> Lab 6 - GEO 580, Spring '02

GEO 580

LAB 6: Surface Analysis

Suggested time for completion: One week

Outline:


6.1.  Purpose

6.2.  Introduction and background

     The term Digital Terrain Model (DTM) is a generalized term for any digital representation of a topographic surface.  Three such representations are DEMs, TINs, and contour coverages:

     Note that while actual real-world topography is the most common surface that GIScientists wish to represent, there is in fact a great variety of continuos spatial information that is well represented as a surface.  A short sampling could include predicted rainfall, temperature, earthquake probability, mineral or pollutant concentration, and human population (or another animal, for example; Ducks Unlimited has done a study where they kriged duck counts to make a map of duck density in the prairie-pothole region -- mountains of ducks, as it were).

     This lab will focus on topography, but keep in mind that many of the issues and techniques discussed will be applicable to these other kinds of continuous geographic information.

Basic information on USGS DEMs
(From the Rocky Mountain Mapping Center)

     The US Geologic Survey is in the process of creating a national database of DEMs.  The source data for this project includes contour maps and stereo pairs of air photos.  Several methods have been used to create DEMs, which you can read about on the USGS page on DEMs.  These methods can be manual (e.g., having a human estimate elevation along a regular grid) or more-or-less automated (interpolating a digitized contour map into a DEM).  Many of these techniques leave artifacts (artificially introduced patterns in the data) in the dataset, however.

     The USGS produces DEMs at a variety of scales.  Much of this is currently free to download in SDTS (Spatial Data Transfer Standard) format. The EROS Data Center hosts the USGS DEMs and DLGs at the USGS Geographic Data Download site.
 

The Future of DEMs: SRTM

     In February 2000, the Shuttle Endeavor flew the Shuttle Radar Topography Mission (SRTM), and in ten days it mapped 80% of the Earth's land surface -- 94.59% of that twice.  SRTM should provide almost complete DEM coverage of areas between 60 N and 60 S, at a high resolution of about 30 m.  This is equivalent to the large scale 7.5-minute DEMs (also about 30 m resolution), but with uniform coverage both inside and outside of the US and without the artifacts characteristic of DEMs generated in other ways.  The 30 m resolution is also equivalent to Landsat TM resolution and thus provides rich opportunities for combining data.  The data is currently being processed and distribution will begin in November 2001.

     Currently, samples of the SRTM dataset have been released:
 
A bug specific to this lab:
  • The spatial reference of the data from occasionally seems to cause problems.  A dataset may have a spatial reference when viewed in ArcCatalog, but then apparently not have it when viewed in a data frame in ArcMap. 
    • Workaround: Create new layer with the same (?) projection info in ArcCatalog, and use that.
    • Also, use a frame with no spatial ref. when your data has none -- e.g., TINs


6.3.  Data
For more information on USGS DEMs:

Download the data for lab 6 to your c:\temp drive.  It contains these layers:

Copy the data to your local work folder.


6.4.  Procedures
 
Your map for Lab 6: 
     For this lab, you will be using ArcMap data frames to make your maps.  Every time you have dropped data into ArcMap, you have put it into the default data frame.  What we will do in Lab 6 is create three more data frames.  Each of your four data frames will hold a separate map, and you will print out and hand in all four maps printed out on one page.  This will be a challenge -- focus on communicating one idea with each frame.  You will also have to think carefully about layout, what information is really necessarily for these maps (you may decide to leave out a few things recommmended in the basic guidelines). 
     An introduction to data frames is provided below.

 
Introduction to Data Frames

     Go to ArcMap.  You will notice that your default view is the analysis view of Data Frame 1, named Layers.

  • Rename the frame Layers to Data Frame 1 (or another name you prefer).
  • Add three more data frames. 
    • Go to the Layout view () in ArcMap.
      • This would be a good time to change the page layout from portrait to landscape (File --> Page Setup).
    • Go to the menu bar and hit Insert --> Data Frame.
      • Position this data frame and the original one in the upper-right and upper left quadrants of your page.  Put the next two frames in the lower quadrants.
      • Note: you could also copy (CTRL-C) your first data frame and paste it (CTRL-V) three times.
    • Give the frames appropriate names like Data Frame 1, 2, etc. (you can change them as you move through the lab).
  • Some notes on frames:
    • Active/Inactive Data Frames: Only one data frame can be active at a time.  A frame must be active for you to drag data into it, view the data, or edit the data.
      • After you return to Data view: to switch the frame you are looking at, right-click on the data frame heading and select Make active data frame (Note: command may be called Activate in 8.1).
    • You can drop layers from ArcCatalog into any of the data frames, or move or copy layers between frames.
    • Once you've arranged the data for each frame the way you like, you can return to Layout view and resize the frames if necessary, as well as add titles, etc. 

