Research On Spatiai And Temporal Features Of Digital Basins And Algorithms For Features Extraction Based On Raster DEMs

Digital basin features are the representation of basin features in the real world, which are the basis of basin hydrologic models and play an important role in hydrological simulation, hydrological problems process and the ecological environment protection. Therefore, extraction of digital basin features is one of the important issue in Geographic Information System (GIS).The research for extraction of digital basin features has made great progress through the efforts of predecessors. However, because the real world is very complex, current research still cannot reflect the basin features truly, which are mainly manifested as follows.(1) It is unable to reflect basins with multiple ownerships caused by rivers branching;(2) it is unable to reflect basins with dynamic ownerships caused by rivers with varied flow directions;(3) there are very little study for bling drainage areas and the extraction of them from raster DEMs. In addition, the current algorithms for extraction of digital basin features from raster DEMs are not efficient, efficiency in processing massive DEMs.According to the defects of current researches in digital basin features, this paper comprehensively studies issues of digital basin analysis, including depression-filling, flow directions analysis, flow accumulation calculation, drainage networks extraction and watersheds division, and make an intensive study of digital basin features and algorithms for extraction, including five aspects as follows.(1) This paper defines the basins with multiple ownerships as "shared basins", and the algorithm of extraction is proposed. In the algorithm, the topological relationship of rivers is built firstly, and then each river is marked with the geocoding which is transformed from downstream rivers to upstream rivers using the topological relationship. According to the geocoding, the rivers with multiple ownerships are defined, and the share basins can be extracted by determining the catchments of these rivers using flow directions from the DEM.(2) This paper defines the basins with dynamic ownerships as " dynamic basins ", and the algorithm of extraction is proposed. In the algorithm, the rivers with varied flow directions are absreacted into "nodes", the nodes and the rivers with fixed flow directions form the rivers network topology. On the basis above, each river is marked with the geocoding which is transformed from downstream rivers to upstream rivers using the topological relationship, and then the rivers with dynamic ownerships are determined. The dynamic basins can be extracted by determining the catchments of rivers with dynamic ownerships using flow directions extracted from the DEM. (3) The bling drainage areas are divided into two categories——topographic bling drainage areas and climatic bling drainage areas in this paper, according to their features, and the algortihm of extraction is proposed. The algorithm firstly restrains the DEM using measured rivers, which is to reduce the elevation of cells corresponding to the measured rivers and make these cells become the depressions. Then the DEM is filled, and the flat surfaces are traversed to recognize the bling drainage areas by judging wether the flat surface contains the inner waterways or the area of depressions in the flat surface are larger than the threshold. The flat surfaces corresponding to bling drainage areas are handled by changing the elevations (equal to the elevations before the DEM filled) to form depressions, and the bling drainage areas are extracted by determining the catchments of these depressions, which are the core regions of bling drainage areas, using flow directions extracted from the DEM.(4) The more efficient algorithm for depressions and flat surfaces processing is proposed. This algorithm takes full advantage of the depression-filling step by recording flow directions from one cell to its neighbor for cells on flat surfaces or depressions. In the step of processing flat surfaces, the flow directions is useful to calculate the distance from cells in flat surfaces to the lower terrain. Moreover, the distance transfrom is used to calculate the distance from cells in flat surfaces to the higher terrain. This algorithm significantly reduces recursive traversals and random data accesses, and increases the efficiency.(5) The more efficient algorithm for flow accumulation calculation is proposed. The algorithm determines the outlets of basins through traversing the edge of the region firstly. Then the basins in the region are processed one by one. Each basin is traversed from the outlet to the interior to build a "flow accumulation tree with direction’" of the basin (The root node of the tree is the outlet of the basin and the leaf nodes are the headstream of the basin). At last, The flow accumulation is calculated by traversing each flow accumulation tree of the basin from the leaf nodes to the root node. The algorithm avoids invalid traversals because the tree determines the process order. Moreover, the algorithm processes basins in the region one by one, which reduces the search scope, efficiently use the resource of the computer, and decreases the exchange of the data in the memory and the data in the disk.The main innovation is as follows:(1) The share basin and the extraction algorithm is firstly proposed.(2) The dynamic basin and the extraction algorithm is firstly proposed.(3) The features of bling drainage areas in raster DEMs are firstly researched, and the algorithm is proposed to extract bling drainage areas. (4) The more efficient algorithm for depressions and flat surfaces processing is proposed, and the more efficient algorithm for flow accumulation calculation is proposed.The experiment shows that the method proposed in this paper can extract dynamic basins, shared basins and bling drainage areas effectively, and performs more efficient in digital basin features.