Gravel parameterization scheme and verification using BCC_CSM

The soil in China contains an abundance of gravels, but it is poorly described in land surface models. To solve this problem, the Beijing Climate Center Atmosphere–Vegetation Interaction Model (BCC_AVIM), which is a land surface model with the gravel parameterization, is coupled to the Beijing Climate Center Climate System Model (BCC_CSM). The simulation ability of BCC_CSM for China using the gravel parameterization is evaluated by comparing the simulation results using default and new schemes with the observed data. The results show that the annual average surface temperature simulated with the new schemes is more consistent with the observation in terms of spatial distribution, and the simulation results are significantly improved, especially in summer. From the perspective of the area-averaged variables, more precipitation simulated using the default schemes is improved except for summer. The high-level and low-level wind fields simulated by BCC_CSM significantly improve the Qinghai-Tibet Plateau. In general, this gravel parameterization is more suitable for areas with high gravel content, and it improves the simulation performance of BCC_CSM in some areas of China.


Introduction
The medium-resolution Beijing Climate Center (BCC) Climate System Model version 2 (BCC_CSM2_MR) is a global climate system model coupled with atmosphere, land surface, ocean, and sea-ice components, which participated in the Coupled Model Intercomparison Project Phase 6 (CMIP6; Wu et al. 2014a). The land surface accounts for about 30% of the global surface area, so the study of land surface models and land-air interactions is far-reaching (Zhang 1998;Dai and Zeng 1996). Coupling land surface models with climate system models by sensitivity tests have verified that different land surfaces have different influences on climate (He et al. 2017). Therefore, optimizing the land surface models and coupling them to the climate models is one of the important ways to improve the climate models (Sun 2002).
Stony soils are widely distributed in China, mainly concentrated in the plateaus as well as northwestern and northern mountainous areas (Fu et al. 2001;Hou 1993;Chen et al. 2009;Gao et al. 2011). The gravel firstly affects the physical properties of the soil, such as soil porosity and soil hydraulic conductivity, which in turn alters the hydrothermal transport in the soil (Mehuys et al. 1975;Hanson and Blevins 1979;Poesen and Lavee 1994). With the rapid development of numerical models, many studies have attempted to simulate the role of gravels on hydrothermal transport processes in the soil. In the Water Erosion Prediction Project (WEPP) model, for example, for describing the porosity of gravelcontaining soils, gravels can be considered non-porous spheres. The presence of gravel reduces the porosity and the available water content of the soil (Alberts et al. 1995). Different models have also been used to simulate the water infiltration in the soils with gravels, and it has been concluded that the water infiltration in soils decreases with the increasing content of gravels (Cousin et al. 2003;Ma and Shao 2008). After simulating at a single site (Naqu station) using the CoLM model, Luo et al. (2009) raised that ignoring the gravel component of the soil may cause some biases in the simulation results. Pan et al. (2015) pointed out that the presence of gravels causes changes in the capacity of mixed soil, which alters the thermal conductivity of the soil and ultimately affects the soil temperature. Yi et al. (2013) considered soils as a mixture of fine soils and gravels. They analyzed the process of multi-year permafrost on the Qinghai-Tibet Plateau after adding the impact of gravels on soil hydrothermal properties to a terrestrial ecological model. The simulated soil temperature and humidity were more accurate in the Bei Lu River.
The previous gravel parameterization formula has been revised to form a gravel parameterization scheme suitable for the BCC_AVIM land surface model. The soil hydrothermal process has been simulated using the observations at the Maduo station as the reference data. (Ma et al. 2020). The results show that the BCC_AVIM land surface model is more accurate after gravel parameterization. It is important to evaluate the applicability of this gravel parameterization scheme for China due to the high gravel content in China, and this is of great significance for improving climate simulations.

