3.1 Annual characteristics of precipitation - soil moisture variation
The soil moisture variation of PSM in 2016 is shown in Fig. 4. It reveals that soil moisture has obvious seasonal trends. The soil from January to March is frozen. The near surface soil moisture recharge is from snowmelt. When the near surface frozen soil starts to thaw, soil at the 20 cm depth is recharged on February 9th, 16th and 26th in 2016. Soil at depths greater than 20 cm remains relatively stable. Frequent precipitation events usually occur from June to November, during which soil moisture changes considerably, and soil moistures at different depths exhibit periodic increase or decrease, regulated by the interplay of precipitation and evapotranspiration. After February 26th in 2016, soil gradually thaws completely. Figure 4 shows that snowmelt can recharge the soil moisture as deep as 160 cm. The soil moisture at 200 cm depth is recharged for the first time after a heavy precipitation event on July 8th in 2016.
According to Fig. 4, the soil moisture content of the upper 200 cm soil layer fluctuates multiple times in 2016. After November, the soil moisture content of the upper 200 cm soil layer fluctuations but DSR is not detected. This is probably due to the error of the EC-5 probe under frozen winter condition. Therefore, the active research period has been revised to a window from April to November each year. Although the precipitation time and intensity are different in the three year period of observation (2016–2018), the variation trends of soil moisture in each season of the three years are basically the same.
3.2 Root distribution characteristics of PSM
Table 1
Vertical distribution of root biomass of PSM
Depth(cm) | Sample 1 | Sample 2 | Sample 2 | Sample 4 | Sample 5 | Average fine root biomass(d༜0.2 cm) | Average total root biomass(g) | Biomass precentage of Whole layer |
0–20 | 1989.31 | 1845.89 | 2000.21 | 1789.45 | 1879.45 | 1600.86 | 1900.86 | 11.14% |
20–40 | 5976.45 | 6120.65 | 5789.34 | 6018.83 | 5729.39 | 5646.93 | 5926.93 | 34.75% |
40–60 | 6783.62 | 6682.50 | 6831.72 | 6391.58 | 6972.91 | 6472.47 | 6732.47 | 39.47% |
60–80 | 192.83 | 189.31 | 194.38 | 179.37 | 189.30 | 29.04 | 189.04 | 1.11% |
80–100 | 184.03 | 198.31 | 187.30 | 179.82 | 179.45 | 9.78 | 185.78 | 1.09% |
100–120 | 168.53 | 157.82 | 165.20 | 177.43. | 165.59 | 0.1 | 164.29 | 0.96% |
120–140 | 166.65 | 155.43 | 158.45 | 168.38 | 158.41 | 0.1 | 161.46 | 0.95% |
140–160 | 188.43 | 167.21 | 155.82 | 163.72 | 154.90 | 0 | 166.02 | 0.97% |
160–180 | 179.78 | 167.66 | 162.60 | 154.72 | 160.80 | 0 | 165.11 | 0.97% |
180–200 | 154.56 | 155.68 | 153.80 | 161.58 | 153.82 | 0 | 155.89 | 0.91% |
200–220 | 152.30 | 159.32 | 143.83 | 152.93 | 155.21 | 0 | 152.72 | 0.90% |
220–400 | 1154.20 | 1042.38 | 1203.75 | 1167.49 | 1212.52 | 0 | 1156.07 | 6.78% |
|
PSM in the study site has one taproot and numerous fine roots. The five excavated PSMs have an average taproot length of 4 m, and the vertical distribution of root reflects the water use range of plants. As shown in Table <span refid="Tab1" class="InternalRef">1</span>, almost all fine roots of PSM are distributed in the shallow soil layer of 0-100 cm depth and the biomass of 0–60 cm root accounts for 85.36% of all the root biomass. This indicates that PSM is effective to utilize soil moisture in shallow layers, but is less effective to utilize soil moisture below 100 cm. Although the main root of PSM has a depth of nearly 4 m, the fine root biomass on the deep taproot (deeper than 100 cm) is very small. Therefore, PSM in this area can be classified as shallow root species. The root distribution also indicates that precipitation is the main water source for PSM in this area. As 90% of the fine roots of PSM are distributed in shallow soil layers (with depths less than 100 cm) in which the water moistures generally fluctuate greatly in a daily and seasonal basis, PSM in the Mu Us sandy land is heavily dependent on precipitation events.
