Main content area

Estimating the spatial means and variability of root-zone soil moisture in gullies using measurements from nearby uplands

Gao, Xiaodong, Wu, Pute, Zhao, Xining, Zhang, Baoqing, Wang, Jiawen, Shi, Yinguang
Journal of hydrology 2013 v.476 pp. 28-41
clay, data collection, gully erosion, highlands, loess, plateaus, runoff, soaking, soil heterogeneity, soil water, vegetation, watersheds
Root-zone soil moisture is of critical importance to vegetation restoration, runoff and erosion processes in gullies, especially in arid and semiarid areas. Nevertheless, data relating to soil moisture in gullies are rare, since it is difficult to record in such environments. We hypothesize that mean soil moisture and its spatial variability within gullies could be estimated reliably on the basis of measurements from adjacent uplands, where data are much easier to collect. In this study, we proposed five different strategies to estimate root-zone (0–20, 20–40, 40–60 and 60–80cm) spatial mean soil moisture in gullies and three different strategies to estimate soil moisture spatial variability at the corresponding depths, using measurements collected from nearby uplands, based on 2-year soil moisture data collection campaigns in the Yuanzegou gully catchment on the Chinese Loess Plateau. The estimation errors were quantified for each strategy. The results showed that spatial mean soil moisture (θ¯g) at each depth in gullies was reliably estimated from observations collected at a single upland location when previous soil moisture data collected simultaneously for gullies and uplands were available. The upland locations that were within a ±5% relative bias error for the estimation of θ¯g had similar average percentage clay content (16.1%) in the surface layer (0–20cm) to that in the gullies (15.7%), and moderately steeper slopes (from 36.6% to 53.1%) compared to the upland average (39.4%). When previous soil moisture data were not available, the estimation was much less accurate. Furthermore, when different number of upland locations were randomly selected to estimate θ¯g, the estimation accuracy was only improved slightly when more than 10 locations were used. Based on the best estimated θ¯g and the quantitative relationship between spatial means and spatial variability of soil moisture in gullies, we were able to obtain the estimates of spatial soil moisture variability (standard deviation and coefficient of variation) with a certain level of accuracy (R2 from 0.300 to 0.625) in gullies at each depth. However, when the quantitative relationship was not available, there were much greater estimation errors. Our results demonstrated that spatial means and variability of soil moisture could be reliably estimated using measurements from nearby uplands provided previous soil moisture datasets for gullies and upland locations are simultaneously available.