The data you need to use is all in example.zip.
There is an optional orthoimage geotiff file with 1 m pixel size covering
the area of the example in sampleimage.zip.
This is big (16808 KB), but quite cool, so download it if you have a fast
connection.
In a new folder extract the contents of example.zip.
You should get 4 files:
sample.asc. ASCII file of digital elevation model data.
samplels.shp, samplels.shx, samplels.dbf.
ArcView shape file components for point landslide sample data.
Open ArcView (with SINMAP extension loaded) and open a new View.
Select File/Import Data Source/ASCII Raster then select the file 'sample.asc'.
Select grid name 'kilp' [This is data from the Kilpala pilot study described
in the manual.]. Respond NO to cell values as integers and
NO
to add grid as theme to the view. From the SINMAP menu 'Select DEM
grid for analysis'. Select grid theme 'kilp'. This adds the
theme to the view, providing names that the remainder of SINMAP will recognize.
From the SINMAP menu select 'Make single calibration region theme'.
Click OK without changing any of the default parameters. [This is
one place where you would set parameters given geologic information.]
This adds a calibration regions theme to the view. The table associated
with this theme contains model parameters. The theme just created
has the same parameters everywhere. If there was soil or geologic
information this could have been used with the 'Create multi-region calibration
region' to specify different parameters for each soil or rock type.
From SINMAP menu 'Select landslide point theme'. Choose the 'samplels.shp'
file. This adds locations of known landslides that will be used in
calibration. If you downloaded it add the sample image theme now.
Select 'Add theme/image data source', file 'kilpsub.tif'. You should
see an orthoimage of this area that may be used to cross check the precise
location of observed landslides. Data importing is now complete.
The processing is divided into two steps. Grid processing is only done
once since it depends only on digital elevation data. Stability analysis
requires iteration on parameters.
Grid processing. Select 'Compute All Steps' from the 'Grid Processing'
part of the SINMAP menu. You should see the blue action bar at the
bottom and various messages flash by. Once this stops 4 themes (pit
filled dem, flow direction, slope and contributing area) have been added
to the view. These are from the Dinfinity method so are different
from the results of the hydrologic modeling sample extension 'hydro.avx'
supplied with ArcView. Pit filling uses a routine we developed because
we found that the ArcView one with the 'hydro.avx' extension has bugs.
[For example download the file pittest.asc, import
as a grid and run the SINMAP pit filling and Hydro pit filling and compare
answers and decide for yourself which is right.] Flow direction is
expressed as an angle in radians, measured counterclockwise from east.
Contributing area is specific catchment area expressed in area per unit
contour length, i.e. it has the same units as 'map units' which for this
work is meters. The 'contour length' used for each grid cell is the
cell size (10 m here). The contributing area has been evaluated using
the proportioning between downslope cells, i.e. a multiple flow direction
approach, that is part of the Dinfinity method. This calculation
also includes a check for edge contamination. This is the
possibility that some terrain outside the bound of the DEM analyzed could
be part of the contributing area of grid cells within the DEM analyzed.
If the program detects this it reports the contributing area as 'no data'.
This is the reason for gaps near the edges and along streams that come
in from the edge of the map.
Stability analysis. Select 'Compute All Steps' from the 'Stability
Analysis' part of the SINMAP menu. You should see the blue action
bar at the bottom and various messages flash by. Once this stops
there should be two views. One is the map view seen previously with
two new themes, stability index and saturation. The other is a slope
versus contributing area x-y plot. The theme in this plot is slope
and contributing area data sampled randomly from the DEM (the dots) and
the landslides (symbols). Select the theme 'Region 1' (this is the
whole area - there is only one calibration region) in SA plot then select
'Add/Delete stability equation lines' from the SINMAP menu. These
lines depict the way the model divides terrain based upon slope and contributing
area into the different stability and wetness zones mapped in the stability
and wetness themes. The idea behind calibration of the model, is
to adjust the parameters and shift these lines to optimize the location
of landslides in zones mapped as unstable, while minimizing the area alienated
in these zones, i.e maximizing the landslide density. Right click
in the SA plot and select 'Statistics'. This displays landslide statistics
information. [Note that the statistics menu is greyed out if the program
thinks parameters have been changed since the grids were last evaluated,
so in certain cases you need to right click and select 'Update grids and
lines' before being able to select 'statistics'.]
Questions to hand in.
-
How many landslides are there mapped in the Kilpala sample area?
-
What are the default stability parameter ranges in SINMAP?
-
Indicate, for the sample area, how many landslides fall in each stability
class using the default parameters?
-
Compute the landside density (in landslides per km2) for combined
defended and upper threshold classes by adding the number of landslides
in each class and dividing by the combined area of the two classes.
