Professional Documents
Culture Documents
Nathan A. Toké
01/22/04
Table of Contents
1. Introduction
2. Downloading data sets using seamless
3. Unzipping and organizing seamless files
4. Importing files into ArcMap
5. Projecting grid files
6. Reducing grid file size
7. Merging grid files
8. Using hillshade and color ramp visualization
9. Comparing SRTM and NED data sets in terms of slope tendencies
• Step 1 – a common DEM projection
• Step 2 – calculating slope files
• Step 3 – slope files integerized
• Step 4 – clipping slope files to one another
• Step 5 – converting raster (grid) to features (shapefile)
• Step 6 – exporting slope points as .txt
• Step 7 – using Vi to clean .txt files
• Step 8 – using Mat Lab scripts to compare the data
Introduction
This document is a tutorial describing how to download and process DEM data
from the USGS web site (seamless.usgs.gov) using the ESRI products: ArcMap,
ArcCatalogue, and ArcToolbox. I downloaded the Shuttle Radar Topography Mission
(SRTM) 30m data for the Rocky Mountain GEON test bed; which extends in latitude
from 31 to 49 degrees north and in longitude from 103 to 115 degrees west. In this
document I will describe how I downloaded, unzipped, and organized these files. Then I
will discuss how to import grid files into ArcMap, reduce file size, re-project, merge,
rename, and hillshade the SRTM 30m grid files. Finally, I will describe how to calculate
slope files from the DEM data and one way in which I compared SRTM slopes to NED
slopes over the same geographical space.
Figure 1: Define area by coordinates allows you to select a specific area for download (right window).
When selecting an area for download, seamless returns multiple zipped files that
make up the area of selection. The amount and the size of these files vary depending upon
the area you select, so it is best to select an area of consistent size and range, this way the
zipped files will be the same size and their boarders will align. I consistently selected
areas of 1 degree of latitude by 12-degree strips of longitude, from 103 to 115 degrees
west (the width of GEON). This returned 8 zipped files for download, each file had
dimensions of 1 degree of latitude by 1.5 degrees of longitude (figure 2). I limited the
selection area to these increments because seamless requests that you pay for the data to
be sent by CD if the selection area is greater than 100mb of zipped data (about 30 square
arc seconds for the SRTM 30m data).
Figure 2: After defining and area by coordinates, seamless breaks the download area into zipped files. Each
file covers a discrete area. Using the same coordinate dimensions for each selection will provide consistent
file dimensions. In this case 8 files (each 1 degree of latitude by 1.5 degrees of longitude) completed the
area of selection that was one degree of latitude by 12 degrees of longitude.
To complete the download click download on the desired zip file. Seamless
opens a pop up window that displays the data retrieval progress. Eventually a file
download window will appear (figure 3), seamless will select a random numerical file
name. At this point it is important to change the file name to one corresponding to the
geographical area of the file, I chose to name files according to their range in latitude and
longitude. It is also important to save the zip files to folders corresponding to that name, I
chose to bin the 8 zipped files for each 12-degree strip of longitude into a folder named
according to that strip’s latitudinal range.
Figure 3: Upon clicking download for one of the desired files (figure 2) a window will appear that shows
you the file retrieval process from seamless (background). Once the file is ready for download the above
option will appear (foreground). Remember to name the files and store them in a systematic way that
makes them easily identifiable with their geographical location.
Projecting files
The files you downloaded from seamless may have a different projection than you
would like to use when viewing the files in ArcMap. The usefulness of a projection
depends upon several factors including what you will use it to view, the geographical
location of the file, and the projection of any other files to be used in conjunction with the
file you are adding. To change a file’s projection first open ArcToolbox. In ArcToolbox
open data management tools, under that toolbox open the projections toolbox and double
click on project wizard (coverages, grids) (figure 4). (Do not use define projection
wizard because this option only renames the projection, it does not actually change it).
c) d)
e) f)
g)
c) d)
e)
Figure 6 (a-e): Steps to follow in projecting a grid to match existing data.
Reducing SRTM30m Grid File Size
Each point of the SRTM30m grid files is assigned an elevation value (in meters)
with many insignificant digits (figure 7). The presence of these extraneous decimal places
over the entire grid file increases each grid file by several orders of magnitude. To reduce
the file size the grids can be manipulated in raster calculator to produce copies consisting
of only integer point values. This maintains precision at the 1-meter level, which is still
more precise than the 10-meter vertical confidence interval for the SRTM data.
Figure 7: SRTM_48_115 grid file, projected in ArcMap. Notice the highlighted layer’s range of elevation
values. The SRTM data is assigned too many insignificant digits. This increases the file’s size by several
orders of magnitude. I fixed this disk space problem by using raster calculator and the integer function to
produce grid file copies without any decimal places. The range in confidence of the SRTM data is 10
vertical meters.
In ArcMap, add an SRTM30m grid file: example srtm_49_115 (figure 7). This 1-
degree of latitude by 1.5-degree longitude file is roughly 80mb in size (with pyramids
drawn). Under the “Tools” menu, open “extensions”. Make sure that the spatial analyst
tool is available (figure 8a). Then open the spatial analyst toolbar: under the “View”
menu, select toolbars and then spatial analyst to open the toolbar (figure 8b).
a) b)
Figure 8: Steps for opening the spatial analyst toolbar: a) turning on spatial analyst, b) opening the toolbar.
