1. Introduction
According to the global assessment of land degradation (GLADA), approximately 1.1 billion hectares of land worldwide have been degraded due to soil erosion (Bai et al. 2008). Soil erosion removes the fertile topsoil, which contains essential organic matter and nutrients. Human-induced accelerated soil erosion has resulted in an estimated total potential loss of 74 pg of soil organic carbon (SOC) between 1850 and 2005 (Naipal et al. 2018). Within the context of China, Yue et al. (2016) found that water erosion removed 0.1–0.26 pg·C·yr−1 of soil carbon over the past two decades, causing a redistribution of atmospheric carbon dioxide flux. At a national scale, carbon dioxide sequestration from water erosion contributes to 8%–37% of terrestrial carbon sinks. The amplified spatial distribution and extent of lateral SOC losses have raised concerns regarding the assessment of carbon stocks and sinks (Van Oost and Six 2023; Wang et al. 2022). These problems will be exacerbated in light of rapid land-use and climate changes. A recent study projected a 30%–66% increase in global water-induced soil erosion due to land-use and climate change by 2070 (Borrelli et al. 2020). Therefore, proactive projection of these dynamics can guide effective soil degradation control measures, assess SOC loss and redistribution, and provide theoretical insights into sustainable land resource utilization. To project soil erosion under climate change, a common and valuable approach is the integration of climate models with different climate scenarios and erosion models. While previous soil erosion simulations (Maurya et al. 2021) have been based on the framework of the phase 5 of the Coupled Model Intercomparison Project (CMIP5), comparisons with the latest CMIP6 data are scarce, which our study addresses. With the introduction of new CMIP6 and shared socioeconomic pathway (SSP) scenarios, there is a need to revisit soil erosion and SOC simulations and projections.
Over the past few decades, advancements in climate modeling have shed new light on soil erosion dynamics. Integrated modeling frameworks, combining hydrologic erosion, climate, and land-use models, have been used to study the potential impacts of climate change on erosion (Borrelli et al. 2020; Eekhout and de Vente 2022; Luetzenburg et al. 2020; Pal et al. 2021). To incorporate future climate change effects, a multimodel, multiscenario approach is often employed, utilizing various global climate models (GCMs) in conjunction with the Revised Universal Soil Loss Equation (RUSLE). However, the coarse resolution of GCMs limits their suitability as input data for erosion models (Boé et al. 2023). Additionally, GCMs fail to adequately capture extreme events, which are known to contribute significantly to soil erosion rates. Regional climate models (RCMs) with higher resolution offer a more accurate representation of local forcing factors, including complex topography and land surface heterogeneity (Li et al. 2022; Nikiema et al. 2017).
The CMIP6 models have demonstrated advancements in model resolution and dynamic parameterization schemes compared to earlier versions such as CMIP3 and CMIP5 (Dong and Dong 2021; Kumar et al. 2023). The High Resolution Model Intercomparison Project (HighResMIP) experiments conducted within the CMIP6 models have shown improvements in simulating extreme precipitation and precipitation distribution, particularly in China (Dong and Dong 2021; Xin et al. 2021). Additionally, the CMIP6 models incorporate additional scenarios using SSPs (O’Neill et al. 2016; Schlund et al. 2020), which consider socioeconomic developments, technological advancements, and other environmental factors like land use (Hamed et al. 2022). CMIP6 provides more comprehensive data on land use, land cover, and land management, allowing for a more realistic representation of the complex interactions between human activities, land-use change, and the climate system (Chini et al. 2021; Kumar et al. 2023). These updated climate projections offer a better understanding of the consequences of climate change policies. However, despite these advancements, there is a lack of studies investigating the impact of climate and land-use changes on soil erosion using the recently released CMIP6 high-resolution models and comparing them to the CMIP5 results. Furthermore, although the impact of soil erosion on lateral SOC redistribution is recognized in previous studies (Li et al. 2022; Van Oost and Six 2023; Yue et al. 2016), few studies have quantified and compared future changes in SOC redistribution rates caused by soil erosion across different CMIPs models.
Therefore, this study aims to achieve three main objectives: 1) Comprehensive analysis and comparison of the impacts and differences of climate and land-use changes on soil erosion in China through the CMIP5 and CMIP6 high-resolution models. 2) Quantitatively analyze the contribution rates of climate change and land-use change to soil erosion under the two frameworks to explore underlying mechanisms. 3) Investigation of variations in lateral redistribution rates of SOC caused by soil erosion using different CMIPs. By addressing these objectives, this study seeks to improve our understanding of the effects of climate and land-use changes on soil erosion, provide insights into the mechanisms driving soil erosion, and explore variations in SOC redistribution rates.
2. Data and methods
a. Study area
China, located on the east side of the Eurasian continent and the west coast of the Pacific Ocean, spans geographically from 73°33′ to 135°05′E and 3°51′ to 53°33′N. The main factors affecting soil erosion are precipitation, soil, topography, vegetation, and land use. The precipitation in China is affected by monsoon climate and topography, and the regional distribution of precipitation is extremely uneven. It decreases from the southeast coast to the northwest inland. The soil types are diverse and display a zonal distribution pattern. Its diverse topography, with numerous mountains, hills, and complex geological layers, particularly the widespread Quaternary loose sediments and slightly cemented clastic rocks, creates favorable conditions for soil erosion. The zonal distribution of vegetation in China is evident, with landscape features transitioning from forests along the coast to grasslands and eventually deserts, reflecting a regional differentiation pattern from coastal to inland areas. China faces imbalances in social and economic development. The eastern region experiences intense economic development, leading to the conversion of significant agricultural land into urban construction areas, and human factors have become the main cause of soil erosion. The western region faces challenges of grassland overloading, land desertification, and degradation. Southwest China still grapples with excessive land conversion, leading to severe soil erosion on sloping farmland. Factors such as frequent rainstorms, dense population distribution, and intensive production activities contribute to distinct soil erosion types and distributions across China. In accordance with the Soil Erosion Classification Standard of China established by the Ministry of Water Resources, the study area is divided into eight water and soil conservation areas, as depicted in Fig. 1.