3.2 Water distribution characteristics of individual soil layers
In order to study the degree of soil moisture response to precipitation in individual layers, this research chooses each layer’s soil moisture at the beginning of each month of 2016 as a representative, to observe whether the soil moisture in a specific layer is recharged. Figure 5 shows the soil moistures at depths of 20 cm, 40 cm, 80 cm, 120 cm, 160 cm, and 200 cm at the beginning of each month. From Fig. 5, the soil of PSM exhibits four distinctive layers: an evapotranspiration layer at 0–40 cm depth, a lateral root activity layer at 40–160 cm depth, a relatively dry soil layer at 160–200 cm depth, and a deep soil layer below 200 cm. The soil layer at 160–200 cm depth is usually in a relatively dry state without DSR replenishment, and it will temporarily converted to a relatively wet state when replenishment is obtained either from heavy precipitation events or from soil moisture during the freeze-thaw process. For the 0–40 cm evapotranspiration layer, the soil moisture increases only under the effect of precipitation or snowmelt. Its moisture content decreases rapidly under the interplay of evapotranspiration and infiltration. For the 40–160 cm root activity layer, the soil moisture is recharged from infiltrated water passing through the upper layer, and it gradually decreases under the effects of infiltration and root moisture absorption. Water absorption of the PSM root system is primarily responsible for depleting the soil moisture for soils at 160–200 cm, resulting in a relatively dry state for this layer of soil. Below the 200 cm, the absorption of the PSM root system diminishes because the PSM root system can rarely penetrate a depth greater than 200 cm. Consequently, the moisture content of the soil at the 200 cm depth is higher than that at the 160 cm depth. The deep soil below 200 cm depth is of native sand soil, and Fig. 5 shows that the soil moisture content of this layer is recharged five times under heavy precipitation events in 2016.
Results show that the shallow soil receives soil moisture replenishment from precipitation, which is quickly converted into evapotranspiration or absorbed by the roots of PSM, with a small amount of precipitation that can reach the deep soil layer. The topsoil moisture of PSM forest changes dramatically, and the 120 cm soil layer stores more water than the deep soil layer. The soil water content at a depth of 160 cm is relatively low, and one can conclude that precipitation cannot effectively replenish the deep soil below the 160 cm depth.
3.3 Precipitation and DSR distribution
The total precipitation in 2016 is 506.4 mm, which is higher than the average annual precipitation, thus 2016 is a wet year. As shown in Fig. 6, the precipitation in 2016 is concentrated from July to August, with 86 precipitation events accounted for over the whole year in 2016. The maximum daily precipitation is 137.2 mm on July 10th, and the minimum daily precipitation is 0.2 mm and it occurs multiple days in 2016. According to the distribution of soil moisture curves in each soil layer in Fig. 6, the moisture content of the soil layer below 120 cm layer fluctuates during the freeze-thaw period and the summer rainy season. Before the heavy rain in July 10th, no precipitation could penetrate the 120 cm depth soil layer, and the freeze-thaw water and precipitation are absorbed by the PSM root system. The results show that there are two replenishment processes of soil moisture, the freezing and thawing replenishment process of surface ice and snow deposits accumulated in winter, and the replenishment process of rainy season precipitation, the heavy precipitation has a significant effect on the replenishment of deep layer soil moisture in this region.
There are only four events of detected DSR in the PSM plot over the entire year in 2016, resulting in an annual DSR of 1.4 mm in the PSM plot in 2016. As shown in Fig. 7, all the measurable DSR events occur after September 21th, and they are close to the winter freeze-thaw period, during which PSM usually ceases to grow. This implies that there are essentially no DSR events throughout the growing season of PSM in 2016. In comparison, in the bare sandy land plot, there are six measurable DSR events before the maximum precipitation event of July 10th in 2016. After July 10th in 2016, there are four considerable DSR events, with a total DSR amount of 281.6 mm. The distinctive DSR difference in the PSM forest plot and the bare sandy land plot shows that the PSM forest absorbs almost all the precipitation-induced infiltration in 2016, while the bare sandy land has a considerable amount of DSR at the same year, which accounts for 55.6% of the annual precipitation. This means that groundwater recharge is profoundly affected after vegetation reconstruction using PSM in the Mu Us sandy land, even for a wet year like 2016.
