Response of deep soil drought to precipitation, land use and topography across a semiarid watershed.

• Watershed scale soil water to a depth of 500 cm was investigated from 2013 to 2016. • Half of the 500-cm profiles were drying out below 125 cm, and dried soil layers responded strongly to rainfall in wet years. • DSLT and DSLFD were controlled by climatic variables in time, and DSL-SWC was controlled by land use and topography in space. • Plan curvature, slope gradient, clay and silt content were the dominant factors of DSLs both in space and time. • Grassland may be the optimum land use type to minimize DSL occurrences. Soil drought caused by climatic change and/or poor land management in arid and semi-arid regions are seldom recognised due to a lack of comparative data on soil moisture (SM) in soil profiles. This lack of information endangers the sustainability of these fragile ecosystems. The current study assessed spatial-temporal variations of soil drought, as indicated by dried soil layers (DSL), at a watershed scale, and tested the hypothesis that soil drought in deep profiles is controlled by the combined effects of meteorological processes, land use, and topography. We measured SM to a depth of 500 cm on 20 occasions at 73 locations from 2013 to 2016 at a small watershed on the Chinese Loess Plateau (CLP). We also collected data on possible environmental factors including meteorological variables, land use, topographical elements, and soil properties. The DSLs occurred at > 90% of the sampling sites within the watershed, and the spatially and temporally averaged DSL formation depth (DSLFD), DSL thickness (DSLT) and soil water content within the DSL (DSL-SWC) were 125 cm, 257 cm, and 10.4%, respectively. This suggests that 51.4% of the 500-cm-profile is drying out below 125 cm. The DSLFD, DSLT and DSL-SWC demonstrated a moderate degree of variability (20% < CV < 84%) in space, and showed a moderate, moderate and weak temporal variability, in time, respectively. The temporal series of the mean spatial DSLT and DSLFD were significantly correlated with climatic variables. The spatial variation of the mean temporal DSL-SWC differed significantly among the land uses and between shaded and sunlit aspects. We found that plan curvature, slope gradient, clay and silt content regulated DSLs in both space and time. This result verified our hypothesis that meteorological processes, land use, and topography play an essential role in shaping DSL variation and distribution pattern. Taking DSL reclamation into account in the study area, grassland would be the optimum land use type. Understanding this information is helpful for watershed soil and water conservation, and soil drought meditation via the best management practices in the CLP and other water-limited regions with deep soils. [ABSTRACT FROM AUTHOR]