Spatio-temporal variability behavior of land surface soil water content in shrub- and grass-land

The knowledge of the relationship between spatial variability of the surface soil water content (θ) and its mean across a spatial domain (θm) is crucial for hydrological modeling and understanding soil water dynamics at different scales. With the aim to compare the soil moisture dynamics and variability between the two land uses and to explore the relationship between the spatial variability of θ and θm, this study analyzed sets of surface θ measurements performed with an impedance soil moisture probe, collected 136 times during a period of one year in two transects covering different land uses, i.e., korshinsk peashrub transect (KPT) and bunge needlegrass transect (BNT), in a watershed of the Loess Plateau, China. Results showed that the temporal pattern of θ behaved similarly for the two land uses, with both relative wetter soils during wet period and relative drier soils during dry period recognized in BNT. Soil moisture tended to be temporally stable among different dates, and more stable patterns could be observed for dates with more similar soil water conditions. The magnitude of the spatial variation of θ in KPT was greater than that in BNT. For both land uses, the standard deviation (SD) of θ in general increased as θm increased, a behavior that could be well described with a natural logarithmic function. Convex relationship of CV and θm and the maximum CV for both land uses (43.5% in KPT and 41.0% in BNT) can, therefore, be ascertained. Geostatistical analysis showed that the range in KPT (9.1 m) was shorter than that in BNT (15.1 m). The nugget effects, the structured variability, hence the total variability increased as θm increased. For both land uses, the spatial dependency in general increased with increasing θm.

Research highlights
► Temporal pattern of soil moisture (θ) behaved similarly for the two land uses.
► More stable pattern of θ was found for dates with more similar soil water conditions.
► Soils for BNT were wetter during the wet period and drier during the dry period.
► Natural logarithmic function is good to describe SD versus θm for semi-arid areas.
► Nugget, structured variability, and spatial dependency increased with increasing θm.