Temporal stability analysis of surface soil water content on two karst hillslopes in southwest China

Knowledge of the temporal variability of soil water content (SWC) at the hillslope scale is essential for guiding rehabilitation strategies and for optimizing water resource management in the karst region of southwest China. This study aimed to use temporal stability analysis to upscale point-scale measurements to represent mean areal SWC on two typical karst hillslopes. Based on a grid sampling scheme (10 m x 10 m) applied to two 90 m x 120 m plots located on two hillslops, the SWC at a depth of 0-16 cm was measured 11-12 times across 259 sampling points, using time domain reflectometry (TDR) from April 2011 to October 2012. Soil texture, bulk density (BD), saturated hydraulic conductivity (K (s) ), organic carbon (SOC), rock fragment content (RFC), and site elevation (SE) were also measured at these locations. Results showed the hillslope with more shrub cover was wetter than the hillslope with mixed grass-shrub cover. This difference was related to the differences in soil texture, soil hydraulic permeability, and topography. Through a comparison of values obtained with the Spearman correlation coefficient (r (s) ), standard deviation of mean relative difference (SDRD), and mean absolute bias error (MABE), we inferred that there is a higher degree of temporal stability for SWC in wet conditions than in drier conditions on the two hillslopes. Based on the values of the index of temporal stability (ITS), which combine the mean relative difference (MRD) and SDRD, the two locations were determined to be representative of mean SWC on both hillslopes. Moreover, these locations captured changes in mean SWC (NSCE = 0.69, and 0.65, and RMSE = 1.96, and 1.96 %, respectively). This demonstrates the feasibility of using the temporal stability of SWC to acquire mean SWC on karst hillslopes of southwestern China. The indirect method, which estimates mean SWC by considering the offset between the mean and the measurement value at a time-stable location, predicted mean SWC (NSCE = 0.86, and 0.76, and RMSE = 1.29, and 1.63 %, respectively) more precisely than the direct method (mean SWC directly measured at a time-stable location), because it eliminates deviation by introducing a constant offset (MRD). We recommended the use of the indirect method to acquire mean SWC values, when an allowable bias of 5 % for both MRD and SDRD can not be achieved. In addition, we found that soil texture, RFC, and elevation affect the pattern of SWC on the shrub hillslope. These results are expected to be useful for monitoring soil water dynamics on karst hillslopes, especially for restoration purposes.