Impact of soil parameter and physical process on reproducibility of hydrological processes by land surface model in semiarid grassland

The land surface model (LSM) takes part in the land-atmosphere interaction on the earth system model for the climate change research. In this study, we evaluated the impact of soil parameters and physical process on reproducibility of hydrological process by LSM Minimal Advanced Treatments of Surface Interaction and RunOff (MATSIRO; Takata et al, 2003, GPC) forced by the meteorological data observed at grassland in semiarid climate in China and Mongolia. The testing of MATSIRO was carried out offline mode over the semiarid grassland sites at Tongyu (44.42 deg. N, 122.87 deg. E, altitude: 184m) in China, Kherlen Bayan Ulaan (KBU; 47.21 deg. N, 108.74 deg. E, altitude: 1235m) and Arvaikheer (46.23 N, 102.82E, altitude: 1,813m) in Mongolia. Although all sites locate semiarid grassland, the climate condition is different among sites, which the annual air temperature and precipitation are 5.7 deg. C and 388mm (Tongyu), 1.2 deg.C and 180mm (KBU), and 0.4 deg. C and 245mm(Arvaikheer). We can evaluate the effect of climate condition on the model performance. Three kinds of experiments have been carried out, which was run with the default parameters (CTL), the observed parameters (OBS) for soil physics and hydrology, and vegetation, and refined MATSIRO with the effect of ice in thermal parameters and unfrozen water below the freezing with same parameters as OBS run (OBSr). The validation data has been provided by CEOP(http://www.ceop.net/) , RAISE(http://raise.suiri.tsukuba.ac.jp/), GAME-AAN (Miyazaki et al., 2004, JGR) for Tongyu, KBU, and Arvaikheer, respectively. The reproducibility of the net radiation, the soil temperature (Ts), and latent heat flux (LE) were well reproduced by OBS and OBSr run. The change of soil physical and hydraulic parameter affected the reproducibility of soil temperature (Ts) and soil moisture (SM) as well as energy flux component especially for the sensible heat flux (H) and soil heat flux (G). The reason for the great improvement on the reproducibility of H and G by the OBSr run is likely to include the ice effect on the soil heat capacity and the thermal conductivity. The reproducibility was lower in winter than summer on the simulation of SM. The low reproducibility of winter SM is not only the performance of LSM but also the observation problem.