Numerical modeling of the effects of soil moisture changes on ecosystems in the study of plant and vegetation ecology in arid zones

Xinjiang is located in northwestern China, in the heart of the Asian-European continent, its geographical position is 73°20′-56°25′E, 34°15′-49°10″N, belonging to the temperate continental climate. The northern and southern regions of Xinjiang are bounded by the Tianshan Mountains, and there are obvious differences in topography and climatic conditions: the northern part of Xinjiang has warmer weather, more abundant rainfall, a large number of plants and semi-fixed deserts, with deserts dominating the overall land type; the southern part of the region has very hot weather, less rainfall, and sparse grass and trees, and it is mostly a fluid desert.

The annual evapotranspiration in the Xinjiang region is 2257*10⁸t, and there is about 13797*10⁸t water vapor in the atmosphere each year, which is about 1/5 and 1/3 of that in the Yangtze and Yellow River Basins.It can be seen that the deficit in water vapor content in Xinjiang, coupled with the fact that the proportion of precipitation (snow) formed by the water vapor is only 17. 6% of the population has made Xinjiang an arid region with scanty rainfall, which is precisely the most fundamental climatic feature of this region. Furthermore, drought has limited the local economic and social development. Under the double influence of global warming and human activities, the losses caused by drought disasters in Xinjiang are increasing. Therefore, it is very important to study the soil water utilization efficiency in Xinjiang for the scientific management of the ecosystem in Xinjiang.

The remote sensing data [18] and reanalysis data used in this study are shown in Table 1.

MODIS products from 2001 to 2020 were selected as the primary data source for this study from the Distributed Activity Archive Center (DAAC) of the Level 1 Atmospheric Archives and Distribution System (LAADS).MOD17A2 provides GPP data with a temporal resolution of 8 days and a spatial resolution of 500 m. The reliability and accuracy of the data product has been evaluated by many studies and compared with ground-based data from a variety of terrestrial ecosystems.The GPP data contained invalid values, and data with values greater than 3000 were considered as no-data areas and were removed, and the 8-day data were synthesized into monthly- and yearly-scale data.

In this study, daily soil volumetric moisture data (m³/m³) were collected from ERA5 at a spatial resolution of 0.25° from 2001 to 2020 to characterize soil moisture. A bilinear interpolation method was used to resample the soil moisture data to 500 m resolution, which is consistent with the GPP data, to exclude the data in the region with no GPP value, and to merge the daily-scale data into monthly- and yearly-scaled data.ERA5 is the fifth and the most advanced global reanalysis product released by the European Center for Medium-Range Weather Forecasts (ECMWF), which adopts the latest model cycle (Cy41r2), and is comparable to ERA-The main improvements of ERA5 over Interim are its higher spatial and temporal resolution, the incorporation of more historical observations, and enhanced bias correction. Previous studies have found that the ERA5 soil moisture product shows good agreement with in situ observations in terms of seasonal variability. To further evaluate the usability of the ERA5 soil moisture dataset, we evaluated the ERA5 dataset using six layers of soil relative humidity data observed at 52 meteorological stations in Xinjiang from 2001 to 2013 at depths of 20 cm, 40 cm, 60 cm, 80 cm, 100 cm, and 130 cm. ERA5 provides five layers of soil moisture data at depths of 0-20 cm, 20-40 cm, 40-60 cm, 60-80 cm, 80-100 cm, and 100-130 cm, and in this study, the soil moisture from these six layers was totaled to obtain the total soil moisture data at depth, SW (mm) [19].

AI is defined as the ratio of precipitation (P) to potential evapotranspiration (PET). The gridded monthly P and PET data (spatial resolution 0.5°) from 2001 to 2020 used in this study were obtained from the Climate Research Unit Time Series Dataset (CRUTSv4.05). Among them, PET was calculated by the Penman-Monteith formula using gridded temperature, vapor pressure, and cloudiness data. In this study, the reliability of the precipitation data from CRU was verified by precipitation observations from 34 meteorological stations in Xinjiang from 2001 to 2020, and the reliability of the potential evapotranspiration (PET) data was verified by searching the literature. Comparing the CRU dataset with other similar gridded datasets, the results show significant consistency between the dataset and the sparse observations. The potential evapotranspiration was calculated based on the observation data from 66 meteorological stations in Xinjiang, which proved the reliability of the PET data from CRU. In this study, a bilinear interpolation method was used to resample the meteorological data to 500 m resolution, which was consistent with the GPP data, to exclude the data in the region without GPP values, and to merge the monthly-scale data into annual-scale data.

