Topography

Physical Template - Topography

Topography and River Networks

* Country Scale * Wangchhu River Basin Scale

The first step is to build the base digital elevation model. From the DEM, so-called river networks and flow accumulation grids need to be built. For the scale of the work here, the SRTM is sufficient. Further, it was useful to start with the SRTM-based product Hydrosheds, which has initial error-corrections included (http://www.worldwildlife.org/science/projects/freshwater/item1991.html).

Country Scale (2.5', ~ 4 km, for application using the VIC Hydrology Model)


Digital Elevation Model, River Networ, and Flow Accumulation

As the VIC hydrological model is being set up at a resolution of 2.5' (~ 4×4.6 km), the DEM was first generalized from 3" to 2.5' as follows (Fig.a):

  • The maximum discharge for each 50×50-cell block was determined.
  • The minimum and mean elevation for each block was calculated.
  • Each large cell was assigned an an elevation value between the minimum and the mean, weighted by the natural log of the discharge.
     

This yields a DEM that tends to respect stream channels better than a DEM derived by other means.
The process for the derivation of the flow direction grid (Fig. b):

  • The starting point was the Hydrosheds conditioned flowdirection grid, derived from the SRTM 3" DEM, modified with the aid of mapped rivers.
  • Mean annual TRMM rainfall (below) is summed down this network to yield discharge.
  •  For each 2.5' block of 2500 small cells, every flow that leaves the block is traced up to 31 steps to determine it's destination.
  • Net flows are calculated, and the maximum value is used to set a flow direction on a coarse scale.

The resulting flowdirection grid (Fig.c ) following the rivers much better than a grid derived from a DEM, even the revised DEM.
 

Delineating Drainage Basins

Derivation of slope (Fig. a) can be derived from the DEM, and used for modeling purposes. Next, the river networks (Fig. b) need to be organied by drainage basin. A small portion of Bhutan drains north into the Tsangpo, reaching the Brahmaputra. Most of Bhutan drains south across the Indian plain, reaching the Brahmaputra through a number of rivers. In order to avoid modeling most of the Brahmaputra in order to cover Bhutan, 17 basins were defined (Fig. c). An alternative is to clip the basins as they leave Bhutan. This generates 95 watersheds of 1 to 199 cells, covering the 2225 cells of Bhutan (including every grid cells that touches Bhutan), plus 682 cells that are upstream from Bhutan (Fig. d). This raises the question: what data layers need to extend throughout watersheds that flow into Bhutan. 

 


High-Resolution DEMS: ASTER

For site-specific future studies, where higher spatial resolution is required of a DEM, ASTER-derived DEMs might be of utility  The ASTER scene contains 14 spectral bands, along with a backlooking band used to create the DEM shown here. The highest areas in the north and south of the ASTER tile are clouds. It would be challenging to build a cloud-free mosaic of Bhutan. Close examination of the ASTER-based Global DEM (http://ai-relief.org/auxdetail.html) would be informative. With the recent release of the ASTER DEM, such work could proceed. For example, Fig. (a) shows the DEM derived from SRTM, with one ASTER DEM tile, while (b) shows the quality of the global ASTER DEM for Bhutan. (c) shows the side--by-side quality of SRTM versus ASTER, while (d) and (e0 show the ASTER DEM at 30m and 15m resolution, respectively.

 

 

 

Wangchhu Basin Scale (150m, for use with DHSVM Model)

 Several steps are required to produce a DEM and derived river network/flow accumulation for the Wangchhu.

Projections. Projections: The first order of business is to concur on the geographic projections to be used in developing the data-frame. Multiple exchanges, starting on August 11, between Deki Wangmo, Harvey Greenberg, and Hans van Noord discussed the issue. The new land cover database is based on the Bhutan's standard projection (DRUKREF) developed by Survey of Bhutan. Topographic projections are in the same projection. There are 23 EPSG codes for DRUKDEF 03, including EPSG:5262 (DRUKREF 03, Geocentric, Cartesian coordinate system), EPSG:5263 (DRUKREF 03, Geographic 3D), EPSG:5264 ( DRUKREF 03, Geographic), and 20 Transverse Mercator systems for the dzongkhags. None of these are provided with ESRI's software. The national system appears to be simply geographic coordinates based on the GRS1980 spheroid (is that correct?). Geographic 3D CRS details were forwarded by Hans. The “final” decision was 5262..

Selection of DEM. To build the digital elevation model of the Wangchhu, for the DHSVM and related applications, it is necessary to select the most appropriate data set to use. The options include:


Option 1: The SRTM 3-second data, pre-conditioned by Hydrosheds. This is the dataset used for VIC, where it was aggregated up to 1/12o. It has an advantage, in that it has already been prepared for hydrologic applications.
Option 2: ASTER DEM, developed from the 2009 release of the global ASTER DEM. A Version 2 has just been released.
Option 3: ALOS. There is an ALOS-derived DEM, but, according to Harvey, it is not clear that it has been georegistered.
Option 4: DEMS derived from local topographic/detailed survey data. I remember having been told about such a dataset (at NLCS?).

Decision factors include:
Factor 1: What are intended uses: (1) DHSVM modeling, (2) multiple others, that would have their own requirements.
Factor 2: Amount of work to produce the “best” product. The most straight-forward would be the SRTM product, but it would produce no better than a 3”(~90m) resolution. If other applications wanted better, it wouldn’t suffice.
Factor 3: Consistent with VIC/country-wide. Not critical, but a factor.

Harvey Greenberg did a detailed analysis of the SRTM/hydrosheds relative to ASTER, relative to defining stream channels and slopes
(http://rocky.ess.washington.edu/areas/Bhutan/streamslopes/index.html) (Figure a-d)..


