120][121][122][123][124][125][126][127]. The fraction of soil carbonates in SOC entering the atmosphere may reduce the intensity of terrestrial carbon sequestration and further increase the CO 2 concentration in the atmosphere, which has a positive feedback effect on climate warming [123,124,128].
Dust emission affects SOC by selectively removing fine particles from the soil surface. In this way, the soil evolves toward a coarser texture [119]. Fine soil particles have a high content of stable SOC [122], which directly affects plant growth and soil biological activities, soil air CO 2 concentration, soil water regimes, temperature, and respiration, and, therefore, carbon flux to the atmosphere [119,128]. Studies have shown that the SOC loss caused by wind erosion is mainly the active component of SOC and the organic carbon components combined with the soil fine-grained particles [48,54,119,[129][130][131]. Dust emissions can affect soil reflectivity, and thus, soil moisture and temperature, thereby accelerating the in-situ mineralisation of residual SOC [48]. Soil desertification and dust emission reduce the soil's water-holding capacity, root depth, and the efficiency of water and nutrient uptake by plants, thus reducing soil productivity, the amount of organic matter returning to the soil, and the rate of POC formation [24,132]. Moreover, severe dust emission removes the topsoil and exposes the calcium carbonate-rich subsurface soil horizon. This can result in increased emission of CO 2 into the atmosphere due to carbon oxidation [48,128].
Despite the significance of dust in the global carbon cycle, wind erosion-induced carbon emissions remain a poorly understood, unquantified component of the global carbon budget. The SOC erosion associated with dust emission in major regions of the world is presented in Table 4 [24,45,120,131]. The difference in an order of magnitude in total min-max dust emission, wind-eroded area, and total SOC erosion across different regions is shown in Table 3. Although several studies have attempted to estimate SOC losses due to dust emission in specific regions, such as China (75 Tg C yr −1 [7]), Australia (1.59 Tg C yr −1 ) [46], the United States (34 Tg C yr −1 ), a small arable catchment in Germany (4.4 g C m −2 yr −1 ) [133], and a dryland farming system in Western Australia (3.6 t C ha −1 yr −1 ) [134], there is significant inconsistency among these results.

Fate of SOC in Dust
Soil losses due to wind erosion do not amount to a net loss of SOC; it is a process of SOC migration, in other words, a non-source and non-sink process [135,136]. The fate of the SOC involved in dust dynamics is determined by a series of complex interactions. As these interactions constitute a dynamic process, it is difficult to accurately estimate the ultimate fate of wind-eroded SOC. In general, the fate of the SOC is mobilised, as dust may include [134]: (1) proximal deposition, from creep and saltation, in the range of tens of meters; (2) deposition in lakes and rivers; (3) transport, in the form of dust, to a distant system; (4) release to the atmosphere by oxidation; and (5) variation in SOC with dust size.
The net change in SOC stocks reflects the balance between carbon sequestration and soil carbon emission. Some studies have indicated that the main losses in the process of dust emission are mainly the active organic carbon of SOC and the organic carbon combined with soil fine-grained components [137][138][139][140]. Soil active organic matter components are the habitat and survival matrix of soil microorganisms. Therefore, the loss of SOC caused by dust emission can significantly reduce soil biological activity. The decrease in soil biological activity and the change in soil structure and water-holding capacity caused by wind erosion can significantly change the biological process of carbon mineralisation and result in the net release of carbon from the soil system to the atmosphere. Therefore, from the perspective of the global carbon balance, more attention should be paid to the loss of mineralised SOC due to wind erosion. The mechanisms of carbon mineralisation during the migration and deposition of wind-eroded material are yet to be determined. This raises the question of how to estimate the effect of dust emissions on the global carbon balance. The current estimates of SOC loss usually ignore the redistribution of SOC generated by dust emissions; consequently, they overestimate the contribution of SOC erosion to atmospheric CO 2 . The fate of wind-eroded SOC is still discussed in merely qualitative terms. Quantitative analysis is limited to smaller space-time scales. In-depth study and quantification of SOC in dust, especially the fate of wind-eroded SOC in the global dust cycle, is essential to quantify the release of CO 2 from SOC dust to the atmosphere, the contribution of SOC deposition to downwind carbon sinks, and the effect of dust processes on the global carbon balance.

Conclusions
Advances in dust modelling in the past five decades have changed the requirements for input data, and increased model complexity and the availability of model outputs.
Owing to the diversity of the required inputs, hybrid observation methods (integrating multiple observation methods) should be adopted to provide dust models with input data. Although the development of dust models has progressed considerably over the past 30 years, the model simulation results are still replete with uncertainties. Dust models developed in a specific region require careful calibration when used to other regions. It is only possible to simulate dust processes in an area after the model's parameters have been localised with the use of observation data. There are no universally accepted parameters for dust models in different regions/countries. Therefore, it is necessary to develop a set of parameters for different regions. It is recognised that anthropogenic activities can also induce dust emissions; as such, they are non-negligible contributors to global dust concentrations [140][141][142][143][144][145]. However, all models reviewed in this study simulated 'natural' or indirect anthropogenic (e.g., cropland and pastureland) dust processes, neglecting the contribution of direct anthropogenic dust (e.g., city construction and transportation). This leads to considerable uncertainties in estimating dust emissions. Therefore, to improve the accuracy of dust emission simulations, the consideration of anthropogenic dust emissions is imperative [146].
SOC loss due to wind erosion is a key component of the global carbon cycle. A better understanding of the role of dust processes in the global SOC flux and carbon budget is needed. Although it is recognised that SOC is transported and redistributed by dust processes, SOC cycling schemes used in land surface models (LSMs) typically only consider the effects of net primary production and heterotrophic respiration. Current estimates of SOC loss results in significant underestimations due to the omission of the effects of dust emission. Moreover, the dust emission flux observation does not include the measurement of SOC concentrations; there is a lack of SOC concentration in different dust sizes, and how dust emission is directly linked to SOC erosion is not well represented. It is necessary to explore the various effects of dust processes on SOC pools, mineralisation rates, and SOC emission to the atmosphere in dust source regions, and on the enrichment of SOC in deposition regions. Currently, although some Earth System Models have the ability to simulate the effects of mineral dust deposition on biogeochemistry [78,147,148], most dust models are limited to estimating dust emission and deposition and do not consider the effects of dust on the global carbon cycle. Similarly, the current carbon cycle models ignore the effects of SOC movement caused by dust processes. Therefore, representing the linkages between dust processes and the carbon cycle in both dust and carbon cycle models is essential.