BCC_CSM model
The atmospheric model component of BCC-CSM2-MR is BCC-AGCM3-MR with a horizontal resolution of T106 (1.125° × 1.125°), increased vertical stratification of 46 layers, and a model layer top of 1.459 hPa. The description of the model dynamical framework and physical processes and an assessment of the basic model performance can be found in the relevant references (Dong et al. 2009;Guo et al. 2011;Wu 2012). The land surface model component is BCC_AVIM2.0, an atmosphere-vegetation interaction model that can simulate land surface processes. AVIM2, a domestically developed dynamic vegetation and soil carbon cycle model, is introduced based on the physical module in community land surface model version 3 (CLM3) of the National Center for Atmospheric Research (NCAR) land surface model (Ji 1995;Ji et al. 2008). The Modular Ocean Model Version 4, a 40-level ocean model (MOM4_L40), has a horizontal resolution of 1° × 1° with a longitude encrypted to 1/3° in the tropics. The sea ice model, Sea Ice Simulator (SIS), has a horizontal resolution of 1° × 1° with one snow layer and two sea-ice layers of the same thickness in the vertical direction, and the model has the same horizontal resolution as MOM_L40.

Data sources
The current surface data used in BCC_CSM does not include the gravel parameter. A new soil dataset with the gravel content needs to be created to apply the gravel parameterization scheme to the global climate system model BCC_CSM. A global soil dataset (BNU) containing sand, clay, and gravel are created by Shangguan using seven soil databases spread over the globe (Shangguan et al. 2014) (from http:// globa lchan ge. bnu. edu. cn/ resea rch/ data). In this study, the BNU data with the 10-km resolution is used. It has eight layers in the vertical direction (0-0.045 m, 0.045-0.091 m, 0.09-0.166 m, 0.166-0.289 m, 0.289-0.493 m, 0.493-0.829 m, 0.829-1.383 m, and 1.383-2.296 m). Figure 1 shows the gravel distribution of the BNU data in China. The gravel content is high in the Chinese region, with large values in Fig. 1 Spatial distribution of the BNU gravel content (unit: %) in China the Qinghai-Tibet Plateau region, and the gravel content increases significantly with depth. Gravel content decreases with depth in the southeastern coastal area. There is a wide range of missing gravel data in the shallow 4.5-16.6-cm layer in the Chinese desert region and the 138.3-cm external area of the Qinghai-Tibet Plateau. We update the BNU global soil data set for the BCC model. Details of the methods will be described in the subsequent sections.

Data processing
• Step 1: The BNU global data (including gravel, sand, and clay) is converted into the precision of 320 × 160 to correspond to the default grid points of the BCC model (Wu et al. 2019). • Step 2: The gravel parameter is created in the surface data of the BCC model. The BNU data were interpolated into ten, to match the ten layers required by the BCC model (node depths of 0.0071 m, 0.0279 m, 0.0623 m, 0.119 m, 0.212 m, 0.366 m, 0.620 m, 1.038 m, 1.728 m, and 2.865 m, using a bilinear interpolation method). • Step 3: Missing gravel data in the BNU dataset, replaced with data from adjacent layers or grid points • Step 4: The sand and clay data in the BNU data will be replaced by the corresponding content in the model default data if they are missing measurements. • Step 5: However, this method may cause the sum of sand, clay, and gravel contents to be greater than 100% at some point, contrary to the fact. To solve it, the values where the sum of the three exceeds 100% reduce in the same proportion. (Shangguan et al. 2014) Figure 2 shows the updated soil gravel data in China. Compared with the original data, the distribution of gravel content and the location of the large-value areas remain. The gravel content is high and widespread in China, with the large-value center mainly in the Qinghai-Tibet Plateau, where the gravel content increases with the increase of depth. The deep gravel content is over 50%, and the mediumdepth and shallow gravel content are relatively uniform. The areas with high gravel content of the deep layer are mainly in the western Qinghai-Tibet Plateau. The shallow gravel content is low in North China, and the content of which increases with depth.

Observational data and reanalysis data
To test the simulation results with the gravel parameterization, the monthly mean wind field data with a horizontal resolution of 2.5° × 2.5° from the National Centers for Environmental Prediction (NCEP) is chosen (Kalney 1996)

Methodology
The gravel-parameterized land surface component-BCC_ AVIM coupled to the BCC_CSM climate system model is completed to simulate global surface air temperature, wind field, and precipitation from January of 2000 to December of 2001 in China. To demonstrate the spatial distribution and temporal variation, China is divided into seven regions: NE, N, SE, ENW, SW, WNW, and Tibet (Fig. 3), taking into account the climate, vegetation, and especially the distribution of gravels shown in Fig. 2 (Wang et al. 2011). Taylor diagrams are used to visualize the performance of simulated temperature, precipitation, and wind wield under different regional default and new schemes (Taylor 2001).

