Temporal stability of soil moisture in various semi-arid steppe ecosystems and its application in remote sensing

Abstract

Summary Monitoring soil moisture is often necessary in hydrological studies on various scales. One of the challenges is to determine the mean soil moisture of large areas with minimum labour and costs. The aim of this study is to test temporal persistence of sample locations to decrease the number of samples required to make reliable estimates of mean moisture content in the top soil. Soil moisture data on four experimental sites were collected during the vegetation period in 2004–2006. The experimental sites are located in a steppe environment in northern China, and are characterised by different grazing management which causes differences in vegetation cover. A total of 100 sampling points per site were ranked with respect to their difference to field mean soil moisture using the time-stability concept. We tested whether: (a) representative sample locations exist that predict field mean soil moisture to an acceptable degree, and (b) these locations are time-stable beyond a single vegetation period. Time-stable locations with a low deviation from mean field soil moisture and low standard deviation were identified for each site. Although the time-stability characteristics of some points varied between years, the selected points were appropriate to predict mean soil moisture of the sites for multiple years. On the field scale, time-stability and the persistence of patterns were analysed by the use of a Spearman rank correlation. The analysis showed that persistence depended on grazing management and the related plant cover. It is concluded that the time-stability concept provides useful information for the validation of hydrological or remote sensing models, or for the upscaling of soil moisture information to larger scales. A preliminary comparison of soil moisture measurements derived from ground-truth and remote sensing data showed that the data matched well in some cases, but that the considerable difference in spatial extent promotes differences in other cases.