For a full description of the major components of CSM, click on the links below:
Terrain Analysis for Global Runoff Routing
by
Kwabena Asante, Francisco Olivera, Jay Famiglietti and David Maidment
Center for Research in Water
Resources
The University of Texas at Austin
Introduction
Preprocessing DEMs
Projecting the DEMs
Identifying Inland Catchments
Filling the Sinks
Computing Flow Parameters Flow Direction
Flow Length Flow Accumulation Defining Drainage Outlets
Delineating Drainage Basins
Checking the Delineation
Defining Modeling Units
Obtaining the Data
References
The National Center for Atmospheric Research (NCAR)
is working in conjunction with a number of universities and research institutions to
develop models to enable us to better understand our environment. A major part of this
effort is the development of a global model to study the effects of variations in
moisture, biogenic emissions and other physical parameters which affect the Earth's
climate. The Climate System Model (CSM) is a
fully coupled model with four major component models and a flux coupler. The use of a flux
coupler enables the component models to be run in different programming languages,
at different temporal and spatial scales, if required. Thus only processes which are
similar in their spatial and temporal scale are grouped together in a given model. To view
the major components of the CSM, click on the image below.
For a full description of the major
components of CSM, click on the links below:
The Center for Research in Water
Resources (CRWR) at the University of Texas at Austin,
is developing a GIS based runoff routing model to determining how much water the land
surface discharges into the ocean. The development of the model is being funded by a
National Science Foundation (NSF) grant. The GLOBAL LAND HYDROLOGIC MODEL (GLHM) will use
input generated by the Land Surface Model to compute fresh water discharges into the
ocean, an input for the Global Ocean Model. The GLHM is being developed in two stages
namely, a GIS based Terrain Analysis and a FORTRAN based Runoff Routing.
The purpose of the terrain analysis is to determine
Both of these pieces of information can be determined from Digital Elevation Models (DEMs) of the land surface. 30 arcsecond (or 1 km) DEMs produced by the USGS were used in this determination. These DEMs were developed as part of the GTOPO30 global terrain mapping project. Following is a description of the major steps involved in the terrain analysis.
The raw DEMs were obtained by ftp (file transfer protocol) from the USGS website at http://edcwww.cr.usgs.gov/landdaac/gtopo30/gtopo30.html. They were then converted from their original image to grid format using the arc info IMAGEGRID command. In the original DEM, ocean cells are assigned a dummy elevation value of 55537, while cells with elevations below sea level have a value of 65536 added to their elevation values. These values are corrected by running the following arc info command at the grid prompt:
grid: out_grid = con(in_grid >= 32768, in_grid - 65536, in_grid)
This command results in all the elevations in the DEM being set to the actual measured
values. There will consequently be negative elevation values in the DEM. For this
delineation, an artificial datum of 1000 m above mean sea level was set to prevent such
negative elevations.
Before DEMs can be used in any type of spatial analysis, they must be projected from their original ellisoid onto a flat surface. The map projection used in this analysis is the Lambert Azimuthal Equal Area Projection. Due to their geographic location, different central latitudes and longitudes are used for the different continents. These centers are chosen to minimize areal distortion of shapes in the projected coverages. The projection parameters used in this delineation were obtained from the HYDRO1K Documentation prepared by the EROS DATA Center of the USGS. The projection files for each continent can also be viewed be clicking on the continent name below.
For hydrologic modeling purposes, DEMs must be devoid of pits except in areas with actual inland catchments. The FILL process available in Arc Info can eliminate artificial pits but it also fills in existing inland catchment. In order to prevent this from happening, the major inland catchments in each continent were identified. A NODATA cell is then placed at the lowest point of each inland catchment. This ensures that the catchment is left intact when the FILL process is run. The inland catchments indentified for each continent are shown below. Click on the images for a closeup view.
After the inland catchments have been identified in the DEM using NODATA cells, the Arc Info FILL process is run to fill any remaining sinks. The result of this process is a modified terrain model which allows water to flow uninhibited on its way to the sinks. This modified terrain model is used to derive hydrologic parameters of interest as described below. Click on the images for a closeup view of the modified terrain grids.
The key flow parameters of interest in this study include the following.
Flow direction, which is a measure of the direction in which a given cell is likely to discharge any incident water. Click on the images below for a closeup view of the flow direction grids for the respective continents. The actual flow direction grids are available from the Digital Atlas Module of this CD-ROM.
Flow length, which is a measure of the distance along the flow path (determined by the flow direction grid) from a given cell to its drainage basin outlet. Click on the images below for a closeup view of the flow length grids for the respective continents. The actual flow length grids are available from the Digital Atlas Module of this CD-ROM.
Flow accumulation, which is a measure of the number of upstream cells draining through a given cell. Click on the images below for a closeup view of the flow accumulation grids for the respective continents. The actual flow accumulation grids are available from the Digital Atlas Module of this CD-ROM.
Outlet cells were defined by intersecting a line coverage of the continental margin with a 3 by 3 degree fishnet. A new boundary reach is defined each time the continental margin encounters a new 3 by 3 degree mesh. The figures below show the outlets defined for each continent.
The drainage area of each of the drainage outlets defined in the previous processing step are delineated from the flow direction grid. This process yields the drainage basin of each segment of coastline or inland catchment. As expected the drainage basins thus obtained correspond to the basins associated with the major river systems. The basin of some smaller streams and rivers may be incorporated into that of the larger basins. However, the process yields a handful of drainage basins that can be used for large scale modeling without excluding any portion of the land surface. The commonly used process of delineating drainage basins from a single outlet point cannot be used here because areas that lie outside of the major basins are not included in such a delineation. The flow from these interbasin areas cannot however be neglected in continental scale routing studies.
The delineated drainage basins were checked against those defined by the rivers in the Arc World river coverage. Additional comparisons were made with the
basins delineated in the UNESCO Atlas of World Water Balance
study of 1977, and in the more recent World Resource Institute (WRI) Watersheds
of the World study of 1998. Some differences were observed between the delineations
and these were resolved by revisiting the location of the inland catchments. In the case
of North America, some minor editing of the DEM was required to achieve an accurate basin
divide between the Mississippi and the Great Lakes.
The delineated drainage areas had to be subdivided into a modeling units. Two conditions had to be borne in mind when defining these modeling units. First of all, the number of modeling units had to be small enough for the routing program to handle. If there are too many units, the routing process becomes exceeding slow and the routing program becomes susceptible to memory overflow errors. For this application, it was determined that intersecting the drainage basins with a half degree by half degree mesh would result in a suitable number of modeling units. The second consideration was that each modeling unit had to be associated with a single cell in the mesh over which input runoff is generated. This is necessary to ensure that each modeling unit receives input from only one runoff generating cell. To ensure achieve this association, the drainage basins were further intersected with the mesh of runoff generating cells. The resulting coverages are shown below. Refer back to the Global Hydrology Module for presentations describing how these coverages are used in the modeling process.
The data resulting from the analysis described above are available on the Digital Atlas Module of this CD-ROM.