Vegetation ecological water demand is mainly affected by vegetation type, soil moisture and climatic elements, this study is based on the theory of vegetation ecological water demand calculation proposed by the Food and Agriculture Organization of the United Nations (FAO), and the vegetation ecological water demand (EWR) was calculated under different vegetation cover types and soil moisture [20], with the formula: 1EWR=Ap⋅∑n=1NETn⋅10-3 where EWR is the ecological water requirement (m³), A_p is the area (m²), and ET_n the ET on day n (mm/d).

Under the given climatic conditions, if an area has extensive land, sufficient fertility and suitable soil moisture to promote vegetation growth to standard healthy production conditions and is free from pests and diseases, the rate of evapotranspiration from vegetation, according to FAO recommendations, can be determined by the following equation: 2ETc=Kc⋅ET0 where ET_c is the evapotranspiration rate of vegetation under standard conditions (mm/day), K_c is the vegetation coefficient, and ET₀ is the potential evapotranspiration (mm/day).

Under non-standard conditions, vegetation growth is influenced by soil moisture content, and when soil moisture content falls below a defined threshold, vegetation will be subject to water stress, quantifying this effect as the soil moisture limitation factor (K_s). In this case, vegetation evapotranspiration is calculated as follows: 3ET=Kc⋅Ks⋅ET0 where ET is the vegetation evapotranspiration rate (mm/day) under non-standard conditions; and K_s is the soil moisture limitation factor.

Potential evapotranspiration (ET₀) was calculated using the revised Penman-Monteith equation recommended by the FAO, and potential evapotranspiration (ET₀) was calculated using measured meteorological data with the following equation: 4ET0=0.408Δ(R-G)+γ900T+273U2(ea−ed)Δ+γ(1+0.34U2) where ET₀ is the potential evapotranspiration (mm), Δ is the slope of the tangent to the temperature/saturated water vapor pressure relationship curve at temperature T (kpa/°C), T is the mean air temperature (°C), R is the net radiation (MJ/(m²*d)), and G is the soil heat flux (MJ/(m²*d)).

Crop coefficients were calculated using the segmented single-valued average crop coefficient method to estimate crop coefficients, and vegetation growth stages can be divided into four stages: early growth stage, rapid vegetation development and growth stage, mid-growth stage, and late growth stage. According to the recommendation of FAO-56, the crop coefficients for the full growth stage of vegetation mainly consist of vegetation coefficient K_e,ini for the early growth stage, vegetation coefficient K_e,mid for the middle growth stage and vegetation coefficient K_c,end for the late growth stage.

The water consumption in the early vegetative growth stage is mainly from soil moisture evaporation because of the low vegetative cover, which is determined K_e,ini according to the wetting interval and Eto reference FAO-56 standards.

K_e,mid There are two types, in which the non-agricultural land area is calculated K_e,mid through the vegetation cover (f_p) in combination with other meteorological factors with the following formula: 5Kc,mid=Kc,min+[ Kc,full−Kc,min ](min[ 1,2,fp,(fp,eff)11+h ])

Where K_c,min is the minimum vegetation coefficient for bare soil in the absence of vegetation, ranging from 0.15 to 0.2; f_p is the actual vegetation cover on the surface, obtained by inversion of MODISNDVI data products; K_e,fall is the vegetation coefficient under full cover conditions; and f_p,eff is the effective vegetation cover. 6Kc,full=Kc,h+[0.04(u2−2)−0.004(RHmin−45)](h3)0.3 $${K_{c,full}} = {K_{c,h}} + \left[ {0.04\left( {{u_2} - 2} \right) - 0.004\left( {R{H_{\min }} - 45} \right)} \right]{\left( {{h \over 3}} \right)^{0.3}}$$

Where, K_c,h is the coefficient of full-cover vegetation under standard wind speed and humidity, K_c,h = 1.0+0.1h, where, h is the height of vegetation (m), u₂ is the wind speed at the ground level of 2 meters, and RH_min is the daily minimum relative humidity, which are all calculated from standard meteorological data. 7fp,eff=fpsin(n) 8sin(n)=sin(φ)sin(δ)+cos(φ)cos(δ) 9δ=0.409sin(2π360J−139) $$\delta = 0.409\sin \left( {{{2\pi } \over {36\,0}}J - 139} \right)$$

Combine the meteorological data with the standard crop coefficients given by FAO to calculate K_e,mid and K_e,cnd with the following formula: 10Kc,mid=Kc,mid(table)+[0.04(u2−2)−0.004(RHmin−45)](h3)0.3 $${K_{c,mid}} = {K_{c,mid\left( {table} \right)}} + \left[ {0.04\left( {{u_2} - 2} \right) - 0.004\left( {R{H_{\min }} - 45} \right)} \right]{\left( {{h \over 3}} \right)^{0.3}}$$ 11Kc,end=Kc,end(table)+[0.04(u2−2)−0.004(RHmin−45)](h3)0.3 $${K_{c,end}} = {K_{c,end\left( {table} \right)}} + \left[ {0.04\left( {{u_2} - 2} \right) - 0.004\left( {R{H_{\min }} - 45} \right)} \right]{\left( {{h \over 3}} \right)^{0.3}}$$

K_c,mid(table) and K_c,cnd(table) are the calculated initial values of different vegetation at different stages, which are found in the FAO-56 standard document according to the actual situation.