BCC_CSM default parameterization
The BCC_AVIM soil hydraulic property parameterization scheme is based on the studies from Clapp and Hornberger (1978) and Cosby et al. (1984). The soil thermal property parameterization scheme is based on the work of Farouki (1981). In each layer, the hydraulic conductivity is a function of the saturated hydraulic conductivity, the soil moisture, and the porosity, as well as the B parameter.
Where the k is the hydraulic conductivity, k sat is the saturated hydraulic conductivity, θ is the liquid volumetric soil moisture, θ sat is the porosity, ψ sat is the saturated soil matric potential (Clapp and Hornberger 1978). Where the B min is the exponent B of the mineral soil, θ sat, min is the porosity of the mineral soil, Ψ sat, min is the saturated mineral soil matric potential, κ sat, min is the saturated hydraulic conductivity for mineral soil (Cosby et al. 1984).
The BCC_AVIM soil thermal property parameterization scheme is proposed by Farouki (1981). Among them, soil mineral thermal properties include soil mineral thermal conductivity, dry soil mineral thermal conductivity, soil capacitance, and soil mineral thermal fusion. Where λ s, min is the solid thermal conductivity of mineral soil, λ dry, min is From the above equations, it can be seen that the role of gravel in soil hydrothermal process is not considered in BCC_CSM, especially in the Qinghai-Tibet Plateau region where the gravel content is high, and ignoring gravel brings errors to the simulation results. Therefore, considering the role of gravel in soil hydrothermal, the original parameterization scheme is modified.

Impacts of gravel on the soil hydraulic properties
The effect of gravel on the saturated soil water content, saturated soil hydraulic conductivity, saturated soil matrix potential, and parameter B is mainly considered in the new scheme, and Eqs.
(3)-(6) are modified as follows: Where the soil saturated water content takes into account the effect of gravel (Eq: θ sat, min = (1 − V g )θ sat, f + V g θ sat, g ), the water content is low and negligible when the gravel is not highly weathered, so the formula is approximated as θ sat, min = (1 − V g )θ sat, f (where the θ sat, f is the saturated water content of fine soil, θ sat, g is the saturated water content of gravel, and V g is the volume content of gravel) (Poesen and Lavee 1994). Peck and Watson (1979) propose an equation to calculate the conductivity of a spherical body embedded in the medium. Assuming that the conductivity after being embedded in the medium is much lower than that the conductivity without the spherical body, the saturated hydraulic conductivity for mineral soil is modified as follows, Cousin et al. (2003) have found that if only the gravel volume is considered and its water-holding properties are ignored, the calculated effective water volume is underestimated by 34%. Considering the role of gravels on soil waterholding properties is essential to accurately model the water transport in gravel-containing soils. The mixed soil matric potential can be calculated using the following equation: is the matric potential of saturated gravels and set to −1.3 mm (Pan et al. 2015).
In BCC_ AVIM, the pore-size distribution coefficient (b m , unitless) of mineral soils is an important dimensionless parameter for describing the hydraulic properties of soils. The magnitude of b m is related to the water-holding capacity of the soil. The b f of fine soils increases with the clay content in the soil. In this experiment, b g is set to 7.5 in the equation due to the effect of gravel parameterization on b m (the pore distribution coefficient is set to 3 for sandy soils and 12 for clay soils).