The total precipitation in 2017 is 309 mm (less than the multi-year average precipitation of 400 mm), as shown in Fig. 8, which signifies a dry year. There are 32 times observed precipitation events in 2017, as shown in Fig. 8. The maximum daily precipitation is 22 mm on July 22th, 2017. The soil moisture fluctuation in 2017 is similar to that in 2016. There is a freezing period from January-March, the surface soil freezes, and the soil water content changes in each soil layers are relatively stable. March to April belongs to the freezing and thawing mixed period, frozen water in the soil layer gradually melts. Especially, the surface layer frozen water functions as a reservoir. The frozen water replenishes the soil moisture of each soil layer and the spring snowmelt recharge depth is 140 cm in 2017. April to November in 2017 is a rainy season with dynamic soil water variation. Under the interplay of precipitation-induced infiltration, soil evapotranspiration and vegetation consumption, the soil moisture fluctuates. After the rainy season, the soil begins to freeze again in December. The results show that in 2017, the total precipitation of the PSM forest land is relatively small, and the of a single precipitation is relatively small as well. Before entering the winter season, the precipitation is absorbed by the PSM root system, and no precipitation-induced infiltration can reach the 120 cm depth soil layer until later winter.
In 2017, the amount of DSR in the bare sandy land is 67.6 mm. In contrast, the amount of DSR in the PSM forest land in the same year is only 0.2 mm, as shown in Fig. 9. Since January 1st of 2017, there has been a continuously recognizable DSR signal in the bare sandy land, indicating that the previous year (2016) is a wet year, thus the soil contains a large amount of water and continues to replenish the groundwater reservoir. However, the PSM forest land has a much lower water storage due to the consumption by PSM, and the DSR below the PSM forest has been significantly reduced in 2017. As shown in Fig. 7, the soil moisture fluctuations intensely in the soil layers at 20 cm, 40 cm, and 80 cm depths. The soil layer from 120 cm to 200 cm is relatively stable, and it is only replenished by soil moisture during the freezing and thawing period (March to April). The 120 cm and 160 cm depth soil layers are replenished (on 14 October) due to continuous summer precipitation. The 200 cm depth soil layer remains relatively dry throughout 2017. The annual DSR underneath the PSM forest is only 0.2 mm in 2017, which is much lower than the DSR for the same plot in 2016. The lack of precipitation in 2017 causes a sharp drop in deep soil moisture infiltration.
The total precipitation amount is 239.8 mm (greater than the multi-year average precipitation of 400 mm) in 2018, as shown in Fig. 10, which is a dry year. There are 42 observed precipitation events throughout the year with a maximum daily precipitation of 20 mm on July 21th, 2018. The soil moisture fluctuation in 2018 is similar to those in 2016 and 2017. There is a freezing period in January-March, in which the soil water content changes in each layer are relatively stable. The March and April belong to the freezing-thawing period. When the frozen soil water gradually melts, the soil below the surface layer is replenished by snowmelt, and the spring snowmelt recharge depth is 160 cm in 2018. Consequently, the soil moistures in various layers rise accordingly. April to November is the rainy season. Under the interplay of precipitation-induced infiltration, soil evapotranspiration and vegetation consumption, the soil moisture fluctuates greatly. The shallow soil layer begins to freeze again in December. The layers of intense soil moisture fluctuations are 20 cm, 40 cm, 80 cm, 120 cm, and 160 cm. The soil moisture change at the 200 cm depth is relatively small, and this layer is only recharged from May 14th to June 6th and on September 7th of 2018.
The DSRs of the PSM forest and the bare sandy land plots are respectively 1.2 mm and 66.2 mm in 2018, as shown in Fig. 11. The results show that the total precipitation in 2018 exceeds the average annual precipitation. Multiple precipitations before the growing season of PSM replenish the deep soil layer moisture. The results show that in 2018, there are two precipitation replenishment processes for DSR, one is the freeze-thaw season from May to June, and the second is the precipitation process in October. The latter is caused by intensive precipitation, which replenishes the entire soil layer.
