| Abstrakt: |
Purpose: Soil water is a critical variable for hydrological and biological processes in arid and semi-arid ecosystems. Information on regional spatial pattern of soil water storage (SWS) and its relationship with environmental factors is important for optimal water management and vegetation restoration in China's Loess Plateau (CLP) region. State-space approach and artificial neural network (ANN) were used to analyze spatial variability of SWS in the CLP region. Materials and methods: SWS in the 0-1, 1-2, 2-3, 3-4, and 4-5 m soil layers was measured during the period from June 2013 to September 2015 at 86 locations along a 860 km long south-north transect of CLP. Results and discussion: The analysis showed that SWS in the 5 m soil profile generally decreased with increasing latitude, driven by decreasing precipitation and soil-water holding capacity. Using various combinations of variables, the state-space model gave a better spatial pattern of SWS than the ANN approach. The best state-space approach, which included clay content, mean annual precipitation, and slope gradient, explained 96.0% of the total variation in SWS. Then, the best ANN approach explained only 76.2% of the variation. Clay content, mean annual precipitation, and slope gradient was the most effective combination for large-scale estimation of SWS under the state-space approach. Conclusions: The state-space model was recommended as an effective method for analyzing large-scale spatial patterns of soil water using soil, climatic, and topographic properties in the CLP region. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink