Elsevier

Geoderma

Volume 369, 15 June 2020, 114320
Geoderma

Using pedo-transfer functions to estimate dry soil layers along an 860-km long transect on China’s Loess Plateau

https://doi.org/10.1016/j.geoderma.2020.114320Get rights and content

Highlights

  • Dry soil layer thickness is harder to predict than the other indices.

  • The artificial neural network approach effectively improved simulation accuracy.

  • Indirect method is a promising development for dry soil layer indices prediction.

Abstract

Dry soil layer (DSL) development as a result of imbalance in water input and output is a widespread pedo-hydrological phenomenon in arid/semi-arid regions such as the China’s Loess Plateau (CLP). To build sufficient data for large-scale DSL estimation, soil water data for the 0–5 m soil profile were collected for the period 2013–2016 along an 860-km long transect on CLP and analyzed for pedo-transfer functions (PTFs). The objective was to determine the effects of environmental factors on DSL variation and to develop an effective PTF for the estimation of DSLs on CLP. Three DSL evaluation indices were calculated — DSL thickness (DSL-T), DSL formation depth (DSL-F) and DSL mean soil desiccation index (DSL-SDI). The results showed that DSLs were mainly distributed in the northcentral part of the transect, with a mean thickness of 2.77 m. We compared the performances of PTFs developed by different approaches — multiple regression (MR), artificial neural network (ANN) and the indirect and direct methods. It showed that the ANN approach effectively predicted DSL formation and indices. The indirect method improved simulation accuracy of DSL indices. The combination of the ANN approach and the indirect method gave the best estimation accuracy for DSL indices. The application of the PTFs not only reduced labor and time needed for field survey of DSL, but also improved DSL research on CLP and beyond. The indirect method based on soil moisture and/or hydrological models was promising for the estimation of DSL indices.

Introduction

Climate change and human activities are dramatically affecting traditional hydrological processes in terrestrial ecosystems (Zavaleta et al., 2003, Wang et al., 2016), resulting in more frequent and intense droughts (Sherwood and Fu, 2014, Anderegg et al., 2015). This has interrupted the balance in water input (e.g., rainfall and irrigation) and output (e.g., evaporation, root uptake and deep drainage) (Brown, 2002, Wang et al., 2010a, Wang et al., 2010b, Schulte et al., 2011), ultimately causing the formation of dry soil layers (DSLs) or soil desiccation in the soil profile (Li, 1983, Yang et al., 1999, Jia et al., 2015).

DSL is a soil layer with soil water content below stable field capacity (Chen et al., 2008, Wang et al., 2010a, Jia et al., 2020). It is mainly caused by a long period of excessive depletion of deep soil water through evapotranspiration and low precipitation recharge (Wang et al., 2010b, Yan et al., 2015). DSL is a global soil hydrological phenomenon that has been reported in southern Australia (Robinson et al., 2006), eastern Amazonia (Jipp et al., 1998), Russia (Yang and Han, 1985), southwestern United States (Querejeta et al., 2007) and China’s Loess Plateau (CLP) (Wang et al., 2010a). For example, Robinson et al. (2006) found that soil water depletion by Eucalyptus robusta caused the formation of low water potential layer to the depth of 2 m in the profile soil, limiting groundwater recharge by incident rainfall in southern Australia. In southwestern United States, Querejeta et al. (2007) observed DSL throughout the 0–2 m soil profile and the Quercus palustris plant derived water from moist soil layers below the 2 m depth. The formation of DSLs can negatively affect water cycle, crop yield and ecosystem succession (Nepstad et al., 2004, Chen et al., 2008, Mendham et al., 2011). Furthermore, if DSL persists for a long time, it could lead to soil degradation, regional vegetation die-off and aridity of local climatic environment (Yan et al., 2015, Jia et al., 2020). Understanding the formation and distribution of DSL is thus critical for sustainable management of soil hydrological and ecological processes at various spatial scales.

CLP is well known for its intense soil erosion, severe water shortage, fragile ecosystem and deep loess deposit. Thus, considerable effort in terms of labor, time and money has been invested into vegetation restoration in the region (Lü et al., 2012, Zhang et al., 2017, Jia et al., 2019). In CLP region, DSL was first detected in the 1960 s (Li, 1962). Since then, numerous studies have been conducted relating to the definition and classification (Li, 1983, Chen et al., 2008), formation and development (Wang et al., 2010b), spatial distribution pattern (Wang et al., 2010a, Zhang et al., 2017), temporal persistence (Jia et al., 2015) and sustainable recovery (Wang et al., 2012b) of DSLs in the region. Wang et al. (2010a) noted that DSL was found in most regions of CLP, but was generally more severe in the central and western regions of the plateau. A widespread formation of DSL with severe negative effects can endanger the sustainability and development of restored ecosystems and ruin the vast investments in environmental conservation in the study area (Wang et al., 2010a, Jia et al., 2015, Yan et al., 2015). Thus, studying the status and spatial distribution of DSLs at regional scale is critical for developing sustainable land management strategies with long-term ecosystem services in CLP and beyond.