3.3 Soil moisture infiltration rate comparison during different seasons
The soil moisture recharge sources in the experimental area are spring snowmelt and summer precipitation. According to Fig. 5, we can clearly see that the soil water recharge in different seasons varies, especially at the end of winter season and in the summer rainy season. The amount of spring precipitation in this study site is small, and snowmelt moisture is the main water source during the spring season. The germination process of vegetation or seeds in the Mu Us sandy land mainly depends on the water source of accumulated snowfall in winter. With the surface soil temperature increasing, the surface ice gradually melts and infiltrates to deeper soil layer. As shown in Fig. 12(A-B), this study chooses two typical processes for comparison: the snowmelt soil moisture recharge from February 26th to March 27th of 2016, and the precipitation recharge from July 3rd to 12th of 2016. During the process of snowmelt infiltration, the soil wetting front moves slowly downward, as shown in Fig. 12(A). It takes 2 days and 7 hours for the wetting point to reach the 60 cm depth soil layer, but for the summer precipitation-induced infiltration, it takes only 1 day for the wetting front to reach the 60 cm depth soil layer.
The soil moisture recharge due to February 26th to March 27th snowmelt in 2016 lasts for 29 days, and the soil moisture recharge depth can reach 160 cm depth. The soil moisture at 200 cm depth does not show any noticeable change, suggesting that DSR has not been generated yet. The start time of the moisture recharge is set at the moment when the soil moisture content starts to increase. The end time of the moisture recharge is set at the moment when the soil moisture content reaches its maximum. These two moisture infiltration processes are shown in Fig. 12. There are many factors affecting the rate of precipitation-induced infiltration. A model that does not adequately consider the most relevant factors can certainly leads to erroneous simulation results. In the future, the knowledge gap between the experimental measurement results and the corresponding model simulation should be filled.
3.4 Recharge intensity of different soil layer infiltration
In a controlled laboratory experiment, one may calculate the precipitation infiltrating into a specific soil layer according to the soil characteristics with a proper mathematical model. In the natural environment, however, there are too many factors affecting the infiltration process, such as temperature and air humidity, wind speed, surface soil moisture, soil heterogeneity, etc. In this study, we will compute the replenishment of each soil layer by analyzing the precipitation-induced wetting point to find the minimum precipitation for infiltration to reach each individual layer, according to the results of our field test.
During the three-year in-situ observation period, there were 394 observable precipitation events in total. According to the soil moisture fluctuation data recorded by the soil moisture probe, there are 294 precipitation events that have infiltrated into the soil layer below 200 cm depth.
Table 2
Precipitation produced moisture increase signal and corresponding minimum precipitation (data from 2016–2018).
Soil layer depths | Sum of soil moisture increase signals on each soil layer | Corresponding minimum precipitation intensity |
20 cm | 74 | 2.6 mm/d |
40 cm | 46 | 3.2 mm/d |
80 cm | 32 | 3.4 mm/d |
120 cm | 16 | 8.2 mm/d |
160 cm | 16 | 10.2 mm/d |
200 cm | 10 | 13.2 mm/d |
|
Infiltration results in elevated soil moisture content. Each time when infiltration reaches a designated depth, it leaves a crest signal of soil water content. Based on the comparison between the time of crest signals and the time of precipitations, the minimum precipitation amount can be determined by the crest signals at different soil depths. Antecedent soil moisture conditions also affect infiltration depth, shallow soil moisture in the semi-arid sandy land will soon return to a relatively dry state because of evapotranspiration. This research describes the general state of soil moisture fluctuations in the experimental area after precipitation. Statistics of precipitation data from 2016 to 2018 and fluctuations in soil moisture content in each layer are shown in Table 2, which demonstrates that for infiltration to reach the soil layers at depths of 20 cm, 40 cm, 80 cm, 120 cm, 160 cm, and 200 cm, the required daily minimum precipitation intensities are 2.6 mm, 3.2 mm, 3.4 mm, 8.2 mm, 8.2 mm, and 13.2 mm, respectively. Infiltration depth and precipitation are not linearly related. This suggests that in the Mu Us sandy land, infiltration will cease to exist when the daily precipitation is less than 2.6 mm, and DSR may be detected only when the daily precipitation becomes greater than 13.2 mm. However, according to the three-year in-situ experimental record, one can see that for some precipitation events whose daily intensities are greater than 13.2 mm, there are no corresponding DSR events detected. Such evidence shows that the precipitation is not the only factor controlling DSR. Besides the precipitation intensity, other factors may also be relevant. Further research is needed to understand those factors impacting the soil moisture dynamics and DSR. One should note that the above infiltration depth is not based on mathematical calculations, but on the basis of the soil moisture fluctuation data monitored by the soil moisture probe at the experimental site.
