Scale-dependence of temporal stability of surface-soil moisture in a desert area in northwestern China

- Used re-sampling technique to analyse the scale-dependence of temporal stability.
- The sampling extent influencing temporal stability was greater than the spacing.
- Temporal stability was susceptible to scale changes when extent was small.
- The number of representative locations (RLs) was influenced by sample size.
- Scale change cannot weaken/intensify the ability of RLs to estimate the field mean.

SummarySoil moisture is an important component in hydrological cycles but is highly variable in space and time. As a tool for optimal soil and water management and effective field monitoring, the analysis of temporal stability has recently received increasing interest. The scale dependence of temporal stability, however, has received little attention. We measured surface-soil (0-6 cm) moisture at 187 sampling locations in a desert region (40 km2) in northwestern China approximately every two weeks from April to October 2012 with a Theta Probe, for a total of 13 sampling campaigns. We grouped the samples by re-sampling analysis under six sampling spacings and six different sizes of sampling area (extents) to analyse the changes in the characteristics of temporal stability at the various sampling scales and to test the accuracy of the field-mean moisture content estimated by measurements at a limited number of locations at each sampling scale. Increasing the spacing had little influence on the temporal stability of both the overall spatial pattern and the individual locations, whereas increasing the extent tended to increase the temporal stability of the overall spatial pattern but to decrease the temporal stability at the individual locations. The characteristics of temporal stability were susceptible to changes in scale when extent was small. Exponential and power equations could well express the relationships between most of the parameters of temporal stability and sampling scale. The number of identified representative locations (RLs) was influenced by sample size and was more sensitive to changes in sample size caused by extent than by spacing. At all sampling scales, the sets of selected RLs accurately estimated the field mean for the entire study period by using offset estimates, and the estimate accuracy was not affected by sampling scale. Increasing the spacing did not change the influence of topography and soil properties on temporal stability, whereas increasing the extent tended to intensify the influence of relative elevation and soil organic carbon content but to weaken the influence of saturated hydraulic conductivity on temporal stability. This study contributes to our understanding of the dependence of temporal stability of soil moisture in desert ecosystems on sampling scale and will help the integration of information from different spatial scales and the design of optimal sampling sizes and strategies for similar studies.