Evaluation of simulated soil moisture and temperature for a Canadian boreal forest

Abstract

• CLASS well simulated soil temperature. • Simulated soil water was underestimated during winter, while its seasonal variation corresponded well with observations. • Thickness of organic layer and soil water freezing point influenced both soil temperature and soil water. • Percentage sand and clay, and drainage parameter had effect on soil water, but not on soil temperature. • Simulated soil temperature and soil water was improved after calibration, especially during winter. Soil temperature ( T soil ) and soil water ( θ soil ) are fundamental variables that have an essential role in many processes in forest ecosystems, as well as influencing the tree species distribution and forest composition over time. We tested the Canadian Land Surface Scheme (CLASS) capacity to simulate T soil and θ soil in the boreal forest using a sixteen-year data set of daily measurements. Sensitivity analyses were also carried out to evaluate the impact of the thickness of organic layer (TOL), soil texture (percentage sand and clay: PS and PC), drainage parameter (DP), and water freezing point (FP) on simulated T soil and θ soil . Finally, the model was also calibrated with a combination of model parameters. CLASS well simulated T soil while its performance for θ soil varied by soil horizon and season. In winter particularly, soil liquid water was greatly underestimated because simulated T soil was below 0 °C. Nevertheless, simulated θ soil seasonal variation corresponded well with observations. Based on sensitivity analyses, TOL had an important effect on both T soil and θ soil . Although PS, PC and DP had almost no effect on T soil , their effects on θ soil were substantial. T soil increased throughout the year and θ soi l increased during the winter with decreasing FP, yet the match between modelled θ soil and observations was not substantially improved. In general, T soil was well simulated by CLASS, except for the freezing during winter. Model calibration improves greatly both simulated T soil and θ soil , especially during winter in all soil layers. However, despite the model calibration, CLASS still requires improvement for modelling T soil and θ soil , hence emphasizing the need to review the equations governing these variables in CLASS.