3.5 Redistribution of precipitation in the PSM forest
Figure 5 shows that during the seasonal frozen-soil period, soil moisture is relatively stable. The monthly average values for soil moistures of different soil layers in January and December of 2016 are used as the start and end soil moisture values. Although the precipitation amount varies from 2016 to 2018, other environmental factors in this area are basically the same, and soil moistures are similar. To figure out the PSM water balance from 2016 to 2018, one has:
P-DSR-ET = δW (2)
where P is precipitation, ET is evapotranspiration, and δW is the whole 200 cm soil layer moisture change. Runoff is not included in above water balance equation because it does not occur during the experiment. As precipitation, DSR, and δW can be accurately measured, ET can be calculated by above equation.
Table 3 shows the precipitation, DSR, δW, and the computed ET for 2016–2018. Based on Table 2, one can see that precipitation has played a role in regulating and replenishing soil moisture for both shallow and deep soil layers. For the shallow soil layer, evapotranspiration in the dry year (like 2018) consumes stored water from previous wet year (2016), but in the wet year (like 2016), precipitation recharges the shallow soil layer. The recorded DSR values in 2016–2018 are very small, as compared to other terms in above Eq. (2), indicating that under the existing vegetation cover and rainfed conditions, the precipitation is barely able to support the shallow groundwater ecosystems, and has almost no capacity to provide recharge for groundwater reservoir in the region. However, in the bare sandy plot, the precipitation indeed can provide moderate recharge for groundwater reservoir, as reflected in the sizable annual DSR values there. In semiarid regions such as the Mu Us sandy land, precipitation varies considerably every year, and the year of 2016 may not be representative of the long-term average behavior of DSR in this region as the precipitation of this year is higher than the average annual precipitation of 400 mm. One can see that in different years, vegetation water consumption patterns are different, and evapotranspiration decreases in dry year and increases in wet year. One common feature among wet and dry years is that precipitation-induced infiltrated water is trapped in shallow soil layers and then consumed by PSM. To understand the long-term behavior of DSR and soil moisture dynamics in the semiarid regions such as the Mu Us Sandy land, one must carry out a multi-year (preferably a decade long) experiment.
As shown in Table 3, comparing the PSM plot and the bare soil land (BSL) plot, it can be found that the ET is 466.94 mm and the DSR is 1.4 mm for the PSM plot in 2016; while the ET is 137.13 mm and the DSR is 237.6 mm for the BSL plotin 2016. From this observation one can conclude that due to vegetation reconstruction, the DSR is significantly reduced and the ET is significantly increased in this year. 2018 was a relatively dry year, with precipitation of 239.8 mm in 2018. Due to the decrease in precipitation, the DSR has decreased in both plots. The DSR in the PSM plot is 1.2 mm and the BSL plot DSR is 55.2 mm, which is significantly lower than the ET and DSR in 2016.
Table 3
Water distribution of the rainfed PSM and bare sandy plot
Year | Plot | Precipitation | DSR | δW | ET |
2016 | PSM | 506.4 mm | 1.4 mm | 38.06 mm | 466.94 mm |
BSL | 506.4 mm | 273.6 mm | 95.67 mm | 137.13 mm |
2017 | PSM | 309 mm | 0.4 mm | -16.00 mm | 324.60 mm |
BSL | 309 mm | 67.7 mm | 61.89 mm | 179.41 mm |
2018 | PSM | 239.8 mm | 1.2 mm | 54.75 mm | 183.85 mm |
BSL | 239.8 mm | 55.2 mm | 96.8 mm | 87.8 mm |