The interactive philosophy of SINMAP includes analysis of individual
landslide points that may be outliers. The trusty dog 'Rex' may be
used to sniff out the locations of points. With both SA plot and
map views open, click on Rex 
in the menu bar. Then click on a landslide point in SA plot to see
it's location highlighted on the map view and visa versa. [This is
a like view version of linked tables].
Question to hand in.
-
What is the x, y location of the landslide point with the largest contributing
area?
Calibration. Parameters may be adjusted by editing the parameter
table associated with the 'calibration regions' theme, or interactively
by right clicking and selecting 'Adjust Calibration Parameters' in the
SA plot view. After parameters have been changed you need to select
'Update grids and lines' to recompute stability and wetness grids displayed
in the map view. Fiddle with the parameters a bit to get a feel for
how they affect the lines on SA plot and the mapped stability and wetness
classes and landslide density of each class. Try and adjust the parameters
to maximize the landslide density for combined defended and upper threshold
classes.
Questions to hand in.
-
Report your best parameter set and corresponding maximum combined defended
and threshold density. [We will have a competition in the class to
see who gets the highest.]
One of the features of SINMAP is that it maps instability based
upon slope and contributing area following the logic that water and hence
contributing area is a significant causitive factor in landslide occurrence.
A simpler model might have ignored contributing area and only considered
slope. Here the goal is to demonstrate the benefit of also using
contributing area for terrain stability mapping. [Benefit is defined in
terms of higher landslide density for regions mapped as unstable.
This translates into a financial benefit for loggers who can minimize the
area 'alienated' from logging due to stability concerns. There are
also environmental benefits in terms of reduced landsliding and sediment
delivery to streams by more accurately demarcating potentially unstable
terrain and limiting logging or other disturbance activities (e.g. road
building) in these areas.]
Determine the slope associated with each landslide. This requires
looking up the value of the slope grid theme at each landslide location.
This is achieved as follows. With the landslide point theme selected
in the map view select Analysis/Summarize Zones. Select the 'Lsrecno'
field as defining zones. Select 'Slope' as the theme containing variable
to summarize. Select cancel on statistic to chart. The result
is a table that contains the statistics within each landslide zone.
Since each zone is a point min, max, mean and sum are the same and range
and std (deviation) are 0. Start editing this table and delete all
fields except lsrecno and mean. In table properties enter the name
'Slope' as an alias for 'mean'. Stop editing, saving the edits.
Use the Lsrecno field to join this table with the 'Attributes of Landslides'
table. Now slope for each landslide is in the Landslides attribute
table. [Note that slope as used here is vertical drop over horizontal
distance, i.e. the tangent of the slope angle.] Select the slope
field then select Field/Sort descending. This sorts the table so
that landslides with the steepest slope are at the top. Read down
n lines, where n is the number of landslides in the combined upper threshold
and defended zones with default parameters. The slope averaged between
this line and the line below gives a slope threshold that will capture
the same number of landslides. Use map calculator to map the area
with slope greater than this threshold.
Questions to hand in:
-
In the sample area, how many landslides occur in combined upper threshold
and defended classes for the default parameters?
-
What is the slope threshold that captures this number of landslides using
a slope only approach to classify terrain stability?
-
What is the area and landslide density (in landslides per km2)
in this unstable class mapped using the slope only approach? Compare
this to your answer to 4 above.
-
Give a layout showing the stability index map, SA plot, slope threshold
stability class map, and table of landslide densities in each stability
class.
Do the same for your optimum parameters.
Questions to hand in:
-
In the sample area, how many landslides occur in combined upper theshold
and defended classes for your optimum parameters?
-
What is the slope threshold that captures this number of landslides using
a slope only approach to classify terrain stability?
-
What is the area and landslide density (in landslides per km2)
in this unstable class mapped using the slope only approach? Compare
this to the landslide density obtained using your optimum parameters.
-
Give a layout showing the stability index map, SA plot, slope threshold
map, and table of landslide densities in each stability category.
This exercise will have you download DEM data from a USGS web site.
The DEM quadrangles needed are:
PORCUPINE RESERVOIR,UT
HARDWARE RANCH,UT
SHARP MOUNTAIN,UT
MONTE CRISTO PEAK,UT
DAIRY RIDGE,UT
Go to http://edcwww.cr.usgs.gov/doc/edchome/ndcdb/ndcdb.html.
This is the site for FTP access to SDTS format DEM data. Look for
the 7.5-Minute Digital Elevation Model (DEM) section link and select a
method to find the quadrangles you want. I find ftp via graphics
easiest. Locate Utah on the US map, then Logan at the North end of
Utah. Click just SE of Logan and you should see these quad names.
Click on a quad name on the map, then in the page and download a file that
has name ending in tar.gz. Extract all files (using Winzip) from
this file into a separate folder named so you know which is which.
One of the files extracted has name ending in cel0.ddf. This is the
largest file. In ArcView make sure you have the SDTS grid import
extension (see installation above) loaded. With a View open select
File/Import data source. Designate the type as USGS SDTS DEM.