The eight soil and water conservation areas of mainland China: northeast China black soil region (I), North China mountainous region (II), northwest China Loess Plateau region (III), North China sandstorm region (IV), South China red soil region (V), southwest China purple soil region (VI), southwest China karst region (VII), and Qinghai–Tibet Plateau region (VIII).
Citation: Earth Interactions 29, 1; 10.1175/EI-D-24-0002.1
b. Datasets
1) Precipitation data related to climate change
In this study, five models of CMIP5 (https://esg-dn1.nsc.liu.se/search/cordex/) and five models of CMIP6 (https://esgf-node.ipsl.upmc.fr/search/cmip6-ipsl/) are selected to simulate the daily precipitation in reference and future periods. The models and their institutions are summarized in Table 1. The five models of CMIP5 are derived from four Coordinated Regional Climate Downscaling Experiment (CORDEX) East Asia experiments and the Providing Regional Climate Impacts for Studies (PRECIS) simulation. The five CORDEX models selected have good performance in simulating precipitation over China (Zhu et al. 2018).
Model names, modeling centers, and the atmospheric resolutions of CMIP5 RCMs and CMIP6-HighResMIP GCMs used to simulate rainfall.
We use five high-resolution ensemble members from the CMIP6 HighResMIP tier (Haarsma et al. 2016). The HighResMIP is a CMIP6-endorsed Model Intercomparison Project (MIP) and for the first time applies a multimodel approach to the systematic investigation of the impact of horizontal resolution. The HighResMIP simulations incorporate the model resolution (grid spacing) that ranges from typical CMIP6 resolutions (∼250 km in the atmosphere and 100 km in the ocean) to considerably higher resolutions (25 km in the atmosphere and 8–25 km in the ocean). The goal of HighResMIP is to assess the performance improvement of climate models at higher horizontal resolutions and reduce simulation uncertainty through multimodel ensemble simulations. High-resolution models are better equipped to capture diurnally forced circulations and the influence of orography on rainfall patterns (Dong and Dong 2021).
Our focus lies on the high-emission pathways, namely, RCP8.5 for CMIP5 and SSP5-8.5 for CMIP6. These pathways allow us to explore the response of climate to high-level warming, such as a 3°C increase above preindustrial levels. SSP5-8.5 for CMIP6 and RCP8.5 for CMIP5 are both high-emission scenarios, with a similar radiative forcing of 8.5 W m−2 by 2100. Although SSP5-8.5 exhibits approximately 20% higher CO2 emissions by the end of the century and lower emissions of other greenhouse gases, the differences between the two scenarios are relatively small.
2) Land-use change data
The CMIP5 and CMIP6 simulations of land-use states rely on two datasets: Land-Use Harmonization 1 (LUH1) (https://doi.org/10.3334/ORNLDAAC/1248) and LUH2 (https://luh.umd.edu/). LUH1, developed for CMIP5, provides harmonized land-use data at a resolution of 50 km for the years 1500–2100. It includes categories such as cropland, pasture, primary land, secondary (recovering) land, urban land, and annual land-use transitions. Building upon LUH1, LUH2 (Hurtt et al. 2020) was developed for CMIP6. It incorporates updates from the History of the Global Environment (HYDE) database for historical agricultural patterns, a new historical wood harvest reconstruction, improved maps, and 20 rates of shifting cultivation. LUH2 covers the timespan of 850–2100 at a higher resolution of 25 km and utilizes remote sensing observations to constrain forest cover transitions. It includes 12 land-use types and encompasses transitions between all combinations of these categories.
In addition to land-use types, CMIP5 (https://esgf-node.ipsl.upmc.fr/search/cmip5-ipsl/) and CMIP6 (https://esgf-node.ipsl.upmc.fr/search/cmip6-ipsl/) models also simulate changes in vegetation cover related to plant litter and leaf area index. Considering the robustness of the results, model ensemble averaging is also used to observe the impact of vegetation cover changes. In this study, models that simulated both litter carbon stock and leaf area index under the same conditions are selected, as shown in Table 2. Vegetation cover-related parameters have a low resolution of 100 km.
Model names and modeling centers of CMIP5 and CMIP6 used to simulate vegetation cover.
3) Soil organic carbon data
The monthly soil carbon density data, with a resolution of 100 km, are obtained from five CMIP5 and five CMIP6 models, as shown in Table 3. To validate soil organic carbon, the Global Soil Dataset for Earth System Modeling (GSDE) is utilized. GSDE provides soil information, including soil particle-size distribution, organic carbon, and nutrients, along with quality control information represented by confidence levels. The data cover a horizontal resolution of 10 km and include eight vertical layers up to a depth of 2.3 m (Shangguan et al. 2014).
Model names and modeling centers of CMIP5 and CMIP6 used to simulate SOC content.
c. RUSLE model
To judge the reliability of the soil erosion obtained from the simulation, two national soil erosion survey datasets accomplished in 1995–96 and 2010–12 are used as validation sets. These two national soil erosion surveys were generated by comprehensively using satellite-based images, field measurements, and the Chinese Soil Loss equation (Wang et al. 2019).
