The Applicability Of Different Parameterization Schemes With Noah-MP Land Surface Model In Arid And Semi-arid Regions Of Northwest China

Land surface provides the boundary conditions for the atmosphere and is an important way connecting the atmosphere and other spheres in the climate system. The simulation quality of land surface process is directly related to the accuracy of climate prediction and numerical weather forecasting. In recent years, land surface observation experiments are carried out in different land surfaces and a lot of observational data were gathered, which greatly promoted the development of land surface process researches and land surface models.Many researches on the land surface characteristics have achieved in the arid and semi-arid region of Northwest China, but there is less research works carried out on the impact of different parameterization schemes. In order to evaluate the performance of different parameterization schemes in arid and semiarid regions, the Noah-MP model with a variety of parameterization schemes options is used, which has a good performance at a single point and global scales but is rarely applied in the arid and semiarid regions of Northwest China. Besides, the observation datasets at the Semi-Arid Climate and Environment Observatory of Lanzhou University(SACOL) and Gobi station of HiWATER in arid regions are selected to validate the model results. The comparative results are as followings:Firstly, the land surface processes at the two sites are simulated using the Noah-MP model with three specified parameterization schemes, i.e. M-O parameterization scheme for the surface sensible heat exchange coefficient, Ball-Berry stomatal resistance parameterization scheme and SZ09 evaporation parameterization scheme from bare soil, and validated against the observations. The simulation results show that the model has a good performance on the simulations of soil moisture and soil temperature in the arid and semi-arid regions. Also the simulated net radiation fluxes are very close to the observed values respectively, but the biases in the simulated latent heat flux and sensible heat flux are significantly large in the arid and semi-arid regions.Secondly, considering the large biases of the simulated surface heat fluxes with Noah-MP model, a series of sensitivity tests with different parameterization schemes of the surface exchange coefficient, the stomatal resistance and the soil resistance are carried out to search for the main reasons. The results show that using the Chen97 parameterization scheme for the surface sensible heat exchange coefficient will improve the sensible heat flux simulations in the arid and semi-arid regions; adopting the Jarvis stomatal resistance parameterization scheme can increase the transpiration of plant with the improvement of latent heat flux and a better distribution ratio of sensible and latent heat fluxes at SACOL. LP92 evaporation parameterization scheme from bare soil is applied into the Noah-MP land surface model with a significantly increase of the evaporation when compared with the original scheme and a better simulation of the latent heat flux as well at SACOL while the latent heat flux at Gobi site has not improved, even increasing the sensible heat flux peak.Finally, the tests with different combinations of the three kinds of parameterization schemes are carried out in order to further improve simulation performance. The results with combination of schemes are much better than the default schemes test in the semi-arid region. Compared to the results of replacing one scheme with the above selected one, replacing 2 or 3 original parameterization schemes at the same time with the selected schemes further reduces the rmse(root mean square error) of the simulated surface sensible and latent heat fluxes, but the simulated soil moisture and soil temperature do not improve accordingly. The simulated latent heat flux with combination of schemes has changed little compared to the default schemes test. The simulation result of the sensible heat flux is improved significantly, but its simulation error and predictability are lower than those with Chen97 scheme in the arid region.