Impact of gravels on the soil thermal properties
In this subject, the effect of gravel on soil bulk density, soil solid thermal conductivity, and solid soil heat capacity is mainly considered, and Eqs. (7)-(9) are modified as follows: Russo (1983) proposed using the gravel volume content, gravel capacity, and fine soil capacity to calculate the capacity of gravelly soils. ρ b = ρ n (1 − V g ) + 2650V g (ρ n = 2700 × (1 − θ sat, f ) is the fine soil capacity.
Due to the lack of soil samples, the above method was simplified to only consider quartz as a rock-forming mineral mineral, respectively. f q and f o are the volume fraction of quartz and other rock-forming mineral, respectively.
The average heat capacity of gravel is in the range of 2.1~2.5 × 10 6 Jm −3 k −1 . Therefore, the average volumetric heat capacity of gravel in the new formula for calculating the heat capacity of soil minerals is taken as 2.2 × 10 6 Jm −3 k −1 (Pan et al. 2017).
In addition to the above, the dry thermal conductivity of mineral soils is modified using the work of Côté and Konrad (2005) ( dry,i = × 10 − sat,m ) as follows: where χ (w·m −1 ·K −1 ), η (unitless) are empirical parameters used to calculate the different soil types. In Côté scheme, the values of χ and η for gravels are 1.70 and 1.80, respectively; for natural mineral soils, χ and η are 0.75 and 1.20, respectively; for fibrous soils, χ and η are 0.3 and 0.87, respectively. In this paper, we consider the general application of this scheme for different soils, so χ and η are 0.917 and 1.29, respectively, taken as the average values of the three soils.
Based on the above theory about the effect of gravel on soil hydrothermal properties, the soil schemes in the default model are modified as follows for new parameterization schemes with the impact of gravels (see Table.1). × 10 6 the average of 2 years, and the gridded surface temperature dataset is used as the reference data (0.5° × 0.5°). The simulated surface temperature values with a resolution of 1.125° × 1.125° are interpolated by bilinear interpolation to the grid points of the reference data. As shown in Fig. 4c, the significant value area of temperature in China is in the southeastern coastal region and gradually decreases inland. Compared with the observed data, the annual mean surface temperature simulated with the new schemes is more consistent with the observed spatial distribution than the default schemes. Due to differences in latitude and topography, the temperature differs considerably in China. In spring and autumn, the differences between simulated and observed values of the two schemes are not significant. The simulated surface temperature changes significantly in summer and winter after adding gravels. In winter (Fig. 5d), the minimum temperature in China appears in the northeast, with the average temperature below −20 °C, and the simulated value of the model is lower compared with the observed data ( Fig. 5h and l). However, the difference between the simulated value with the new schemes and the observed value is significantly reduced after adding the gravel parameterization except region III. The reason is that the soil exerts heat to the outside in winter, and the temperature in the deep soil layer is high, so the heat is transferred to the shallow layer (Chen et al. 2009). The addition of gravel increases the thermal conductivity of the soil (Ma et al. 2020). The heat transfer from the deep soil to the surface layer gets greater, increasing the surface soil temperature and making the near-surface temperature higher through interaction between the ground and the atmosphere. In summer, as shown in Fig. 5b, the simulated temperatures in most regions are about 2 °C higher than the observed data, but it is negative in the region of I. The soil absorbs heat from the outside in summer, so the heat is transferred from the surface to the deeper layers of the soil (Chen et al. 2009), leading to soil thermal conductivity increasing after adding gravels and the surface temperature decreases. So, the difference between the new schemes and the observed value decreases in the positive area, the new schemes perform better, and the difference increases in the region of I where the difference is negative.

Verification of the surface temperature simulated with the new and default schemes
In terms of regional mean temperatures (Fig. 6), the BCC_CSM model simulated temperatures have warm deviations throughout the year, except for region I. The simulation is better with the addition of gravel, especially in summer. The surface temperature decreases after adding the gravel parameterization. The reason is that this gravel parameterization scheme may have affected the plants. In addition to the reasons analyzed in the previous paragraph, the addition of gravel affects thermal conductivity. Jackson et al. (1972), Danalatos et al. (1995), andHeisner et al. (2004) reported that gravel has a positive effect on vegetation growth, as gravel mulch was found to contribute to both soil water conservation and higher soil temperatures, in turn, providing favorable conditions for vegetation growth. Therefore, the surface temperature values simulated by the default model in all regions except I are higher than the observed values. The addition of gravel causes an increase in vegetation, which makes the temperature decrease and is more consistent with the observed values. In contrast, the surface temperature values simulated by the default model in areas I are lower than those observed. The soil temperature decreases after adding gravel, which is ineffective, and more detailed soil water and heat transport processes need to be addressed to be further investigated.