1) Rainfall erosivity factor
2) Soil erodibility factor
3) Slope length–steepness factor
4) Cover management factor
5) Conservation support practice factor
The conservation support practice factor P represents the ratio of soil loss with soil conservation measures to soil loss without any measures. The P values range from 0 to 1, where 0 indicates no soil erosion in the area and 1 signifies the absence of soil conservation measures. Land-use types are commonly used as indirect indicators of the P factor. In this study, based on the characteristics of the study area and relevant references (Ghosal and Das Bhattacharya 2020; Tang et al. 2015; Xue et al. 2018), land types without soil and water conservation measures, such as primary and secondary forest land, are assigned a P value of 1.0. The P values for secondary nonforest land, cropland, managed pasture, rangeland, and urban land are set at 0.1, 0.4, 0.8, 0.5, and 0.3, respectively. The observed land-use types include paddy, dryland, woodland, grassland, water, urban land, and unutilized land, with corresponding values of 0.05, 0.4, 1, 1, 0, 0.3, and 1.
d. Detachment of SOC by erosion
e. Contribution of climate and land-use change to soil erosion
3. Results
a. Simulations of soil erosion and SOC
Figure 2 shows a comparison of soil erosion data derived from CMIP5 and CMIP6 simulations with the baseline values for each subzone. The results indicate that CMIP5 and CMIP6 can effectively capture the spatial distribution of soil erosion in the North China mountainous region, the northwest China Loess Plateau region, and the southwest China purple soil region. However, both simulations tend to exhibit varying degrees of overestimation in the North China sandstorm region and the Qinghai–Tibet Plateau region. This discrepancy may be attributed to the dominance of wind erosion in the North China sandstorm region, where the applicability of the RUSLE model is limited to hydraulic erosion areas. In contrast, the Tibetan Plateau presents a diverse range of erosion types, coupled with fewer field measurements of soil erosion, making it challenging to establish a reliable baseline value. In addition, the results based on CMIP6 simulations in the South China red soil region surpass those based on CMIP5. There is a serious overestimation of CMIP5 in the South China red soil region. The CMIP5 ensemble model classifies this region as having poor vegetation cover and management, reflected in a significantly higher value of factor C compared to CMIP6. During the historical period (1986–2005), the average annual total soil erosion observed in the study area is 285 t km−2 a−1. The numerical value calculated from the CMIP6 ensemble model is 291 t km−2 a−1. Conversely, the spatial average from CMIP5 reaches 748 t km−2 a−1, primarily influenced by the North China sandstorm region and the Qinghai–Tibet Plateau region. Considering the integrated simulation of soil erosion caused by climate and land-use changes, CMIP6 demonstrates better performance in simulating both the amount and spatial distribution of soil erosion in the study area than CMIP5. Compared to the ensemble model of CMIP5, CMIP6 has better validation results as the high-resolution GCM can capture regional details and its interactions with large-scale circulation when simulating rainfall, and CMIP6 considers more complex social scenarios when executing the plan.
Comparison of soil erosion output from CMIP5 and CMIP6 models with reference values [Each bar represents the average rate of soil erosion, color-coded to distinguish reference value (gray), CMIP5 (orange), and CMIP6 (blue) model outputs].
Citation: Earth Interactions 29, 1; 10.1175/EI-D-24-0002.1
The simulated values for rainfall erosivity, vegetation cover and management factor, and soil and water conservation measures factor obtained from the CMIP5 and CMIP6 ensemble models, respectively, are given in Table 4. The rainfall erosivity simulated by CMIP6 is 4586 MJ mm hm−2 h−1 a−1, which is higher than the 4276 MJ mm hm−2 h−1 a−1 simulated by CMIP5. Notably, CMIP5 fails to capture orographic effects and local landmass changes that influence the spatial variability and distribution of rainfall, consequently leading to an underestimation of extreme rainfall by the RCMs. In contrast, the CMIP6 simulations better reproduce the impact of rainfall erosivity on soil erosion, particularly concerning critical soil and water loss resulting from heavy rainfall events. Recent studies have highlighted improvements in CMIP6 compared to CMIP5, particularly in simulating the spatial variability of average mean precipitation in both arid and wet regions (Gusain et al. 2020). Regarding the cover management factor, the simulated results from the two models exhibit a mixed trend across different regions. Nevertheless, CMIP6 exhibits a closer alignment with the observations in the P factor compared to CMIP5, particularly in arid and semiarid regions. This improvement is attributed to the inclusion of newly added management factors, higher spatial resolution, increased detail, and additional management layers in CMIP6, contributing to a more accurate representation of the P factor (Hurtt et al. 2020).
Factors A, R, and P based on the CMIP5 and CMIP6 ensemble models in each area for 1986–2005.
Figure 3 depicts the spatial distribution of SOC density in the study area based on observational data and simulations from two CMIP ensembles for the historical period. According to the GSDE dataset, the topsoil’s SOC concentration is 5.0%, while CMIP5 and CMIP6 suggest SOC values of 4.8% and 3.9%, respectively. Compared with GSDE data, both CMIPs underestimated the SOC content in the northeast China black soil region and the Qinghai–Tibet Plateau region. In the North China mountainous region, northwest China Loess Plateau region, South China red soil region, and southwest China purple soil region, CMIP6 demonstrates some improvements compared to CMIP5, but it still underestimates the SOC content in the North China sandstorm region and southwest China karst region. Nonetheless, it is important to note that there is a certain error in the soil organic carbon content obtained by GSDE. A notable degree of uncertainty persists in simulating SOC content in CMIPs. Therefore, caution should be warranted when interpreting SOC projections derived from climate models. Acknowledging these uncertainties and limitations is vital when interpreting SOC data from CMIPs, as they bear implications for comprehending carbon storage dynamics and ecosystem responses to climate change. Further research and refinement of modeling approaches are imperative to enhance the accuracy of SOC simulations in future climate projections.
Comparison of SOC contents (%) from CMIP5 and CMIP6 models with reference to GSDE [Each bar represents the average rate of SOC moved due to soil erosion processes, color-coded to distinguish reference value (gray), CMIP5 (orange), and CMIP6 (blue) model outputs].
