Keywords: 
• soil moisture dynamic /
• stochastic modeling /
• probability density function /
• Horqin Sand Land

Abstract: The soil moisture dynamics in the root zone of plants is studied through model simulation and observations. The simulation is done with the stochastic model for soil moisture dynamics developed by Laio et al. (Laio model). The observation covers a range of elements including growing season soil moisture and daily precipitation in fixed dunes and sandy grasslands of Horqin Sand Land for the period 2006-2010. The relationship between the soil moisture dynamics and the precipitation pattern is also analyzed. The probabilistic density function (PDF, or p(s)) of soil moisture is estimated using the Laio model. A sensitivity analysis is performed for the 13 parameters of the model. Results show that: ① the seasonal distribution of precipitation appears to be very uneven in the study area. The precipitation amount between April and October can account for 93% of the annual total precipitation. Precipitation events of 5 mm or less contribute only 25% of the annual total precipitation, but can comprise 73% of the total events. 38% of dry periods (intervals between precipitation events) are 10 days or less, which can account for 87% of the total dry periods during a year; ② on fixed dunes and sandy grasslands, the root zone depth is between 0-10 cm and 0-70 cm, respectively. The value of root-zone soil moisture in sandy grasslands is significantly higher than that in fixed dunes. In both cases, the highest soil moisture occurs in July relative to other growing season months; ③ the values of root-zone soil moisture during growing seasons follow a normal distribution on both fixed dune and sandy grassland. The peak value, the position and the scope of p(s) obtained by the Laio model are in good agreement with observations, indicating the Laio model is able to estimate the p(s) of soil moisture in Horqin Sand Land; ④ the estimation of the peak value of p(s) is more sensitive to the 5 parameters of the Laio model, which are the arrival rate of rainfall events (λ), the mean rainfall depth (α), the average daily maximum evapotranspiration rate (E_max), the soil moisture level at which plants begin closing stomata (s^*) and the permanent wilting point (s_w).