Abstract
In order to develop hydrological models for flood and water resource analysis of river watersheds continental scale watersheds such as Yangzhy River(China). Mekong River, Brahmaputra/Ganges River, "Drainage Model", which has surface flow direction for each pixel in raster base, is very much required even in rough resolution of 1 km. In this paper, procedure to produce the drainage model is reported. In the first step, method to generate the flow direction using only the GTOPO30-1km digital elevation data-is introduced. However the generated river stream is not match with the stream lines in DCW(Digital chart of the world) and/or classified rivers in the land cover map using NOAA data. In the second step, a method to correct the primary drainage model is introduced, in which DCW and USGS 1 km land cover map is used to make reasonable stream lines. After the drainage model is established, topographical characteristics of watersheds, such as terrain features, land cover distribution and NDVI seasonal changes are analyzed for main rivers in the world. The developed drainage model can be applied to hydrological modeling.
1. Introduction
Drainage Direction Matrix (DDM) is defined as mesh data where each pixel has flow direction showing where the rain and the water body on a mesh flows. By making DDM, boundary of watershed can be delineated, calculation of watersheds area and hydrological analysis based on watersheds will be possible in computer. In order to make DDM, Digital Elevation Model (DEM) is used. If DEM express the detail of the real topography of the land, the flow direction of each pixel can be easily set. However, DEM usually include some error, and it represents a specific heights (ex. Mean Maximum or Minimum Height ) depending the mesh size, so it is difficult to obtain a consistent DDM from DEM.
The problems are the existence of concave pixel which has lowest height among the surrounding 9 pixels and no flow direction, and the existence of flat area where the heights of the neighboring 9 pixels are same. Both of the problems are quite complex and are picked up as subject for many study papers (Hayakawa, 1995, Lu, 1995).
And some solutions are also suggested (Nogami, 1998 ). But even the problems are solved, the computer extracted river systems are not always similar as actual river systems or map extracted river systems. In this case, manual correction comparing the computer products with reliable products are required however it will be difficult if the data size become huge.
In this paper, global DDM excluding Antarctica are produced using GTOPO30-global 30 second are DEM provided by USGS and some other global data sets. A similar products are now available as "HYDRO 1K, Elevation Derivative Database" at USGS which includes data sets of (1) Aspect, (2) Flow Direction , (3) Flow Accumulations (4) Slope (5) Compound Topographic Index. The URL of the site is :
www.edcwww.cr.usgs.gov/landdaac/gtopo30/hydro/index.html
However the available data of "HYDRO-1K" are only two-North America and Africa- and other continents -South America, Asia, Europe ad Australia- are still under the processing (October 1998 ). Furthermore, the algorithm and the accuracy of river network systems are not clear in "HYDRO- 1K" at this moment. This paper describes the detail of the developed algorithm for DDM generation and reliability of the extracted river systems.
2. Used Data
The used data to generate the global DDM are as follows :
GOTO30 30 seconds are Digital Elevation Model of the world lands, provided by USGS. The accuracy of the data is not qualified because the data sources are varied. GOTO30 is one of the precise mesh elevation data covering the entire world land area. The data is also available at following site, and detail of the data can be obtained.
www.edcwww.cr.usgs.gov/landdaac/gtopo30/gtopo30.html
DCW River Data Digital Chart of the world (DCW) includes the world river system (stream line ) data as "DNNET" coverage in LINE feature. Because the data is ARC/Info format, it is converted to 30 seconds grid raster format
WBDb-II River Data world Boundary Databank-II is data sets provided by U.S. Department of state (DoS)/ Central Intelligence Agency (CIA), and it includes the global river system data. Because the data is ARC/Info format, it is converted to 30 seconds grid raster format. The data density is no so high as DCW river data, and it means that river systems in WBDb-II data set are more significant river lines. The data can be obtained from UN/GRID
USGS 1 km Land Cover The data is provided. It is classified image using 1 km 10 days compo9site data from April 1992 to march 1993 combined with other global data sets such as GTOP30 and existing vegetation maps. There are 6 maps provided based on 6 category systems such as 1. Global Ecosystem Legend, 2. IGBP Land Cover Legend , 3. USGS Land Use/Land Cover system Legend, 4. Simple Biosphere Model Legend, 5. Simple Biosphere 2 Model Legend, 6. Biosphere atmosphere Transfer Scheme Legend. In this paper , " 3. USGS Land Use/Land Cover System lend" is used to get information of "water body" distribution which is not described in DCW nor WBDb-II data sets. The data can be obtained from following URL:
www.edcwww.cr.usgs.gov/landdaac/glcc/glcc.html