Citation: Earth Interactions 29, 1; 10.1175/EI-D-24-0002.1
b. Projections of soil erosion
We utilize the outputs of CMIP5 and CMIP6 to assess changes in soil erosion across China for the near future (2031–50) compared to the base period (Fig. 4). Under the RCP8.5 scenario, the CMIP5 ensemble projects an upward trajectory in all subregions except the Qinghai–Tibet Plateau region, with the northwest China Loess Plateau region showing the most pronounced increase in soil erosion rate at 250 t km−2 a−1. Conversely, the CMIP6 projections indicate an overall decreasing trend in the study area, with the most substantial reduction in the northwest China Loess Plateau region and the southwest China purple soil region.
Annual soil erosion changes (t km−2 a−1) projected by CMIP5 and CMIP6 models for 2031–50 relative to 1986–2005 [Each bar represents the rate of change in soil erosion, color-coded to distinguish between CMIP5 (orange) and CMIP6 (blue) model outputs].
Citation: Earth Interactions 29, 1; 10.1175/EI-D-24-0002.1
In terms of relative change (Fig. 5), the CMIP5 ensemble simulations indicate the most significant increase in the southwest China purple soil region, with a 30% surge in soil erosion rates relative to the historical period. The South China red soil region and the southwest China karst region anticipate increases of 24% and 20%, respectively. The northeast China black soil region exhibits a modest relative change, with an increase of only 4%. Similarly, CMIP6 projects the largest change in the southwest China purple soil region, showcasing a remarkable 47% decrease. Following this, the southwest China karst region and the South China red soil region anticipate decreases of 20% and 13%, respectively, while the Qinghai–Tibet Plateau region experiences the smallest decrease, less than 1%.
Relative changes (%) in annual soil erosion projected by CMIP5 and CMIP6 models.
Citation: Earth Interactions 29, 1; 10.1175/EI-D-24-0002.1
Divergences in soil erosion projections between CMIP5 and CMIP6 stem from the different weighting of management factors within the CMIP frameworks, emphasizing that effective mitigation measures could counteract adverse climate change impacts. It is worth noting that soil erosion, influenced by both human disturbances and heightened rainfall erosivity, is less severe from CMIP6 than from CMIP5. These findings underscore the potential efficacy of implementing measures to mitigate soil erosion and the importance of incorporating land management practices into climate change adaptation policies.
c. Effects of climate and land-use change on soil erosion
To project changes in soil erosion, the influence of the R, C, and P factors in the RUSLE model is considered, primarily through alterations in rainfall patterns and land-use types. Utilizing the control variate method, the contribution rates of erosive rainfall and land-use change to soil erosion changes are calculated and presented in Table 5. As per the CMIP5 simulation, rainfall change contributes significantly to the exacerbation of soil erosion in the study area, constituting 51.75% of the total impact. While land-use change contributes negatively, signifying a −28.85% mitigating effect on soil erosion. Even though CMIP6 simulates a larger rainfall erosivity than CMIP5, CMIP6 amplifies the mitigating effects of land use, with a contribution rate of −54.01%, and reduces the contribution rate of rainfall erosivity to 32.75%.
Contributions of climate change and land-use change to soil erosion (%) in each area.
Theoretically, the increase in temperature and precipitation could lead to a heightened vegetation cover (Ren et al. 2023). With intensified rainfall and rising temperatures, CMIPs project a decrease in the C factor. When vegetation grows vigorously, surface vegetation can effectively trap rainfall and diminish the impact of rainfall on soil surface particles subsequently. The protective influence of vegetation evolves with its growth and development. CMIP6 further amplifies the effect of vegetation cover on soil erosion compared to CMIP5. Despite improvements in simulating vegetation structure and distribution, CMIP6 models still tend to overestimate the global average LAI (Song et al. 2021). Therefore, concerted efforts are essential to enhance the simulation and resolution of vegetation structure and distribution.
Land-use change plays a pivotal role in shaping soil erosion dynamics. Practices that tend to reduce erosion include reforestation and agricultural abandonment (Eekhout and de Vente 2022). Reforestation increases vegetation cover, which helps to trap rainfall and bind soil particles, thereby reducing the impact of rainfall on the soil surface and decreasing erosion rates. Agricultural abandonment allows natural vegetation to recover, which also has a similar effect in reducing soil erosion. On the other hand, activities such as agricultural expansion and deforestation often amplify erosion rates. Agricultural expansion may involve clearing of natural vegetation, which exposes the soil to the erosive forces of wind and water. Deforestation removes the protective cover of trees and vegetation, increasing the susceptibility of the soil to erosion. As awareness of environmental protection grows, implementing scientific soil and water conservation measures becomes crucial in controlling soil and water loss. Soil conservation practices have the potential to offset approximately 64% of the estimated increase in soil erosion (Borrelli et al. 2017). In most regions, the mitigating effect of the land-use change on soil erosion is more pronounced under the SSP5-8.5 scenario than under the RCP8.5 scenario in the near future. Especially, the southwest China purple soil region exhibits the largest differences. Given the smaller size of the purple soil area in comparison to other regions, the proportional change in the overall area becomes particularly noticeable. Under the SSP5-8.5 climate scenario, a considerable portion of land in the region has transitioned from undisturbed natural vegetation to nonforest land that has undergone gradual recovery under human influence.