Verification of the precipitation simulated with the new and default schemes
The spatial distribution of precipitation and the location of rain-bands are important indicators for assessing the simulation ability of the model (Wu and Zhang 2012). Gravels can alter the thermal and hydraulic conductivity within the  (Pan et al. 2017;Yamanaka et al. 2004). As shown in Fig. 7., the spatial distributions of the simulated precipitation with the default and new schemes from 2000 to 2001 are consistent with the observation overall. The simulated rainfall in the III coastal region of China is less than 60 mm on average, while the observed surface precipitation is above 100 mm. The precipitation simulated with the gravel parameterization shown in Fig. 7b increases in region III, reaching 40 mm above and increasing by 10-20 mm than the original schemes; it is more consistent with the observed values shown in Fig. 7c. The differences between spatial distributions simulated with the default and new schemes are insignificant in other areas. Seen from the distributions of precipitation in different seasons in Fig. 8, the model can simulate the seasonal variation and distribution pattern of precipitation in China compared with the observed data. The precipitation reaches the maximum in summer, and spatially, it shows a decreasing trend from Southeast China to Northwest China. Compared with the observed data, the rain band simulated by the model is weaker in summer, but there is more precipitation in Northwest China. The new schemes improve this problem to some extent. In winter (Fig. 8d), precipitation is concentrated in region III. The biases of the simulations with the default and new schemes from the observation in these regions are both large. The simulated values are slightly larger than observed in regions VI and VII, and the bias of the simulated precipitation from the observation gets less with the new schemes.
Seen from the area-averaged precipitation (Fig. 9), compared with the observed data, the simulated precipitation is less in summer but more in winter. The precipitation simulated by the new schemes is less in all seasons except for summer compared with the precipitation simulated with the original schemes. The main reason is the weak precipitation in these seasons. The soil hydraulic conductivity is the leading cause for changes in soil moisture (Mehuys et al. 1975). The gravel increases the soil hydraulic conductivity, decreases the soil moisture, and affects processes such as the evaporative runoff in the surface soil (Pan et al. 2017;Ma Wu et al. 2014b), resulting in a decreasing trend of precipitation. The default model simulates more precipitation than observed is improved, except in summer. The simulation and trend of the maximum summer precipitation are also more consistent with observation. The precipitation in summer is considerable, and the change in soil hydraulic conductivity caused by adding gravels is not the most crucial cause for the soil moisture change. The evaporation and runoff processes are more complicated, and the physical processes need to be further explored. The simulation results over the Qinghai-Tibet Plateau show that the new scheme reduces the summer precipitation on the plateau by about 20 mm. The month with the maximum precipitation shifts from July to August, which is more consistent with the observation. The simulated precipitation is locally improved after adding gravels, and the improvement is more evident in areas with high gravel content (Yuan et al. 2021).

Verification of the wind simulated with the new and default schemes
In total, 850 hPa and 200hPa wind fields are the essential climatic element to characterize the model simulated monsoon circulation (Sun et al. 2003;He et al. 2007). In this The area-averaged precipitation (mm) in eight sub-regions paper, the simulation results in 2000 and 2001 are averaged and compared with NCEP wind field data in four seasons to test the simulation ability of the model before and after adding gravels. Figure 10 shows the wind fields from the NCEP data at 850 hPa in four seasons. The northwesterly monsoon forms in northern China under the influence of the Siberian airflow during autumn and winter (Cao et al. 2014). The simulation results in spring are shown in Fig. 10e and i. After adding gravels, the bias of the simulated wind speed from the NCEP data over the VII region significantly gets smaller than that simulated with the default schemes. The wind speed simulated by the BCC_CSM model in summer ( Fig. 10f and j) is slightly lower than the NCEP data all over China, and the change is not obvious after adding gravels. In autumn, as shown in Fig. 10g and k, the new schemes simulate closer to the NCEP data, with significant improvement in III, V, and VII regions. In winter, the wind speed in China is larger than in other seasons, especially over the VII region, where the wind speed is greater than 10 m/s. The difference between the wind field simulated with the new schemes and the NCEP data decreases in the II, V, and VII regions, compared with the default schemes ( Fig. 10h and l). Figure 11 shows the wind fields at 200 hPa in four seasons, and we can see that the model can reasonably simulate the westerly jets over China and the South Asian High over the plateau in summer. In summer, the wind speed over the plateau increases with an average increase of 2-3 m/s after adding gravels, making the location of the South Asian High move by 2-3 latitudes to the north compared with the default schemes. The wind speed simulated with the new schemes in Northeast China is closer to the NCEP data. In winter, the problem is also improved that the simulated wind speed in the Northeast China region is much larger compared with the NCEP data. However, the new schemes are not effective in region III, probably because of the low gravel content in the region, and the gravel parameterization is developed from the highland subsurface experiment, which may not apply to regions with less gravel (Pan et al. 2017;Ma et al. 2020;Yuan et al. 2021).  Figure 12a and b shows the correlation coefficient and normalized standard deviation results for surface air temperature, precipitation, and wind field, respectively, in winter and summer for eight sub-regions in China.