Worldwide DEMs such as the SRTM DEM, especially the publicly available 3" (~90m) version, leave one struggling the define channels, much less calculate slope (a). We have leaned heavily on the HydroSHEDS version of the SRTM DEM. Longitudinal profiles of rivers were created by tracing from cell center to cell center. These profiles never (in the conditioned DEM) run uphill, and their geographic coordinates follow the actual rivers as well as human experts could determine. However, slopes cannot be trusted, if only because they are often stairstepped, with apparent flat spots and bogus waterfalls. Even a perfect DEM will generate problems in all but the steepest rivers when elevations are rounded to the nearest meter. Once a profile graph is extracted from the GIS, there are myriad ways to smooth it. We adopted the strategy of extending our smoothing window to the next highest and lowest points, and imposing the additional constraint of looking a minimum distance (1, 2 or 4 km) upstream and downstream. The results can be seen below.

Results were then compared to the 1" (~30m) ASTER Global DEM. Maintaining the stream points from Hydrosheds as our official river course (a), we
• Converted the points to 1" cells in the space of the Global DEM. As the GDEM cell centers fall on the edges of Hydrosheds cells, there is a uniform 0.5" shift, but this does not worry us. In fact it partly compensates for the 1.5" NE shift of the HydroSHEDS data.
• Calculated the locus of 1" grid cells closest to each streampoint, clipping to an arbitrary buffer of .05 degrees. These regions, called Thiessen polygons or Voronoi cells before clipping, were computing with the eucallocation function in ARC/INFO.
• Found the lowest elevation within each region. (zonalmin)
• Assigned that elevation back to the point
• For diagnostic purposes, those lowest cells were converted to points and are displayed below as stars. In the case of ties, more than one point may be shown.
 

For the Wang Chhu, the buffer size looks good. Too small a buffer will fail to catch useful points. Too large a buffer will catch points that are legitimately lower than the river channel.
 

Here are three versions of the resulting profile of the Wang Chhu (b) . The HydroSHEDS profile looks well behaved, though the flat regions are unlikely to be real. The SRTM DEM is shows some spikiness. It would have looked better if we had been able to obtain a 3" DEM generated as the minimum of the 9 component 1" values. The ASTER GDEM is more spiky, and it shows some low spots that we would like to take a look at.
 

To the left is the SRTM image of an area in the second box of the context map (c) . The graph on the right is a detail of the full river profile. This area concerns us. If the GDEM contains bogus pits, our data filtering becomes much more difficult. The image on the right shows the 3" hydroSHEDS DEM, the profile created from it, and red points selected where the GDEM elevations were selected. The two big blue points are the troublesome pit on the graph.

 This is the 1" Global DEM with a special color stretch (a) . The dip in our river profile is clearly visible as a feature of the river channel, comprising many data points and centered on the channel.

This leads us to conclude that these are legitimate elevations, and that apparently higher areas downstream are artifacts of tree cover. The two pixels have 9 and 10 components values (The minimum value from 9 ASTER scenes was used.). This is typical for the area, though it seems that high points on the profile have lower data quality. We will look at this statistically later.

Here is one artifact of the method. Profile points have a small search area on the inside of curves. If the actual channel lies to the inside of the curve, that particular HydroSHEDS point is inhibited from searching the channel for a data point. The result is a spike in the profile. This is not a big problem, as we expect upward spikes and remove them.

 

Here is a similar artifact. Because of a poor match between the profile and the actual channel, a profile point has searched across an oxbow to find a data point. Such an error could result in a downward spike, but here it merely produces an upward spike.

Here we see our methodology start to fall apart as the Wang Chhu becomes a braided river near the Indian border. The river has moved considerably between the Landsat (~2000), SRTM (Feb, 2000?), and multiple ASTER passes. This is the first place where I suspect that a floodplain is lower than the surface of the river. This is way upriver at about 2000 meters.

For the most part, the GDEM points define the channel better than HydroSHEDS, but there is some unacceptable noise in their path. Would it be possible to automate a redefining of the channel course? Would it be worthwhile to build a GIS team in Bhutan to refine the data sets? Don't forget tie scores. And what if we constructed perpendicular bisectors at each and searched for an alternate lowest point? Not so good, I suppose. Now we have re-visit our smoothing algorithm. For now we will assume that all bumps are artifacts and that no low points will be discarded. As our data is in integer meters, we cannot (except where points are discarded) represent non-zero slopes less than one meter over thirty meters. This is unacceptable even in Bhutan. We have run this procedure on most of the Brahmaputra basin, and are starting to evaluate the results. Version 2 of the ASTER Global DEM looks much better. We are looking forward to testing it with our existing algorithms.

Harvey briefly evaluated new Version 2 of the one-second ASTER DEM for Bhutan and transformed it to their Transverse Mercator coordinate system. It looks much better than version one, but it would take a lot of work to make it behave as well as the HydroSHEDS 3-second conditioned DEM.

There are, of course, multiple applications for a DEM. For the purposes of the DHSVM 150-m modeling, it is judged that the SRTM Hydrosheds 90-m product is the most suitable, as any possible incremental gain of the Version 2 ASTER product would not be worth the effort.

But there are other applications. As noted by Pema Wangdi, applications of the Version 2 ASTER at 1- arc-second (30m), ½ arc-second (15m), and the 1-arc-second SRTM Digital Terrain Elevation Data (SRTM DTED, void filled version of the National Geospatial- Intelligence Agency) in the future. Further, there is 30-m national DEM (of unclear orign). DGPC has very high resolution DEMs, developed for specific hydropower siting applications. These should all be made available in the future.