According to previous studies, three indices are usually employed to characterize DSLs — DSL thickness (DSL-T), DSL initial formation depth (DSL-F) and DSL mean soil water content (DSL-SWC) (Robinson et al., 2006, Zhao et al., 2007, Wang et al., 2010a, Yan et al., 2015). The determination of DSL indices is driven by profile soil moisture survey. Such survey is often labor intensive, time consuming and costly, especially for larger-scale and deep-profile applications (Saxton and Rawls, 2006, Wang et al., 2010a, Zhao et al., 2016a). Difficulties in soil profile survey affect DSL research efficiency. As an alternative therefore, pedo-transfer function (PTF), which uses predictive functions and soil survey data to determine soil properties, is now widely considered in soil study (Minasny and McBratney, 2000, Chapuis, 2012, Duan et al., 2012, Yao et al., 2015, Zhao et al., 2016a). PTF is built on the premises that the main driving factors can be determined. Wang et al. (2010a) noted that the primary and secondary factors with significant impact on DSL formation and/or development were precipitation and soil type. Yan et al. (2015) observed that mean annual precipitation is negatively correlated with DSL-T, but positively correlated with DSL-F and DSL-SWC. Thus, based on potential relationships between environmental factors and DSL indices, PTFs of DSL indices can be built from environmental factors.

Although studies have simulated DSL formation and development (Wang et al., 2012b, Li et al., 2007), remains little known about PTFs of DSL indices at regional scale. The uniqueness of regional geographical and environmental variability on CLP makes DSL research difficult, especially in terms of cost and time (Jia et al., 2015). The determination of PTFs from dominant factors of DSL indices is therefore a theoretical and practical requirement for DSL research. On CLP, a policy-driven large-scale vegetation restoration has been going on since 1999, resulting in 10.8% reduction in cropland, 4.9% increase in forestland and 6.6% increase in grassland (Lü et al., 2012). There is widespread DSL in the restored ecosystems with significant negative effect on the local ecology and environment. This lays the need for developing robust PTFs to effectively and easily predict DSL indices that can enhance DSL evaluation and soil water management on CLP.

The specific objectives of this study were to: 1) characterize regional-scale variations in DSLs along an 860-km south-north transect of CLP; 2) determine the relationship between environmental factors and DSL in the study area; and 3) develop effective PTFs for the estimation of DSLs in the region.

Section snippets

Study area

The study was conducted in CLP region (34°0́–45°5́N, 101°33́–114°33́E), with a total area of 64 × 104 km2 (Fig. 1). The geomorphic landforms include the Yuan (a large flat surface with little or no erosion) and Mao (an oval or round loess hill), Liang (a long narrow range of hills), and hills/gullies of various forms (Wang et al., 2018). The climate in the study area is continental monsoon, with mean annual precipitation ranges from the northwest to the southeast of 150–800 mm, 55–78% of which

SDI and DSL distribution along transect

The profile distribution of the four-year average SDI is shown in Fig. 3. In the transect, DSL (with SDI < 1) was mainly distributed in the northcentral part (regions between sites 46–75) of CLP study area and was generally thicker and more severe in the area. For some sites, there was DSL right through the 0–5 m soil profile — e.g. sites 46, 47, 49, 50, 54, 56, 60, 62 and 65 (Fig. 3). Compared with the northcentral region of the plateau, DSL was scattered across the southcentral region and had

Spatial patterns of DSL

In CLP study area, DSL most widely occurred in the northcentral region (Fig. 3). Wang et al. (2010a) also noted that DSLs were mainly in western and northwestern regions of the plateau, although they were relatively thinner in interior regions such as Guanzhong Plain. Zhao et al. (2019) concluded that soil desiccation in the northern region was more severe than in the southern region of the Loess Plateau.

The distribution of DSLs can be explained in terms of the spatial distribution of

Conclusions

The study showed that DSL existed mainly in the northcentral half of the transect and had an average thickness of 2.77 m on CLP. The ANN approach improved the simulation accuracy of the model and its performance was better than the MR approach in terms of recognition of DSL and simulation of DSL indices in the study area. The indirect method was effective for the prediction of DSL indices. The use of these PTFs not only reduced the labor and time needed for field investigations, but also

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This study was supported by the National Natural Science Foundation of China (41877016 and 41907009), the China Postdoctoral Science Foundation (2018M641460) and the Youth Innovation Promotion Association of the Chinese Academy of Sciences (2017076). We thank the anonymous editors and reviewers for the value added to the work.

References (52)

  • J.I. Querejeta et al.

    Hydraulic lift may buffer rhizosphere hyphae against the negative effects of severe soil drying in a California oak savanna

    Soil Biol. Biochem.

    (2007)
  • P. Schulte et al.

    Applications of stable water and carbon isotopes in watershed research: weathering, carbon cycling, and water balances

    Earth Sci. Rev.

    (2011)
  • X. Tan et al.

    Simulating soil water regime in lowland paddy fields under different water managements using HYDRUS-1D

    Agr. Water Manage.

    (2014)
  • X. Wang et al.