Under the spatial analyst toolbar (figure 9), select “Options”, at the bottom of the
menu. In the “General” tab of “Options” Pick a working directory where you want to
store files created by spatial analyst (figure 10a). Select the “Extent” tab and pick
“Intersection of Inputs” since you will be using raster calculator to make a copy of each
grid value (figure 10b), click ok.
Figure 9: The spatial analyst toolbar contains the functions shown in the above menu.
a) b)
Figure 10: Choose a working directory for files created in spatial analyst/raster calculator to be stored (a)
and set the analysis extent as intersection of inputs (b).
Now open the “Raster Calculator” under the spatial analyst toolbar (figure 9).
Type newsmallfile = int([oldbigfile]) into the raster calculator (make sure your new file
name corresponds to its geographical location, there is one space on each side of the
equals sign, and that there are parenthesis and brackets as shown) and click evaluate
(figure 11). ArcMap will then build a table and evaluate the solution. Once this process is
done a new grid file should appear. The output file should be exactly the same as the
input file less many megabytes in size. This can be assured by subtracting the input file
from the output file using raster calculator and the following equation: Difference =
[oldbigfile] - [newsmallfile]. The difference file will produce a grid from the values
produced by subtracting each of corresponding points of the two grids; all points should
be zero if everything has been entered correctly. The integer truncated grid files are about
20mb, one-fourth the size of the original SRTM30m files. Because there is no difference
between the grid files and since the vertical error in the SRTM 30m data is greater than
the 1-meter integer function truncation used to reduce the grid size, we can be confident
that reducing the grid size is a valuable task for the interest of data management.
Figure 11: “Raster Calculator” interface with an example for how to reduce grid file size for the SRTM
30m data.
Hillshade files are excellent for viewing drainages and patterns of relief (slopes)
over a region (figure 13a), but they do not provide information about the magnitude of
elevation differences over those regions. DEMs provide this information. To make a
really nice figure where both elevations and the patterns of relief can be seen one must
overlay these two maps. Hillshading looks the best if it is done over a DEM in UTM
projection. Hillshading does not work well with units of decimal degrees because as you
move away from the equator the degrees of latitude and longitude are not equivalent and
because the x/y coordinates are much longer than the z units. This results in small
differences in relief appearing as vast hill shadowing effects. In order to see the DEM
through a Hillshade map (or vise versa) one must make the top layer (the Hillshade)
transparent. This can be done by right clicking on the top layer, selecting properties and
under properties selecting the display tab. Now make the topmost layer transparent,
somewhere between 40 and 70% transparent should work well. Now you would be able
to see the information provided in both layers at once if they did not both have the same
color scale. To fix this problem change the color ramp of the DEM layer. This option is
also found under the layer’s properties, this time in within the symbology tab. Make sure
that the values are shown as stretched and then pick a color ramp which displays
elevation differences well. It is best to tinker with this until you find a color scheme that
brings out the regions of elevation best for you (figure12b). Now both the regions of
elevation as well as the patterns of relief are visible (figure 12c).
Figure 12a: An example of a Hillshade map created from SRTM30m data. Notice that you can see the
patterns of relief very well. Drainages can be distinguished and other topographic features are evident.
Figure 12b: A color ramped DEM clearly shows regions of high elevation and low lands based upon color
schemes. White are mountain tops and pale greens represent the low lands.
Figure 12c: A Hillshade map made 40% transparent with a color ramped DEM underneath allows hill
slopes, drainages, and other features of relief to be seen along with the different regions of elevation.
Many other manipulations are available in the Arc products that allow for
improved visualization and can be used to study the geomorphology, hydrology, and
tectonic features of a region using digital elevation models and other GIS data.
Comparing SRTM and NED data sets in terms of slope tendencies
An interesting comparison of the SRTM 30m DEM data to the NED 30m data is
to analyze them in terms of their differences in slope. This comparison reveals differences
between the two data sets answering questions such as: how do SRTM slopes in regions
of high relief compare to NED slopes of high relief? This analysis is important because it
is important to know the tendencies of each data set over different regions of relief. Do
SRTM data sets provide reliable data over regions of high relief? What about the NED
data? As long as this grid point by grid point analysis is done over the exact same
geographical area this analysis is possible. The following steps outline this process. Refer
to the accompanying SRTM and NED comparison report for analysis. File organization is
important, especially with the number of files created for each slope comparison
Step 1: Project SRTM and NED into a common projection. UTM works well
for this analysis. Follow the projection steps from earlier in this report.
Step 2: In ArcMap, calculate slope files for both of these UTM projected files.
To do this first open the UTM files, then select surface analysis from the spatial analyst
toolbar, and choose the slope option. The output measurement should be in degrees, z
factor = 1, cell size should equal the cell size of the input file, and remember to specify an
output file.
Step 3: The slope files must be integerized and multiplied by 1000 in Raster
calculator (Figure 13) (will divide by 1000 in Mat lab). Remember to select the extent of
calculation and the working directory options under the spatial analyst toolbar.