According to the meteorological data of the study area and the actual investigation, the vegetation in the study area starts to grow and develop in April, and the growth and development of vegetation basically stops in October.

Soil moisture limiting factor K_s is calculated as: 12Ks=ln(S−SwS*−Sw×100+1)ln(101) where S denotes the actual soil moisture content (m³/m³) ; S_W denotes the soil wilting coefficient; and S* is the critical soil moisture content (m³/m³), which is usually 70% to 80% of the field water holding capacity.

It is extremely difficult to obtain higher resolution soil moisture data at the monthly scale in the study area, therefore, in this study, the temperature drought vegetation index (TVDI) was selected as the independent variable for mathematical modeling based on the SM dataset in order to obtain a soil water content model applicable to Xinjiang. Compared with models that characterize the degree of land aridity, such as apparent soil thermal inertia and crop water stress index, the advantages of using TVDI to invert soil water content are: the mathematical model form is simple; the model parameters can be obtained directly from remote sensing data; it can meet the requirements of use in a large scale with high temporal resolution; and the spatial resolution of inversion using spectral data can reach a high level. 13TVDI=Ts−TminTmax−Tmin where T_s is the surface temperature, T_max is the dry-side equation, and T_min is the wet-side equation.

The critical soil moisture content (θ_c) is taken as 14.5%, the soil wilting coefficient (θ_c) is taken as 7.7%, combined with the calculated soil moisture content θ, the soil moisture limitation coefficient Ks of each site is calculated by using the formula, and the average value of Ks of each site is taken as the soil moisture limitation coefficient of each region for many years, and the soil moisture limitation coefficient is shown in Table 2. Then use the spatial analysis function of Arc GIS software to interpolate it, and get the spatial distribution of the multi-year average of soil moisture limitation coefficient in the whole Xinjiang region as the basis for understanding the soil moisture limitation coefficient in each region. As can be seen from Table 2, the multi-year average value of soil moisture limiting coefficient in the whole Xinjiang region varies between 0.344 and 0.402, with a small range of variation. The maximum value of soil moisture limitation coefficient is 0.402, which occurs in Zhaosu County in the central north, and the minimum value is 0.344, which occurs in Ruoqiang County in the south.

According to the FAO recommended values of the relevant plant families, combined with the local meteorological data, relevant calculations were carried out to obtain the plant coefficients of the three types of plants at different fertility stages across Xinjiang as shown in Table 3, with the serial numbers 1, 2, and 3 representing the three types of plants of grasslands, shrubs, and trees, respectively. 1)

In the whole Xinjiang region, the plant coefficients of the early stage of grass growth ranged from 0.1 to 1.05, of which only seven counties, including Iwu, Ruoqiang, Bachu, Minfeng, Pishan, Tazhou Kurgan and Hotan, had coefficient values less than 0.2, and Zhaosu had a large coefficient value of 1.08. In the early stage of shrub growth, the plant coefficients of the early stage of grass growth ranged from 0.1 to 1.08, and in Zhaosu, a coefficient value of 0.1 to 1.08 was observed. In Zhaosu, there was a great value of 1.08, and the initial plant coefficient of trees changed in the same way as that of shrubs.

The multi-year average values of soil moisture limitation coefficient, plant coefficient and other parameters have been derived in the paper, based on the above data, the formula was used to obtain the ecological water demand of vegetation per unit area of various types of vegetation in the whole Xinjiang region in the full reproductive stage, and the specific values are shown in Table 4.

Table 4 shows that the ecological water demand per unit area varies from 51-106 mm in the full fertility stage of grasses, from 125-247 mm in the full fertility stage of shrubs, and from 181-393 mm in the full fertility stage of trees.

Vertical changes in soil water content in different types of arid zones are shown in Figure 1. Soil water content in different types of arid zones varies with depth, showing obvious hierarchical characteristics. Generally, the soil water content in the surface layer is low. It gradually increases with the increase of soil depth, but decreases after a certain depth layer. The study shows that the thickness of the dry sand layer in the surface layer of the sandy land in this region is 0-30cm, the layer of drastic change in moisture is 50-80cm, and the layer of relative stability in moisture is 80-130cm.

Figure 1.