Statistical comparison between simulation and observation
In winter, except for precipitation in some sub-regions, the simulations correlate 0.6 to 0.9 compared to the observed data in most cases. The new scheme shows a stronger correlation and standard deviation of temperature for the surface temperature, closer to 1 for all eight regions except China I. For precipitation, the correlation is weaker than for temperature over the eight sub-regions, and the correlation and standard deviation show minor differences in the Fig. 11 The same as Fig. 10, but for 200-hPa wind field two schemes. For the wind field, the simulation of the new scheme is excellent in China VII, with a standard deviation closer to 1, but, over China IV, the correlation values of U850 and V200 are weaker.
In summer, all three regions of IV, V, and VI show strong correlations greater than 0.8 in surface temperature simulation, and the standard deviation is approximately 1. The simulation of the new scheme performs better in eight subregions. For precipitation, the correlation is weaker than for temperature over the eight sub-regions. Compared to the new scheme, the default scheme shows a worse correlation, i.e., less than 0.6 for all eight regions, and lower standard deviations. The correlation is better for the low-level wind field than the high level for the wind field. The correlation and standard deviation show significant differences in the two schemes for the wind field of 850 hPa in the eight subregions. In region VI, the correlation of the new scheme is about 0.4, which is twice as high as that of the default scheme. In contrast, the new scheme performs worse for the low-altitude wind fields in regions II and V. Overall, for the entire domain VIII, the new scheme shows a stronger correlation with a normalized standard deviation ratio close to 1. The correlation and standard deviation show slight differences between the two schemes.
In general, the simulation of surface temperature has been improved to a great extent by adding gravel, which has little effect on the simulation of precipitation but has an improvement effect on the whole region of China. The wind speed simulated by the new scheme is poor in some regions but significantly improves in most regions of China, especially the Qinghai-Tibet Plateau region. From the regional perspective, the surface temperature, precipitation, and wind field are improved to different degrees after the gravel parameterization in the China domain.

Conclusions
Gravel is an essential component of soils in China, yet its impacts are poorly described in current land surface models. In this study, a surface dataset containing gravel data is firstly created for BCC_CSM, and then BCC_AVIM, which had completed gravel parameterization, is coupled to BCC_ CSM. Finally, the impacts of this scheme on the simulation results of meteorological elements in eight sub-regions of China are discussed. The following conclusions are drawn.
Compared with the default schemes, the annual mean surface temperature simulated by the new schemes is more consistent with the observation in terms of the spatial distribution pattern. The surface temperature changes significantly in summer and winter after adding gravels. The new scheme performs better in winter except for southeast China (III) and poorly in summer in the northeast (I).
Regarding regional mean temperatures, the BCC_CSM model simulated temperatures have warm deviations throughout the year in most areas except for the northeast. The addition of gravels reduces surface temperatures and is more consistent with observed data, especially in summer, as shown in the Taylor plot.
Precipitation shows more significant regional variability than temperature. The mean precipitation simulated with the original schemes throughout the year is lower in the southeastern coastal regions. The precipitation simulated with the gravel parameterization is more consistent with the observed data. The precipitation simulated by the BCC model is less in summer and more in winter. The new schemes have improved the overall bias of the original simulations except in summer. The maximum values and trends of summer precipitation simulated with the new schemes have improved in some local areas.
For the low-level wind field simulated by BCC_CSM, the simulation is significantly improved in the plateau region, which is closer to the NCEP data. The average deviation of the simulated values from NCEP data in the new schemes is closer to 1 than in the default schemes. The model can reasonably simulate the westerly jets over China at 200 hPa and the South Asian High over the plateau in summer. The wind field over the plateau increases after adding gravels in summer, with an average increase of 2-3 m/s, making the South Asian High moves 2-3° north compared to the default schemes. The problem of high simulated winter wind speed values for the new schemes in northeast China has been improved.
As the climate system model can only be run on mainframes, the simulation time is not long enough. This paper only briefly tests the impact of this gravel parameterization and does not discuss the physical mechanisms. Based on the sensitivity test, a more in-depth study will be carried out later.