Comparative Analysis And Numerical Simulation Test Of Land Surface Parameterazations Of Loess Plateau In Semi-arid Region

Land-atmosphere interactions are not only affected by the meteorological conditions, but also affected by the vegetation type, vegetation coverage and physicochemical characteristics of soil. In a general land surface process model, land characteristic parameters, such as albedo(α), aerodynamic roughness(z0m), κB-1 and soil heat conductivity(λs), etc., can influence the mass and energy flux exchange on land surface and in soil. Hence the accuracy of land surface process parameters is vital to simulation performance of land surface process model. In order to improve the simulation performance of land surface model in semi-arid Loess Plateau, this paper studied three aspects as following, with the basis of the observation data in SACOL (Semi-Arid Climate and Environment Observatory of Lanzhou University) during June to September from 2007 to 2009 and the land surface model Noah LSM. Firstly, the effects of land parameters’ disturbances on simulation performances are analyzed in detail on the basis of disturbance ratio method. Secondly, using the observed data, the characteristics of four parameters are analyzed, respectively, and comparisons of different parameterizations are done subsequently. Finally, combining different parameterizations of three parameters(α,λs and κB-1), and then analyzing the applicability of each parameterization combination in SACOL, an optimal combination for typical land surface in semi-arid Loess Plateau is proposed at last. Main conclusions are as follows:(1) Sensitivity test analyses of how Noah LSM responds to disturbance of land surface parameters indicate that, disturbance of α, λs, z0m and κB-1 have significant influences on land surface model. Disturbance of α has a prominent effect on the simulation performance of energy flux, surface temperature and soil temperature, an increase in a is associated with decrease of the simulation value of energy flux, surface temperature and soil temperature. Disturbance of λs has a remarkable effect on the simulation performance of soil temperature and soil heat flux, a decrease in λs is consistent with decrease of the simulation value of soil temperature and soil heat flux. Disturbance of z0m and κB-1 have an important effect on the simulation performance of sensible heat flux and surface flux, a decrease in z0m (or an increase in κB-1) results in decrease of the simulation value of sensible heat flux and increase of the simulation value of surface temperature.(2) Analysis of land surface parameters’characteristics showed, the monthly average a from June to September which derived from the data of 2007 to 2009 in SACOL site are,in order,0.187, 0.183,0.172 and 0.170; α are both in inverse proportion to solar elevation and soil moisture content. The monthly average λs, in 5cm depth are, in order,0.381,0.371,0.543 and 4.20; λs is in direct proportion to soil moisture content. Comparing three determination methods of z0m proposed by Chen(1993), Yang(2008) and Shang(2010), we found that there is little difference among these methods, z0m of SACOL is 0.012 m when taking the average of three methods. Comparing the parameter values derived from different parameterizations and the parameter values derived from observation data, we found that MULT scheme of a is the closest approach of the observed one; when compare the monthly average value of soil heat conductivity in 5 cm below calculated by observation data and three parameterizations, the parameterization called J75_NOAH is the best one compared with others; the kB-1 value estimated by Y03 is closest to the value derived from sensible heat flux.(3) Comparisons of different parameterization combinations’ influences to the simulation performance showed that, the simulation performances of different α parameterizations have little difference, W05 scheme performs best in relative term. The choice of λs parameterizations is vital to the simulation of soil heat flux, latent heat flux, surface temperature and soil temperature in deeper layer, and J75_NOAH scheme has the best simulation performance. Choosing kB-1 parameterizations is vital to the simulation of sensible heat flux, net radiation flux and surface temperature, and Y03 scheme has the best simulation performance. When we choose W05 for α, J75_NOAH for λs and Y03 for kB-1, the modified Noah LSM has the best simulation performance at SACOL site, especially for the sensible heat flux and surface temperature.