Distributed Hydrological Modeling Based On Spatial Information Technologies

With the change of global environment and the impact of human activities, hydrological processes are becoming more complex. Traditional lumped hydrological models cannot handle these problems effectively any more. Distributed hydrological models cooperated with geography information system (GIS), remote sensing and digital valley technologies could consider the spatial heterogeneity of concerned catchments and are becoming more popular and necessary for hydrological researches. In this dissertation, the theories and techniques of distributed hydrological modeling are systemically discussed, espically the usage of spatial information technologies. The main contents and result are as follows.Extracting digital stream networks and catchment boundaries from digital elevation model (DEM) is the key of distributed hydrological modeling. In this dissertation, DEM acquirement approaches and features of different DEM resources were reviewed. Methods of digital stream networks extraction with DEM were discussed. The HuaTianHe sub-catchment of QingJiang River basin was selected as a case study to extract digital stream networks from DEM. It is found that varying drainage area threshold should obtain different stream networks and the threshold, which makes the drainage density and the average river slope stabile, should be used. The impact of DEM spatial resolution for stream networks extraction was also discussed and the result shows that derived stream networks with various DEM spatial resolutions should be different, which should be affected by the terrain feature of the study area.As DEM is the fundament of distributed hydrological modeling, acquiring high quality DEM data conveniently has become an important task. At present, generating DEM with spatial remote sensing technology has become one of the most promiseful methods. SRTM, the newly released free DEM dataset with a high spatial resolution and the near global coverage, is an excellent and important DEM resource for hydrological researches. To assess the quality of SRTM and its ability for hydrological researches, SRTM was compared with the DEM derived from digital contours. The result shows that SRTM have high quality except the data voids, which should make the SRTM data not suitable for hydrological researches. The widely used voids filling method, which fills SRTM data voids by merging SRTM with lower spatial resolution global DEM datasets, always leads to many errors unsuitable for hydrological researches.Two newly SRTM data voids filling methods were proposed in this dissertation to overcome this shortcoming and make SRTM more suitable for hydrological researches. The first method is filling data voids with ASTER, which is another free high-resolution global DEM dataset, through analyzing the spatial feature of the elevation difference between SRTM and ASTER. The second method is filling data voids at the aid of actual stream networks with a two-step interpolation method. In the second method, voids along stream networks are filled firstly and rest voids are patched with the lower resolution global DEM. The result shows that not only filled elevation values with both proposed methods are better than the traditional method, but also the derived stream networks are similar with the actual stream networks.TOPMODEL, a widely used semi-distributed hydrological model, was used to perform the rainfall runoff simulation in the area upstream of ZhongJianPing in the QingJiang River basin. Both SRTM and the DEM derived form digital contours were used to computer the topographic index for TOPMODEL. The simulation result shows that although topographic indexes derived from different DEMs were much different, the simulation results were similarly good. Moreover, the effect of DEM spatial resolution to the topographic index distribution and TOPMODEL simulation result were studied. It is found that although the derived topographic index and calibarated model parameters are much different, TOPMODEL is not sensitive to the DEM spatial resolution.To improve the simulation quality of TOPMODEL in large area catchments, a distributed TOPMODEL incorporating GIS and remote sensing technologies was constructed. The proposed model splits the whole catchment into several sub-catchments at first, then applies the traditional TOPMODEL in each sub-catchment and routes the runoff water of sub-catchments to the outlet of the whole catchment through the extracted stream networks at last. In addition, to simulate the impact of vegetation to the rainfall interception process, a canopy interception sub-model was incorporated into the proposed model. The land cover, vegetation and soil information derived from remote sensing data was used to calibarate model parameters, through linking model parameters with the actual catchment conditions. The simulation result shows that the revised distributed TOPMODEL should improve the simulation quality of rainfall runoff modeling for large area catchments. To analyze the effect of land use and land cover change to hydrology cycle, the Soil and Water Assessment Tool (SWAT), a widely used distributed hydrological model, was used to simulate the daily runoff process of the study area. The land cover information was derived from the TM remote sensing image and the soil information was extracted with the FAO global soil dataset. Based on the sensitive analysis of model parameters, the model was calibrated and validated with observed hydrological datasets. The result shows that SWAT model is suitable for runoff simulation in the study area.With the calibrated SWAT model, the effect of land use and land cover change to hydrology cycle is analysis. Two TM remote sensing images acquired in 1989 and 2004 were used to obtain the land cover change information in the study area. With the SWAT model, hydrological processes in different land cover conditions were simulated. The result shows that in the past 15 years, the area of agriculture and city expanded and annual runoff decreases while the peak runoff increases. To further study the impat of land cover change to hydrological processes in the study area, five simulated land cover scenario were established. The simulation result shows that, with the increase of vegetation coverage, evapo-transpiration increases, annual runoff decreases and the peak runoff decreases, while the expanding of city, the deforestation and forest degradation should increase the annual runoff and peak runoff.