In the file dialog select the file with name ending cel0.ddf corresponding
to the quad you want to import. Add the themes to the view.
You will need to do this once for each quadrangle (Sorry for the tedium
- but this is the best we can do so far). Once all 5 quads are imported
and displayed as themes, select all grid themes and do Edit/Merge grids.
[This is a feature added by SINMAP]. You should end up with a merged
grid named Mosaic of .... This grid most likely still has some
problems where cells are left out at the joins. Download the program
gapfill.exe.
This is a C program that fills the gaps by averaging from the grid cells
on either side. Run this program from a MS-DOS prompt, giving the
name of your mosaiced grid as a command line argument. [Theme/properties
tells you the name of the grid file associated with the grid theme in a
View]. You will be prompted for the name of an output grid.
(I used 'lb' - for Little Bear). Add the new grid 'lb' to your View
as a Theme. This is the grid you need to work with and you can delete
all others. Data importing is now complete.
Do SINMAP/Select DEM grid for analysis and select this grid. Then
do SINMAP/Compute All Steps from the Grid Processing section of the SINMAP
menu. We are not going to do stability analysis for this area.
SINMAP just provides an effective way to get the pits filled and slope
and contributing area needed for TOPMODEL analysis.
Now to identify and delineate a watershed we will use TARDEM.
From the MS-DOS prompt run:
d8 lb
This computes single D8 flow directions over the grid. These
are saved in a grid file with suffix 'p', i.e. 'lbp'. Then run
aread8 lb
[Ignore the VAT warning message - we do not need a VAT]. This
computes contributing area using the D8 method. The result is saved
in a grid file with suffix 'ad8', i.e. 'lbad8'. Add the grid lbad8
to the view and fiddle with the color scheme legend so that you can see
the channel network. Select the lbad8 theme and zoom in to the region
around coordinates x=438000 y= 4596500. Determine the precise coordinates
of the grid cell that has a contributing area of 183727 grid cells.
This will be our outlet.
Question to hand in:
-
What are the coordinates of the outlet cell in the Little Bear area near
x=438000 y= 4596500 with contributing area 183727 of grid cells.
Now run:
aread8 lb [xvalue yvalue]
where xvalue and yvalue are the coordinates determined above.
This calculates the contributing area only for the portion of the watershed
draining to this cell. The map calculator [1.AsGrid/( [Lbad8] > 0)]
can be used to build a mask of grid cells in the watershed, i.e. that have
contributing area greater than 0. I use the 1/ logic so that the
grid has No Data, rather than 0's outside the watershed. [1/0 ArcView reports
as No Data]. Save this calculation grid as 'lbmask'
Channel network delineation. From the MS-DOS prompt run the TARDEM
command
netsetup lb -m 4 0.4 0.1 0.05 20 -xy [xvalue yvalue] where xvalue
and yvalue are coordinates of the outlet.
arclinks lb -i
The first command delineates channel networks using only upwards curved
grid cells. The second command imports the resulting files into ArcView.
Name the Output data source 'lbliuc' when prompted. [The uc is for
upwards curvature.] When the import is complete, add feature data
source theme 'lbliuc'. This is the delineated channel network.
The table associated with this theme contains data such as the length of
each link, slope, contributing area, stream order (using Strahler's method),
link magnitude (the number of sources upstream of each link), etc.
The first link in the table is the outlet link. The order and magnitude
attributes of the outlet link are also referred to as the order and magnitude
of the entire network that drains to that link. Field/Statistics
may be used with this table to determine the total channel length.
An alternative scheme for channel network delineation evaluates stream
order for a network starting at each pixel then prunes the network by applying
a threshold to that order. This may be obtained from the following
TARDEM commands run from the MS-DOS prompt.
gridnet lb
netsetup lb -m 5 4 -xy [xvalue yvalue]
arclinks lb -i
A constant support area threshold may be used to delineate channel networks
using the following TARDEM commands run from the MS-DOS prompt.
netsetup lb -m 1 [threshold] -xy [xvalue yvalue]
arclinks lb -i
Experiment with a few different thresholds, e.g. 50, 200, 1000 (whatever
the thousand-million rule suggests). Plot 50 m contours for this
area and display the networks you extract together with the contours.
Evaluate the network you extract (subjectively) against the contour crenulations
and where you expect water to flow in a channelized manner rather than
overland. You may also try some of the other netsetup methods described
in the TARDEM
documentation.
To hand in:
-
A layout showing the upwards curved grid network, and whatever other method
you thinks performs best for delineating channel networks for this watershed.
-
A table giving the following network information for three networks:
Number of links in the network, the stream order, the link magnitude,
the total length of channels, the watershed area and the drainage density.
The three networks tabulated are to be the upwards curved network delineated
first, the pruned grid order network and then whatever network you think
is best for this topography.