6.4.1. Methods of DEM display

Answer Question 1 : While you were classifying the DEM, you may have noticed an unexpected pattern in the histogram of the data.  Go back and look at the histogram again.  The distribution of elevations is not what you would expect if the elevation had been directly measured for every pixel (as would be the case in, for example, SR-TM data).  Describe the pattern and explain why it is there.
     There are many other options for displaying DEMs, although using them in ArcInfo8 will depend upon the availability of various extensions such as 3-D Analyst (More information on ArcInfo extensions).  Some common ones you will see:

6.4.2. Data Structures For Digital Elevation Data

     Review the Introduction to this lab and the Lab 2 to make sure you have a clear idea of what data structures are and what the main kinds of DTMs are.  Often, DEMs will retain artifacts that would not be seen if the DEM had been generated from actual elevation measurements [estimates] for each pixel of the DEM.  Also, converting elevation data between different data structures, while useful for many purposes, will always result in loss of data and often have severe effects on accuracy or other important features.

Contours. First, we are going to create a contour coverage from our DEM sbdem.  We will space our contours every 200 vertical meters.
 
 

Possible bug:  If you have difficulty creating the contour layer due to "not having a licence for TIFFZW" or due to not having the "TIN Wizard" installed, a contour coverage (sbcontour) has been provided in the dataset.  Use this for the lab.

 
First, find the Contour Wizard in ArcToolbox.  If you have trouble finding it, go to ArcToolbox's Menu Bar --> Tools --> Find.
    • Use sbdem as your input surface.
    • Make the contour interval 200 m.
    • Keep the base contour at 0.
    • Keep the output item as CONTOUR. Note that this will be an item in the contour coverage's AAT.
    • Change the default weed tolerance to 50, but click on the help button for this step and read about weed tolerance.  There will be a question on this later.
    • Keep the output z-value in meters.
    • Name the output coverage sbcontour.  Make sure it ends up in your lab6 directory.
Double-check that you did everything right by reading through the last step, the Summary of your input.  Click Finish.
Your map -- Data frame #2: 
     Use a contour map of mainland Santa Barbara county elevations for your second data frame. 

TIN Creation. Now, we're going to create two TINs from our elevation data.
 
 

Possible bug:  When converting from Grid to TIN, first try the tool in the IMPORT TO TIN folder, rather than then EXPORT FROM RASTER folder.  In theory it is the same tool, but when running from one rather than the other you will occasionally receive an error that you are not licenced to use TIFFLZW.  If you receive this error, try the tool in the other folder.  If you still can not make the tool work, consult with your TA.

 
First, from our DEM: Find the ArcToolbox tool Grid to Tin.  
    • Use sbdem as the input grid.
    • Keep the defaults.
Name your output sbdemtin.

 
 
Possible bug:  When using the Create TIN Wizard, you may encounter an error that you are not licenced to use TIFFLZW.  If this occurs, use the sbtin dataset provided in the lab data.

 
Second, from our contour coverage sbcontour.  Find the ArcToolbox Create Tin Wizard:
    • Step #1: Click the  button.
    • Step #2: Use sbcontour as your input coverage.  Leave the other options as defaults and click Next.
    • Step #3: Leave the Specify a Z-value constant window blank, but click the toggle for the second item,
      • like so: 
      • Select CONTOUR as your Z-value item.  Click Next.
    • Step #4: Click the first toggle and select 'Softline': 
      •  Click Next.:
    • Step #5: Click Next.  We don't need to select specific features.
    • Step #6: We are now back to choosing a dataset for the TIN.  You have just added sbcontour, so now select it and click Next.
    • Step #7: Name the output TIN sbctin.  Click Next.
      Click Finish.

 
Answer Question 2 : DEMs and TINs are often referred to as "2 1/2-D" rather than "3-D."  Why do you think this is?  Hint: In a surface, how many z-values are there going to be for a given (x, y) coordinate?  In a system that requires a truly 3-D structure, how many z-values are possible per (x, y)? 