Vertical distribution of soil moisture content in dry areas

This vertical change rule for soil moisture in this area is basically consistent with the conclusion of the study on the stratification law of moisture change in arid areas.

Table 5 shows the coefficients of variation in water content of different types of soil layers. It can be seen from Table 5 that the coefficient of variation gradually decreases with the increase of soil layers, which is because under the same standing conditions, rainfall has a greater effect on the water content of the surface soil than on the deep soil, and rainwater infiltration during rainfall increases the water content of the surface soil rapidly, and after rainfall, due to evaporation from the ground and downward infiltration of water under the effect of gravity, the water content decreases, which results in larger fluctuations in the water content of the surface soil, and the coefficient of variation is thus significantly larger than that of the water content of the lower soil. The coefficient of variation was significantly greater than the coefficient associated with lower soil moisture content.

In this study area, due to the thicker layer of dry sand, the drastic change in the moisture of the sandy soil layer was shifted downward. However, in terms of the difference in water content between each layer, the largest difference was found between the 80 cm depth layer in the dry zone and the control site. This is because this layer is the main distribution layer of the root system of most sandy shrub plants, and there is an obvious decrease in soil moisture due to the strong absorption effect of the root system. In the mobile sandy land, due to the sparse vegetation, the utilization of water is relatively less, coupled with the inhibiting effect of the surface dry sand layer on the evaporation of the lower water capillary, thus making the soil coefficient of variation of the 80 cm depth layer in the mobile sandy land higher, which is 0.5676. Overall comparison, the soil moisture content of different types of arid zones in descending order is as follows: mobile sandy land > sandy land with two years of drought > sandy land with five years of drought > naturally restored land > artificial + naturally restored land > artificially restored land.

The pearson correlation coefficients between soil moisture contents of different soil depths are shown in Table 6, and the correlation analysis shows that the correlation coefficients between the moisture contents of the surface soil (0-20cm) and the soil moisture contents of the layers below it in the sandy land are decreasing in order. And the closer to the bottom layer, the less soil moisture is affected by surface soil moisture. Among them, the correlation coefficients between the layers of sandy land in 5 years of drought were decreasing, while the correlation coefficients of other types were decreasing first and then increasing. ** and * indicate significant (two-tailed) at the 0.01 and 0.05 levels, respectively.

In the process of vegetation restoration in arid sandy land, the soil moisture content changed accordingly with the increase of drought years. Figure 2 shows the change in soil moisture content in arid sandy land during different drought years during vegetation restoration. Under the same standing conditions, the water content of all soil layers in the sandy land with 5 years of drought was significantly lower than that in the sandy land with 2 years of drought, especially in the depth layer of 60-80 cm, the decrease was most significant, with a difference of up to 5 g.kg^-1. Regardless of the duration of drought, the soil water content of all layers in dry sandy land was lower than that in mobile sandy land. This is because with the increase of drought duration, the water consumption of vegetation also increases, especially in the drought period, the strong transpiration of plants leads to the gradual decrease of deep water storage due to the role of supplemental regulation of plant water demand, and the amount of precipitation is not large enough to make the soil surface layer of the water deficit can be supplemented to a certain extent, but due to the limitation of precipitation, it is difficult to make the deep soil moisture to be effectively supplemented, and thus make the deep soil water content decrease. In a prolonged dry climate, it may also lead to the emergence of a soil drying layer. This indicates that as the number of years of drought increases, the utilization of soil moisture by vegetation increases, resulting in a tendency for soil moisture to decrease.

Figure 2.

Soil moisture content of different dry years

In total, 70 herbaceous plant species were found and identified in the surveyed samples in the study area, and the box plots of the distribution of plant species richness in shrublands (n=18), grasslands (n=29), damlands (n=3) and other grasslands (n=26) in the study area are shown in Figure 3. Herbaceous plant species richness was significantly different (P<0.05) between restored shrubland and restored grassland, and shrubland species richness was significantly higher than grassland communities. The topography of the Loess Plateau region is complex, with long gullies and ravines, and a large number of silt dams have been constructed to minimize soil erosion. In order to test whether this kind of ecological engineering measures can maintain a high level of biodiversity, the investigated grassland sample plots (n=29) were divided into two groups: dam land (n=3) and other grassland (n=26). The results showed that there was a significant difference (P<0.05) in species richness between the dam site and other grasslands, and the species richness of the dam site was significantly higher than that of the other sloping grassland communities, with the mean value of species richness reaching 17.67. The reason is that the dam site is generally located in the lower elevation of the watershed, and a large amount of plant seeds, nutrients and water are enriched in the soil with runoff, so the dam site provides environmental conditions that can maintain a higher level of species diversity.

Figure 3.

A box diagram of the distribution of species richness