As an important food production base in China, in order to realize the sustainable use of black-soil resources, the intensity of cultivation in the black-soil region can be appropriately reduced, and fallow land can be returned to grassland to increase grass cover. Rainfall will exacerbate soil erosion in the North China mountainous region and South China red soil region. With high population density and rapid socioeconomic development in these region, more active soil and water conservation planning and actions are needed to safeguard the relatively fragile urban soils. After decades of soil and water loss control in the northwest China Loess Plateau region, soil and water conservation has achieved outstanding results, and high-quality development of soil and water conservation and ecological protection should be adhered to in the future. In the updated SSP5-8.5 climate scenario, the contribution of land-use change is obvious in the southwest China purple soil region. This shows that the sustainability of agriculture and water projects in the region has received significant attention, taking into account more realistic socioeconomic activities. Soils in the southwest China karst region are thin and easily erodible, but the soil that can be eroded is essentially less. To furnish comprehensive projections, it is recommended that assessments of soil losses due to water erosion incorporate a diverse array of climate change and land-use change scenarios. Such an approach will facilitate a more accurate understanding of the potential impacts and support informed decision-making.
d. Lateral redistribution of SOC
Soil carbon detachment resulting from erosion is calculated by an empirical method with soil erosion rates derived from the RUSLE and CMIPs models. Figure 6 depicts the rate of lateral redistribution of soil carbon content due to erosion in each region between 1986 and 2005, based on the output of the CMIP5 and CMIP6 models. The mean simulated rate of SOC redistributions caused by water erosion is 0.29 t km−2 a−1 based on CMIP5 and 0.11 t km−2 a−1 based on CMIP6. These rates display a spatial pattern akin to soil erosion rates, with lower SOC redistribution rates observed in the eastern regions and higher rates in the Qinghai–Tibet Plateau region, northwest China Loess Plateau region, southwest China purple soil region, and southwest China karst region. Yue et al. (2016) have also highlighted notably elevated erosion-induced lateral SOC loss in regions characterized by steep slopes, high-relief topography, and high initial SOC densities, such as the Tibetan Plateau and karst areas. However, the low resolution of the simulations results in both CMIPs failing to capture the elevated SOC redistribution rates in the Northeast China black soil region. Although CMIP6 demonstrates improved accuracy in simulating soil erosion compared to CMIP5, there remains substantial room for improving the simulation of SOC changes. It is crucial to continue refining the models to achieve a better representation and understanding of the intricate dynamics between soil erosion and SOC redistribution.
Simulated lateral redistribution rates of SOC (t km−2 a−1) in the historical period (1986–2005) based on CMIP5 and CMIP6 model outputs [Each bar represents the average rate of SOC moved due to soil erosion processes, color-coded to distinguish between CMIP5 (orange) and CMIP6 (blue) model outputs].
Citation: Earth Interactions 29, 1; 10.1175/EI-D-24-0002.1
Figure 7 depicts the projected changes in soil carbon content resulting from soil erosion in the future relative to the historical period. Both CMIP5 and CMIP6 models show a decline in the total rate of SOC displacement, with reductions by 0.04 and 0.02 t km−2 a−1, with the most substantial change in SOC redistribution rates projected in the Qinghai–Tibet Plateau. Combining changes in SOC content and soil erosion rates, CMIP5 projects the most pronounced increase in soil organic carbon displacement rate by erosion on the Loess Plateau, which is 0.12 t km−2 a−1. In contrast, CMIP6 shows the most significant decrease in soil organic carbon displacement rate by 0.1 t km−2 a−1 in the southwest China purple soil region. The CMIP5 and CMIP6 models present divergent estimations concerning future changes in SOC redistribution rates, consistent with earlier projections on soil erosion. The removal fluxes of SOC in China are different from those estimated by Yue et al. (2016). The erosion rate of SOC is higher in areas with intensive erosion intensity and high initial SOC density. However, the SOC erosion rate projected by CMIP5 and CMIP6 models in this study is consistent with the change direction of soil erosion projected before. This is mainly due to the coarse resolution and poor accuracy of the CMIPs model in simulating SOC content. The relative changes in the SOC erosion rates in each subregion are shown in Fig. 8. Both CMIP5 and CMIP6 simulations show the most significant changes in the southern part of the study area. CMIP5 projects a 25% increase in the South China red soil region, southwest China purple soil region, and southwest China karst region, while CMIP6 sees a 10%–15% decrease in the South China red soil region and southwest China karst region, with the largest decrease of up to 44% in the southwest China purple soil region.
Projected changes in the SOC erosion rate (t km−2 a−1) from two models for the period 2031–50 relative to 1986–2005 [Each bar represents the change rate of SOC moved due to soil erosion processes, color-coded to distinguish between CMIP5 (orange) and CMIP6 (blue) model outputs].
Citation: Earth Interactions 29, 1; 10.1175/EI-D-24-0002.1
Percentage changes in the SOC erosion rate projected by CMIP5 and CMIP6 models.
Citation: Earth Interactions 29, 1; 10.1175/EI-D-24-0002.1
When analyzing SOC redistribution in the study area using the latest simulations and projections from CMIP6, it is vital to carefully consider the disparities compared to the results from CMIP5. These disparities will directly influence the development and implementation of soil and water conservation measures.
4. Limitations and uncertainties
We acknowledge that modeling soil erosion and the loss of soil organic carbon using CMIPs remains challenging. The uncertainty of the article mainly comes from the use of the RUSLE model, the empirical method for estimating SOC detachment, and the differences in the resolution and parameterization of CMIP5 and CMIP6 models.
The RUSLE model is established based on empirical equations, and its parameterization process is affected by multiple factors. During the simulation process, it may not be able to fully capture the dynamic changes and interactions of all these factors, resulting in a certain deviation between the simulation results and the actual situation. However, in large-scale regional studies, physical erosion models are limited by computational resources, and choosing the RUSLE model will have obvious advantages (Borrelli et al. 2017).