For additional help, you may wish to try (linked 9/19/00): http://www.users.bigpond.com/dsandison/Chap3_final.htm

Your map -- Data frame #3: 
     After answering the questions dealing with the comparison of the two TINs, leave the best TIN in frame 3.  We will use it as part of our viewshed analysis map (see below).

 
Answer Question 3 : Upon visual inspection, what are the differences between sbdemtin and sbctin?  Consider: level of detail and information lost in the creation of each TIN.  Also use the identify tool and click around on the Santa Barbara Channel for both TINs.  Explain the elevation results for the ocean in both TINs and the reason for the differences between them.

 
Answer Question 4 : Examine the histograms for your two TINs.  What are the unnatural patterns in each and what are the likely causes of them?

Topographic Shading in TINs. You can see that your TINs display topographic shading by default:

Answer Question 5: What kinds of geographic information would be most useful to you if you were seeking to mitigate the major problems you have discovered with the TINs above?  Especially consider: what would be the best source data to use for generating TINs?

 
Definitions (from the ESRI GIS Glossary):
Displaying Slope and Aspect
     When we created our TINs, slope and aspect attributes were created for each triangle along with elevation.  These can be displayed as their own layers or used for analysis.  Common topics where slope and aspect are important features to consider include erosion and landslide hazard assessment, vegetation modeling in ecology (the north face of slopes is often cooler and wetter and may grow different species), and forestry (aspect and slope will influence the quality of a site for tree growth, and may favor certain species or varieties).  As mentioned previously, slope and aspect are also used in the calculation of topographic shading.
     We are going to practice displaying and analyzing slope.
 
To Display Slope: 
    • Create a new layer from sbdemtin in ArcCatalog.  Name it slope_sbdemtin.
    • Still in ArcCatalog, go to Properties-->Symbology for this slope_sbdemtin.
    • Highlight Elevation, like so:
    • Uncheck Edge types, Elevation, and Show hillshade illumination so that they won't display. 
    • Click the  button.  The Add Renderer comes up:
    • Highlight the line "Face slope...", then , then .
    • Pick a logical color ramp that communicates degree of slope (e.g., red for high slope, blue for flat).  You may want to return to the Style Manager to find or create a good one.  Precipitation, with the color order flipped, seems to work well.
      • Note: a nifty trick is that you can Flip Symbols by right-clicking in the Symbology editing window (see graphic below).
    • Decide on how many classes you want to group slope into, and classify the ranges manually if necessary (look under the  button.
    • Make sure the  button is unchecked (it appears to re-check itself if you switch between Elevation and Slope displays in the Symbology tab).  Your final Symbology might look something like this:
    • Click OK.  You have now created a layer file that will retain your Symbology setup when you transfer it to ArcMap. 
    • Drag it into ArcMap.

 
Answer Question 6 : At 8 p.m. on July 31st, 2000, a landslide occurred about 26 miles west of Santa Barbara (near the Tajiguas landfill).  "Hundreds of tons" of earth slid towards the beach, leaving "a 300 foot section of track hanging in midair."  Fortunately, sensors stopped two approaching freight trains, but an Amtrak train had passed over the tracks "just hours earlier."  (Van de Kamp, Santa Barbara News-Press, 2000)  It was a perfectly calm day in the middle of a dry summer. 
     Imagine that you have been hired by Union Pacific Railroad (http://www.uprr.com/ ) to try and use GIS to try and identify other areas where landslides are a possibility.  Obviously in real life, one would need specialist training or advice to adequately do such a job; but you should be able to lay out a general strategy for such an analysis (what data layers you would need and what kinds of analysis you would do).  Do so in one paragraph.

 
Your map -- Data frame #4: 
     Use your slope map for frame 4.  Make sure that the color scheme will show when printed in black & white printing.  Also make sure that your map communicates a reasonable balance of slopes -- e.i., don't have a 'high slope' class that is so restrictive that only an insignificant part of the map shows up on it.

6.4.3.  Analysis with DEMs

     First, we will conduct a viewshed analysis.  Viewshed analysis is useful for several purposes, such as determining the views available from roads, lookouts, and housing developments.  For example, logging or building may be restricted in National Scenic Areas or areas viewable from highways.
 