    Validation of the EPIC model using a long-term experimental data on the semi-arid loess plateau of china

    Math. Comput. Model

    (2011)
  • Y. Wang et al.

    Large-scale spatial variability of dried soil layers and related factors across the entire Loess Plateau of China

    Geoderma

    (2010)
  • Y. Wang et al.

    A preliminary investigation of the dynamic characteristics of dried soil layers on the Loess Plateau of China

    J. Hydrol.

    (2010)
  • Y. Wang et al.

    A new index to quantify dried soil layers in water-limited ecosystems: a case study on the Chinese loess plateau

    Geoderma

    (2018)
  • C. Zhao et al.

    Long-term temporal variations of soil water content under different vegetation types in the Loess Plateau, China

    Catena

    (2017)
  • C. Zhao et al.

    Using pedotransfer functions to estimate soil hydraulic conductivity in the Loess Plateau of China

    Catena

    (2016)
  • C. Zhao et al.

    Particle size distribution of soils (0–500 cm) in the Loess Plateau, China

    Geoderma Regional

    (2016)
  • W. Anderegg et al.

    Tree mortality predicted from drought-induced vascular damage

    Nat. Geosci.

    (2015)
  • K. Brown

    Environmental data - water scarcity: forecasting the future with spotty data

    Science

    (2002)
  • K. Burnham et al.

    Multimodel inference: understanding AIC and BIC in model selection

    Sociol. Method. Res.

    (2004)
  • R. Chapuis

    Predicting the saturated hydraulic conductivity of soils: a review

    B. Eng. Geol. Environ.

    (2012)
  • R. Duan et al.

    Comparison of methods to estimate saturated hydraulic conductivity in Texas soils with grass

    J. Irrig. Drain. E-Asce.

    (2012)
  • FAO/ISRIC/ISSS. 1998. World reference base for soil resources. World Soil Rec. Rep 84. FAO,...
  • Cited by (15)

    • Prediction of deep soil water content (0–5 m) with in-situ and remote sensing data

      2023, Catena
      Citation Excerpt :

      Five vegetation zones, forest, forest steppe, typical-steppe, desert-steppe and steppe-desert, are distributed from southeast to northwest (Feng et al., 2017). More complete details on the study area have been published in past studies (Fu et al., 2011; Zhao et al., 2020). A total of 86 aluminum neutron-probe access tubes (5.0 m in length) were installed at 10 km intervals from south to north across the CLP transect (Fig. 1).

    • Building pedotransfer functions for estimating soil erodibility in southeastern China

      2022, Ecological Indicators
      Citation Excerpt :

      MLR and machine-learning technologies are commonly used for building models. MLR has the advantages of simple calculation, strong explanatory power and convenient application and has been widely used to build PTF models for estimating soil properties (Qiao et al., 2019a, 2019b; Ostovari et al., 2019; Zhang et al., 2019b; Barbosa et al., 2019; Rivera and Bonilla, 2020; Zhao et al., 2020). The machine-learning technologies including adaptive neuro-fuzzy inference systems, artificial neural networks (ANNs), support vector machines and random forest techniques (Szabo et al., 2019; Rivera and Bonilla, 2020).

    • The coupling interaction of soil organic carbon stock and water storage after vegetation restoration on the Loess Plateau, China

      2022, Journal of Environmental Management
      Citation Excerpt :

      However, Feng et al. (2016) noted that vegetation restoration has approached the sustainable water resource limit. Zhao et al. (2020) noted that the drastic increase in vegetation has led to severe soil water depletion and widespread dry soil layers. It is worth noting that the decreasing SWS has become the main obstacle hampering vegetation restoration, thereby threatening SOC sequestration (Zhang et al., 2016, 2020b; Feng et al., 2017; Wu et al., 2018).

    • Prediction of the van Genuchten model soil hydraulic parameters for the 5-m soil profile in China's Loess Plateau

      2022, Catena
      Citation Excerpt :

      Field measurements are direct and highly precise, but are time-consuming, labour-intensive, and costly, especially when performed over long-term period and at large spatial scales. Numerical simulations can alternatively be used for identification of missing hydrological processes by using input data (Romano, 2014; Zhao et al., 2020a,b). In numerical simulations, the processes of water flow are expressed by mainly using the water budget method or fundamental equations such as the Richards’ equation.

    • Soil enzyme activities of typical plant communities after vegetation restoration on the Loess Plateau, China

      2022, Applied Soil Ecology
      Citation Excerpt :

      On the Loess Plateau, even though vegetation restoration has resulted in abundant nutrient inputs, potential water shortages have gradually emerged in recent decades (Feng et al., 2016; Feng et al., 2017). In particular, plants with exuberate root systems can contribute to the generation of dry soil layers, which substantially aggravate the root absorption capacity, causing severe vegetation degradation (Li et al., 2017; Xiao and Hu, 2017; Zhang et al., 2019; Zhao et al., 2020). Therefore, it is of great necessity to match suitable species with afforestation sites when implementing vegetation restoration (C. Zhang et al., 2020).

    View all citing articles on Scopus
    View full text