The study assumes that the rate of SOC erosion is proportional to the soil erosion rate and the SOC content. However, in reality, the rate of SOC erosion can change due to various factors (Yang et al. 2020; Wang et al. 2019). For example, soil texture can affect the separation and transport of SOC. Soils with finer textures may have a higher ability to retain SOC, while soils with coarser textures may be more likely to cause SOC to separate and be lost. Soil aggregates also play an important role. Well-aggregated soils can protect SOC within the aggregates and reduce its susceptibility to erosion. Management practices such as tillage may disrupt soil aggregates and increase the vulnerability of SOC to erosion. These factors are not fully considered in the estimation, which may lead to inaccuracies in estimating the amount of SOC detachment.
The spatial and temporal resolution of CMIPs models has a gap with the fine-scale nature of soil erosion. For example, there are obvious errors in the simulation of rainfall, which is closely related to soil erosion. Traditional global climate models and regional climate models rely on cumulus convection parameterization schemes (Qing and Wang 2021). Although this scheme can describe the average characteristics of convection, it is difficult to capture local storms and short-term heavy rainfall events. When simulating climate processes, the use of convection parameterization schemes inevitably introduces some uncertainties. Previous studies have pointed out that convection parameterization schemes usually cause some typical simulation errors, such as difficulty in accurately reproducing the diurnal cycle of convective rainfall, overestimating the frequency of low rainfall intensity events, and underestimating the hourly rainfall intensity and the number of dry days (Fosser et al. 2015; Leutwyler et al. 2017; Sun et al. 2016). Therefore, explicitly resolving deep convection processes and improving the model resolution are the expected directions for CMIPs climate models.
Although the CMIP6 model has improved in some aspects compared with CMIP5, there is still room for improvement. For example, when simulating vegetation structure and distribution, CMIP6 models tend to overestimate the global average leaf area index (Song et al. 2021). Vegetation cover is an important factor in soil erosion as it can mitigate erosion by intercepting rainfall and binding soil particles. Therefore, inaccurate representation of vegetation can affect the estimation of soil erosion rates and subsequent organic carbon redistribution.
These limitations and uncertainties highlight the need for caution when interpreting the results of this study. Future research could focus on improving the models by incorporating more detailed physical processes related to soil erosion and SOC dynamics. Combining physical and empirical models may provide a more comprehensive and accurate understanding of soil erosion in a larger spatial range. Additionally, further efforts to enhance the resolution of climate models and improve the representation of land use and management factors would help reduce uncertainties and provide more reliable projections for soil and water conservation decision-making.
5. Conclusions
CMIP6 is able to better reproduce the soil erosion distribution in the study area compared to CMIP5. Notably, the high-resolution GCMs utilized in CMIP6 exhibit superior validation results by capturing regional intricacies and their interactions with large-scale circulation during rainfall simulations. Additionally, CMIP6 incorporates more intricate social scenarios, further refining its predictive capabilities. Furthermore, CMIP6 rectifies the overestimations observed in soil organic carbon content modeling, a limitation seen in CMIP5.
Both CMIP5 and CMIP6 project declining trends in future soil erosion within the study area. The average projected soil erosion changes are estimated at −87 t km−2 a−1 from CMIP5 and at −39 t km−2 a−1 from CMIP6, exhibiting notable geographical heterogeneity. However, CMIP5 projects an increase in the rate of soil erosion in regions outside the Qinghai–Tibet Plateau by the midcentury. The contribution rates of land-use change and climate change to soil erosion are quantified in CMIP simulations. In CMIP5, climate change contributes to 51.75% of the increased soil erosion rate, with land-use change mitigating by −28.85%. Similarly, CMIP6 attributes 32.75% of the increase to climate change and −54.01% to land-use change. Despite CMIP6 simulating higher rainfall erosivity compared to CMIP5, the negative contribution of land-use change is more pronounced in CMIP6. This suggests a stronger mitigating impact of the P and C factors on soil erosion in CMIP6, particularly in the near future. Implementing scientific and rational soil and water conservation measures, such as agricultural abandonment and reforestation, can help reverse the impacts of climate change on soil and water loss.
The average decrease in SOC displacement rate is projected to be 0.04 t km−2 a−1 in CMIP5 and 0.02 t km−2 a−1 in CMIP6. The CMIP6 projection indicates a milder soil erosion and SOC displacement situation due to the moderating effect of land-use change. Decision-makers are recommended to update impact studies for water and soil conservation by incorporating CMIP6 high-resolution models alongside CMIP5.
Acknowledgments.
This research was funded by the National Natural Science Foundation of China (Grant 42301021) and the Guangdong Natural Science Foundation (Grant 2023A1515012046). Data used in this paper are freely available at Mendeley Data (https://doi.org/10.17632/w948kfrsv6.1). We acknowledge and thank the climate modeling groups in the Coordinated Regional Climate Downscaling Experiment and Coupled Model Intercomparison Project for generating their model outputs and making them available. All authors declare no competing financial interests.
Data availability statement.
Datasets analyzed during the current study are available in the phase 5/6 of the Coupled Model Intercomparison Project. These datasets were derived from public domain resources (https://aims2.llnl.gov/search/cmip6/ and https://esg-dn1.nsc.liu.se/search/cordex/).
REFERENCES
Bai, Z. G., D. L. Dent, L. Olsson, and M. E. Schaepman, 2008: Global assessment of land degradation and improvement: 1. Identification by remote sensing. Blumea GLADA Tech. Rep. 5, 78 pp., https://www.isric.org/sites/default/files/isric_report_2008_01.pdf.
Boé, J., A. Mass, and J. Deman, 2023: A simple hybrid statistical–dynamical downscaling method for emulating regional climate models over Western Europe. Evaluation, application, and role of added value? Climate Dyn., 61, 271–294, https://doi.org/10.1007/s00382-022-06552-2.
Borrelli, P., and Coauthors, 2017: An assessment of the global impact of 21st century land use change on soil erosion. Nat. Commun., 8, 2013, https://doi.org/10.1038/s41467-017-02142-7.
