Analysis and improvement of runoff generation in the land surface scheme CLASS and comparison with field measurements from China

SummaryThe Canadian Land Surface Scheme (CLASS) is evaluated comprehensively using both site-specific data and observed hydrograph from the monsoon region of China. Two versions of CLASS, with and without hydrological modification allowing for interflow and baseflow, are used in this study. We modify standard CLASS for interflow generation using a field capacity threshold together with a spatial probability distribution function to represent sub-grid variability in soil field capacity. We also introduce a linear reservoir at the bottom of the third CLASS soil layer to model the baseflow. The 1998–99 HUBEX/GAME (HUaihe river Basin EXperiment/Monsoon Asian GEWEX Experiment) IOP (Intensive Observation Period) data set is used in this study. CLASS is driven by observed atmospheric forcing in a stand-alone mode. Both standard and modified versions of CLASS are first evaluated using the observed hydrograph at the outlet of the Shiguanhe sub-basin, part of the Huaihe River Basin in China, and soil moisture measurements at three sites over the sub-basin. The two versions are further validated for their ability to simulate net radiation, sensible and latent heat fluxes, ground heat flux, and soil temperature and moisture over four measurement sites with different land covers in the Huaihe River Basin. The total runoff in the Shiguanhe sub-basin simulated by the two versions is similar to each other, and is in close agreement with the observed value. The main difference between the two versions is in the partitioning of runoff components among the surface runoff, interflow and baseflow. Modified CLASS simulates better both peak flows and flood timing. The two versions of CLASS give similar results for the net radiation, sensible and latent heat fluxes, ground heat flux, and soil temperature and moisture, which compare well with observations over the study sites. Our modification of standard CLASS thus leads to an improvement in runoff and hydrograph simulation, while preserving its existing positive features.