Soil Water Modelling In Arid/Semiarid Regions of Northern China Using Land Information System (LIS) - A Minor Field Study in Shiyang River Basin

At present China suffers from severe desertification or land degradation. About 27% of the total territory was exposed in 2004, mainly the northern provinces. Consequences of desertification, such as floods or sandstorms, require huge effort and financial assets. Soil moisture understanding plays a key role in combating desertification and is necessary in order to implement a sustainable water management. Land Surface Models (LSMs) are one approach to survey and quantify soil moisture. A LSM calculates the surface state from physical conceptual equations based on satellite derived input. Land Information System (LIS) is a framework for global modelling with LSMs. LIS has a high spatial and temporal resolution and the ability to simulate soil moisture and other water related parameters in a near real time manner. There are several kinds of LSMs but currently only two are implemented in the Land Information System (LIS); namely Noah and CLM. LIS is designed to be flexible in terms of atmospheric input data and can use one of many sources. The main aim with this thesis is to investigate LIS as a tool in water management in arid /semiarid regions. The two LSMs within LIS were simulated and compared over several investigation points widely distributed over Shiyang river basin, northern China. This was done in order to find possibilities and limitations with LIS and potential differences between the LSM interpretations in this setting. The study consists of two parts. Part one is a field study in an arid area in Gansu, China, during September 2006. Part two is computer simulations using the model framework LIS and its different LSMs with altered atmospheric input over the Shiyang river basin. The aims of the simulations were first to find a good configuration for modelling the area and then to investigate differences in LSM interpretation. Noah and CLM were compared in a four year simulation starting January 1st 2000 and in a created rain scenario to observe infiltration patterns. The field measurements showed average soil moisture of 6.6% in the top ten cm and 11.8% in the 10-30 cm layer during September. The simulations showed the forcing option GDAS to give best performance of precipitation interpolation accuracy. A slightly higher initial soil moisture value than the regional average could give a quicker spin up time. The four-year simulation indicated differences between Noah and CLM in spin up time and soil moisture patterns. The constructed rain event revealed Noah to percolate more rapidly and to a greater extent than CLM. CLM lost water and the reason could be traced to surface and subsurface runoff, rather than evaporation. LIS is still in a developing state and updates are released regularly. Necessary input data was unavailable during the research, due to server problems. Further investigation of soil moisture fluctuation is therefore needed to ensure if any LSM is more preferable in this region. One advantage using LIS is however the possibility to run simulations with different set up and consider all results.