Borrelli, P., and Coauthors, 2020: Land use and climate change impacts on global soil erosion by water (2015–2070). Proc. Natl. Acad. Sci. USA, 117, 21 994–22 001, https://doi.org/10.1073/pnas.2001403117.
Chen, P., and Coauthors, 2023: Tendency of soil erosion dynamics by coupling radioisotopes and RUSLE model on the southeastern Tibetan Plateau in response to climate warming and human activity. Catena, 223, 106954, https://doi.org/10.1016/j.catena.2023.106954.
Chini, L., and Coauthors, 2021: Land-use harmonization datasets for annual global carbon budgets. Earth Syst. Sci. Data, 13, 4175–4189, https://doi.org/10.5194/essd-13-4175-2021.
Dong, T., and W. Dong, 2021: Evaluation of extreme precipitation over Asia in CMIP6 models. Climate Dyn., 57, 1751–1719, https://doi.org/10.1007/s00382-021-05773-1.
Eekhout, J. P. C., and J. de Vente, 2022: Global impact of climate change on soil erosion and potential for adaptation through soil conservation. Earth-Sci. Rev., 226, 103921, https://doi.org/10.1016/j.earscirev.2022.103921.
Fosser, G., S. Khodayar, and P. Berg, 2015: Benefit of convection permitting climate model simulations in the representation of convective precipitation. Climate Dyn., 44, 45–60, https://doi.org/10.1007/s00382-014-2242-1.
Getu, L. A., A. Nagy, and H. K. Addis, 2022: Soil loss estimation and severity mapping using the RUSLE model and GIS in Megech watershed, Ethiopia. Environ. Challenges, 8, 100560, https://doi.org/10.1016/j.envc.2022.100560.
Ghosal, K., and S. Das Bhattacharya, 2020: A review of RUSLE model. J. Indian Soc. Remote Sens., 48, 689–707, https://doi.org/10.1007/s12524-019-01097-0.
Gusain, A., S. Ghosh, and S. Karmakar, 2020: Added value of CMIP6 over CMIP5 models in simulating Indian summer monsoon rainfall. Atmos. Res., 232, 104680, https://doi.org/10.1016/j.atmosres.2019.104680.
Haarsma, R. J., and Coauthors, 2016: High Resolution Model Intercomparison Project (HighResMIP v1.0) for CMIP6. Geosci. Model Dev., 9, 4185–4208, https://doi.org/10.5194/gmd-9-4185-2016.
Hamed, M. M., M. S. Nashwan, S. Shahid, T. b. Ismail, X.-j. Wang, A. Dewan, and M. Asaduzzaman, 2022: Inconsistency in historical simulations and future projections of temperature and rainfall: A comparison of CMIP5 and CMIP6 models over Southeast Asia. Atmos. Res., 265, 105927, https://doi.org/10.1016/j.atmosres.2021.105927.
Hurtt, G. C., and Coauthors, 2020: Harmonization of global land use change and management for the period 850–2100 (LUH2) for CMIP6. Geosci. Model Dev., 13, 5425–5464, https://doi.org/10.5194/gmd-13-5425-2020.
Kumar, N., V. G. Singh, S. K. Singh, D. K. Behera, and M. Gašparović, 2023: Modeling of land use change under the recent climate projections of CMIP6: A case study of Indian River basin. Environ. Sci. Pollut. Res., 30, 1 07 219–1 07 235, https://doi.org/10.1007/s11356-023-26960-z.
Leutwyler, D., D. Lüthi, N. Ban, O. Fuhrer, and C. Schär, 2017: Evaluation of the convection-resolving climate modeling approach on continental scales. J. Geophys. Res. Atmos., 122, 5237–5258, https://doi.org/10.1002/2016JD026013.
Li, L.-L., J. Li, and R.-C. Yu, 2022: Evaluation of CMIP6 HighResMIP models in simulating precipitation over central Asia. Adv. Climate Change Res., 13 (1), 1–13, https://doi.org/10.1016/j.accre.2021.09.009.
Luetzenburg, G., M. J. Bittner, A. Calsamiglia, C. S. Renschler, J. Estrany, and R. Poeppl, 2020: Climate and land use change effects on soil erosion in two small agricultural catchment systems Fugnitz—Austria, Can Revull—Spain. Sci. Total Environ., 704, 135389, https://doi.org/10.1016/j.scitotenv.2019.135389.
Maurya, S., P. K. Srivastava, A. Yaduvanshi, A. Anand, G. P. Petropoulos, L. Zhuo, and R. K. Mall, 2021: Soil erosion in future scenario using CMIP5 models and Earth observation datasets. J. Hydrol., 594, 125851, https://doi.org/10.1016/j.jhydrol.2020.125851.
Naipal, V., P. Ciais, Y. Wang, R. Lauerwald, B. Guenet, and K. Van Oost, 2018: Global soil organic carbon removal by water erosion under climate change and land use change during AD 1850–2005. Biogeosciences, 15, 4459–4480, https://doi.org/10.5194/bg-15-4459-2018.
Nikiema, P. M., M. B. Sylla, K. Ogunjobi, I. Kebe, P. Gibba, and F. Giorgi, 2017: Multi-model CMIP5 and CORDEX simulations of historical summer temperature and precipitation variabilities over West Africa. Int. J. Climatol., 37, 2438–2450, https://doi.org/10.1002/joc.4856.
O’Neill, B. C., and Coauthors, 2016: The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6. Geosci. Model Dev., 9, 3461–3482, https://doi.org/10.5194/gmd-9-3461-2016.
Pal, S. C., R. Chakrabortty, P. Roy, I. Chowdhuri, B. Das, A. Saha, and M. Shit, 2021: Changing climate and land use of 21st century influences soil erosion in India. Gondwana Res., 94, 164–185, https://doi.org/10.1016/j.gr.2021.02.021.
