AbstractSoil‐water content (SWC) is a key factor in restoring degraded vegetation in alpine meadow ecosystems, but it has rarely been spatially simulated on a hectometer scale. We simulated SWC for typical dry and wet days in an alpine meadow using multivariate linear regression and autoregressive state‐space equations based on SWC and other soil, terrain, and vegetation parameters to evaluate the efficiency of these two methods in dry and wet soil‐moisture conditions. SWC measured on a typical dry day (SWC‐D) and a wet day (SWC‐W) increased and decreased with depth, respectively, and SWC‐D was similar to SWC‐W at a depth of 50 cm. Both SWC‐D and SWC‐W were significantly correlated with soil bulk density (BD), capillary porosity, silt content (Silt), gravel and stone content (GSC), pH, and organic carbon density (OCD), and both SWC‐D and SWC‐W were significantly auto‐correlated and cross‐correlated with BD, Silt, GSC, pH, and OCD at more than one lag distance. Multivariate linear regression using three variables in both dry and wet conditions had the highest accuracy, and the accuracy was generally higher for dry conditions than it was for wet conditions. The bivariate state‐space model was the most accurate for both dry and wet soil conditions, but the expression variables were totally different, with pH and OCD for dry day and BD and Silt for wet day. The conditions of soil moisture should thus be considered when choosing variables with which to simulate SWC, instead of only considering the relationships between SWC and other variables.