Validation of soil moisture simulations from the PAMII model, and an assessment of their sensitivity to uncertainties in soil hydraulic parameters

Soil moisture simulations by the Canadian prairie agrometeorological model (PAMII) were validated against in situ observations from three sites on the Canadian prairies. Skill scores indicate that PAMII shows significant skill in simulating the evolution of bulk root-zone soil moisture content during the growing season, and that PAMII also captures the salient features of the soil moisture at each site. Specifically, correlation coefficients between simulated and observed bulk root-zone soil moisture content varied between 0.65 and 0.90, while the relative mean absolute errors were typically less than 10%.We adopted an ensemble approach to quantify the uncertainty in simulating soil moisture that results from errors in the soil hydraulic properties. The ensemble was assembled by running PAMII using soil hydraulic properties from 20 pedotransfer functions (PTFs). Our results suggests that the uncertainty in soil moisture estimates is typically less than 10% of the bulk root-zone soil moisture content, and that there appears to be merit in using a PTF ensemble to improve estimates of the soil's hydraulic properties. In addition, the simulated soil moisture was quite sensitive to the reference stomatal resistance term. For the grassland sites, a reference value of 100 s m−1 yielded the best results. The accuracy of the soil moisture simulations was insensitive to errors <5% in the initial plant-available soil moisture content.The skill of PAMII is strongly modulated by errors in the estimated permanent wilting point and field capacity, with the skill of model simulations typically increasing as the errors decrease. Consequently, the correct modelling of soil moisture is contingent on accurately specifying the site-specific soil hydraulic properties.