Phinzi, K., and N. S. Ngetar, 2019: The assessment of water-borne erosion at catchment level using GIS-based RUSLE and remote sensing: A review. Int. Soil Water Conserv. Res., 7, 27–46, https://doi.org/10.1016/j.iswcr.2018.12.002.
Qing, Y., and S. Wang, 2021: Multi-decadal convection-permitting climate projections for China’s Greater Bay Area and surroundings. Climate Dyn., 57, 415–434, https://doi.org/10.1007/s00382-021-05716-w.
Ren, H., Z. Wen, Y. Liu, Z. Lin, P. Han, H. Shi, Z. Wang, and T. Su, 2023: Vegetation response to changes in climate across different climate zones in China. Ecol. Indic., 155, 110932, https://doi.org/10.1016/j.ecolind.2023.110932.
Riahi, K., and Coauthors, 2017: The Shared Socioeconomic Pathways and their energy, land use, and greenhouse gas emissions implications: An overview. Global Environ. Change, 42, 153–168, https://doi.org/10.1016/j.gloenvcha.2016.05.009.
Schlund, M., A. Lauer, P. Gentine, S. C. Sherwood, and V. Eyring, 2020: Emergent constraints on equilibrium climate sensitivity in CMIP5: Do they hold for CMIP6? Earth Syst. Dyn., 11, 1233–1258, https://doi.org/10.5194/esd-11-1233-2020.
Shangguan, W., Y. Dai, Q. Duan, B. Liu, and H. Yuan, 2014: A global soil data set for Earth system modeling. J. Adv. Model. Earth Syst., 6, 249–263, https://doi.org/10.1002/2013MS000293.
Song, X., D.-Y. Wang, F. Li, and X.-D. Zeng, 2021: Evaluating the performance of CMIP6 Earth system models in simulating global vegetation structure and distribution. Adv. Climate Change Res., 12, 584–595, https://doi.org/10.1016/j.accre.2021.06.008.
Sun, X., M. Xue, J. Brotzge, R. A. McPherson, X.-M. Hu, and X.-Q. Yang, 2016: An evaluation of dynamical downscaling of central Plains summer precipitation using a WRF-based regional climate model at a convection-permitting 4 km resolution. J. Geophys. Res. Atmos., 121, 13 801–13 825, https://doi.org/10.1002/2016JD024796.
Tang, Q., Y. Xu, S. J. Bennett, and Y. Li, 2015: Assessment of soil erosion using RUSLE and GIS: A case study of the Yangou watershed in the Loess Plateau, China. Environ. Earth Sci., 73, 1715–1724, https://doi.org/10.1007/s12665-014-3523-z.
Tian, P., and Coauthors, 2021: Soil erosion assessment by RUSLE with improved P factor and its validation: Case study on mountainous and hilly areas of Hubei Province, China. Int. Soil Water Conserv. Res., 9, 433–444, https://doi.org/10.1016/j.iswcr.2021.04.007.
Van Oost, K., and J. Six, 2023: Reconciling the paradox of soil organic carbon erosion by water. Biogeosciences, 20, 635–646, https://doi.org/10.5194/bg-20-635-2023.
Wang, X., T. A. Quine, H. Zhang, G. Tian, and W. Yuan, 2019: Redistribution of soil organic carbon induced by soil erosion in the nine river basins of China. J. Geophys. Res. Biogeosci., 124, 1018–1031, https://doi.org/10.1029/2018JG004781.
Wang, Z., Y. Zeng, C. Li, H. Yan, S. Yu, L. Wang, and Z. Shi, 2021: Telecoupling cropland soil erosion with distant drivers within China. J. Environ. Manage., 288, 112395, https://doi.org/10.1016/j.jenvman.2021.112395.
Wang, Z., X. Li, Y. Mao, L. Li, X. Wang, and Q. Lin, 2022: Dynamic simulation of land use change and assessment of carbon storage based on climate change scenarios at the city level: A case study of Bortala, China. Ecol. Indic., 134, 108499, https://doi.org/10.1016/j.ecolind.2021.108499.
Xin, X., T. Wu, W. Jie, and J. Zhang, 2021: Impact of higher resolution on precipitation over China in CMIP6 HighResMIP models. Atmosphere, 12, 762, https://doi.org/10.3390/atmos12060762.
Xue, J., D. Lyu, D. Wang, Y. Wang, D. Yin, Z. Zhao, and Z. Mu, 2018: Assessment of soil erosion dynamics using the GIS-based RUSLE model: A case study of Wangjiagou watershed from the Three Gorges Reservoir Region, southwestern China. Water, 10, 1817, https://doi.org/10.3390/w10121817.
Yang, Y., Q. Zhu, J. Liu, M. Li, M. Yuan, H. Chen, C. Peng, and Z. Yang, 2020: Estimating soil organic carbon redistribution in three major river basins of China based on erosion processes. Soil Res., 58, 540–550, https://doi.org/10.1071/SR19325.
Yue, Y., and Coauthors, 2016: Lateral transport of soil carbon and land–atmosphere CO2 flux induced by water erosion in China. Proc. Natl. Acad. Sci. USA, 113, 6617–6622, https://doi.org/10.1073/pnas.1523358113.
Zhang, H., and Coauthors, 2020: Simulating erosion-induced soil and carbon delivery from uplands to rivers in a global land surface model. J. Adv. Model. Earth Syst., 12, e2020MS002121, https://doi.org/10.1029/2020MS002121.
Zhu, J., G. Huang, X. Wang, G. Cheng, and Y. Wu, 2018: High-resolution projections of mean and extreme precipitations over China through PRECIS under RCPs. Climate Dyn., 50, 4037–4060, https://doi.org/10.1007/s00382-017-3860-1.