 

First, we have to create a point feature coverage from which to do the viewing.
    • Create a point coverage for the point from which you want to do the viewshed analysis.  Pick a location you are familiar with (your home, campus point, the top of the Highway 154 pass, etc.).
      • Remember that new features must be created in ArcCatalog, then transferred to ArcMap.  They are then edited to create the geometry.
      • Knowing this, create the (empty) point coverage in your lab6 directory (Don't remember how? Hint: When in doubt, right-click)
      • Name it viewpoint.
      • Use sbcontour to define tics, boundaries, etc.
        • If the sbcontour's projection has not been defined, use sbdem to define the projection
      • This will be a point feature (although one can perform viewshed analysis on a line feature such as a road or river, for example).
    • Put viewpoint in ArcMap.
      • (Note: If you get the warning like "One or more layers lacks spatial reference information" -- this means either that you didn't define a projection or datum for one of your layers, or a bug occurred and this information didn't get transferred with the layer.  If the latter is the case, you will probably have to use the Project[ion] Wizard tool to create a new coverage, and then use that in ArcMap.)
      • Make sure the Editor toolbar is displayed and start editing viewpoint.
      • Place a viewpoint in a familar location.
      • Save your edits and stop editing.
Question #7 : What is your assessment of the viewshed?  Does it fit with what you can actually see from that point?  What would account for differences?

 
Your map -- Data frame #3: 
     In frame 3, use the TIN you have left here to represent elevation.  Then display your viewshed and viewpoint on top of this.

More on ArcInfo 8.x Extensions:
     Much of the GIS functionality needed for scientific research comes from extensions to ArcInfo.  These are software add-ons that make new tools accessible.  When installed (they must be purchased from ESRI, of course) they will be accessible as new tools in ArcToolbox and/or as new toolbars in ArcMap (the latter are those drag-and-drop toolbars, such as the Layout, Draw, Effects, and Network Analysis toolbars, which you have accessed from ArcMap's Menu Bar --> View --> Toolbars).
     You are probably familiar with ArcView extensions such as Spatial Analyst.  As ArcView is folded into ArcInfo 8.1, ESRI intends to make the functionality of these extensions available in ArcInfo extensions.
ArcNews article including extensions in 8.1
More on ArcView extensions from ESRI's webpage
More on ArcInfo extensions from ESRI's webpage
Extensions in ArcView/ArcGIS/ArcInfo 8.1
     Our second analysis will examine the relationship between elevation and vegetation.  Drag sblandcov (we worked with it in Lab 4) into ArcMap.  Because the version of the software as of this writing lacks functional extensions such as Spatial Analyst, we won't do direct queries of grids (this could be done with ArcInfo Workstation, however).  We will use the skills we have learned for displaying and converting data and then investigate the questions by visual inspection.  You will have to decide how best to display the vegetation and sbdem to answer the questions below.  You will also have to join the table landcov.cnddbnames to the sblandcov coverage. You will use the CNDDB1 field for the join field.
 
 

Question #8 : Describe the vegetation profile as you move along a straight line from Campus Point to the east edge of Lake Cachuma.  Make a list of the vegetation types in the order that the line crosses them, and give approximate elevation ranges for each of the vegetation types. You may use the following coordinates (on the dem cs): from 105000/3620500 to 102000/3637000

Note: There are several decent ways to do this. 


 



6.5.  Conclusion

    In this lab, you learned about the basic data structures used for surfaces in ArcInfo 8.  You can see that each has advantages and disadvantages in terms of conveying information, in conducting various analyses, and in accurately representing the real world.  You should also have an appreciation for the uses and hazards of converting surface data between structures.  Surface modeling is a key part of many GIS analyses, but the results attainable will depend upon the data and software tools available.  Most important, however, is the intelligent use of the data and tools by the GIS user.



What to turn in
More info on DTMs and Surface Analysis:
Sources
Weibul, R., and Heller, M.  "Digital terrain modelling."  In Geographical Information Systems: Principles and Applications, edited by Maguire, David J., Goodchild, Michael F., and Rhind, David W.  London: Longman Scientific and Technical, 1991, pp. 269-297.

Lab originally created by Nicholas Matzke and Sarah Battersby
UC Santa Barbara, Department of Geography
© 2000, Regents of the University of California; redistributed by permission

Last update: May 28, 2002
http://dusk.geo.orst.edu/buffgis/Arc8Labs/lab6/lab6.html