Introduction to
Geographic Information Systems in Forest Resources |
Exercise:
Remote Sensing
- Open ArcView 3.x and enable the Image Analyst and IMAGINE Image
Support extensions
- Add some image analysis themes
- Explore image properties (cell values)
- Alter resampling for image display
- Alter the image legend
- Alter band combinations
- Create an NDVI grid theme
- Find similar areas
- Create an NDVI grid theme (manual method)
- Convert an image theme to a series of grids
- Perform Tasseled Cap analysis
Open ArcView 3.x and enable the Image Analysis and IMAGINE Image Support extensions
- Open theStartup Project from the CD.
- Set the working directory to your removable disk.
- From the File menu, select Extensions.
- Enable the Image Analysis, IMAGINE Image Support, and Spatial
Analyst extensions.
Image Analysis includes particular image analysis routines that are not included
in the core ArcView 3.x application.
IMAGINE Image Support allows you to add Erdas image formats img, gis,
and lan. These images cannot be added to a view unless the IMAGINE
Image Support extension is enabled.
Add some image analysis themes
- Open a new view. Change the view's name to View 1991. This view will contain
data from 1991.
- Add the Image Analysis Data Source landsat_1991.img. Note
this is a new Data Source Type you have not encountered before.
- Add the Stands data source to the view, and display it as unfilled
polygons with a bright green outline. Zoom to the extent of the Stands
theme. Note how the pattern of the stands is similar in the image and the
vector data.
The image theme is displayed in false color where band 4 is displayed
in red, band 3 is displayed in green, and band 2 is displayed in blue. The
colors you see are the results of mixing of different values of red, green,
and blue. Note that the image legend displays the current pixel resolution
as well as the band assignments.
Bare ground appears blue-grey, young vegetation appears bright red to pink,
mature vegetation appears dark red, and water appears nearly black. Actively
photosynthesizing vegetation has a relatively high reflectance of NIR radiation,
which is why the stands with vigorous vegetation appear to have a deeper red
color.
- Create views for 1998 and 2000 as well, and add the landsat_1998.img
and landsat_2000.img data sources to these respective views.
- Copy and paste the Boundary theme to the other two
views (this will save you the time of altering the legend more than once).
- Look at the different images from different years and see how some of the
areas change from bare ground to young vegetation.
You have just added a few themes, including an image analysis theme. Image
analysis themes have special properties, even though they originate from any
type of supported image data source.
Alter resampling for image display
- Zoom into the image in View 1991. As you zoom in, you will not see
pixellation, but an interpolated transition between pixels. Note there is
also a resolution statement in the legend.
This behavior is one of the special properties of an image analysis theme.
- Make the image theme active and then click the Theme properties button
.
Check the Hide Image Resolution box. This will hide the pixel resolution
on the legend for the theme.
- Click the Information icon to view image properties.
This shows data similar to what you have seen with grid theme properties.
A few differences are the display of the number of bands (an image can have
several bands, whereas a grid has only one band), and Pyramid Layers presence/absence.
Pyramid layers are additional data subsets created for images at resolutions
that allow for more speedy display when zoomed out at small scales.
- Click OK to dismiss the Theme properties dialog.
- In order to see the individual cells, you will need to turn off image interpolation.
From the Image Analysis properties menu, select Preferences.
Select Nearest Neighbor from the dropdown. Click OK to dismiss
the preferences.
- Now you will see the individual pixels. If you don't see the pixels, zoom
in or out until you do.
You have just viewed and altered theme properties for an image analysis theme.
Explore image properties (pixel values)
- Use the Identify tool and click on one of the pixels. You will see
the Identify results dialog.
The dialog shows the values for each of the 7 LandSat bands. The values are
expressed on a scale of 0-255 (known as 8-bit values; 8^2 = 256).
These values represent the relative strength of reflectance in each "slice"
of the spectrum measured by the LandSat satellite sensors.
- Click several different cells to see their reflectance values.
The dialog also has entries for Latitude and Longitude values. Because the
header for this image is incomplete, the latitude and longitude are null.
- Use the identify tool on one of the newer images in another view and you
will see the lat/long values displayed
Like any of the other themes we have used, image analysis themes can be identified
on a pixel-by-pixel basis.
Alter the image legend
- Open the legend for the image theme.
You see that the legend editor is quite a bit different from the legend editor
for feature or grid themes. It contains more controls. Hit the <F1>
key to view the help topic that describes the image theme legend.
- For the Assign bands for color: control, click the Natural
button. Note how the band assignments change, as well as the image display.
This combination, also known as True color, shows the bands that closely
represent red, green, and blue wavelengths as red, green, and blue, respectively.
It looks similar to a color photograph.
Compare the true color image above with the false-color image below.
- Alter the Brightness and Contrast to see how this alters the
display. Make sure to click the Apply button to apply the changes.
- Click the Statistics button to view value statistics for various
bands. Use the Layer dropdown to select the band for description. This
will give you basic descriptive statistics about the values in each band.
- Turn off the Stands theme and experiment with the different Stretch
options to see how different patches appear in different colors. Different
band combinations and different stretch values can allow you to detect differences
among various patches.
Here are a few different stretch choices applied to the 1991 image. You should
review the different stretch types listed in ArcView 3.x 's help.
|
standard deviation
|
Gaussian
|
gamma
|
- Click the Advanced button to view the histograms for the different
bands. Depending on the stretch you have applied, your histograms may differ
from these:
The histograms show input and output values for each band. The black histograms
show the actual count of the number of pixels (Y-axis) at each reflectance
value level (X-axis). The colored histograms show the values as they are displayed
The yellow line controls the input-display mapping. The input values are on
the X-axis. The display values are on the Y axis (though the units are not
displayed). In the image above, a red input value of about 180 (on X) translates
to a value of 255 (on Y, though unmarked). Altering the yellow line alters
the input-display values for the intensity of each color. This can help display
different patches distinctly. The different stretch values are really just
shortcuts for different input-display pixel color intensity maps.
- Alter the Legend Type dropdown from Mutli-band to Single-band.
Display Layer_5. This is the reflectance in the 1.55 - 1.75 micron
range.
Band 5 is an infrared band. Green vegetation absorbs infrared light, so green
vegetation will have a low reflectance value. Make the Stands theme
active and identify some of the darker stands. You should see they are older
than the stands that appear very bright in shade. Also be aware that the image
is from 1991, whereas the stand boundaries are from 2000-01.
- Experiment with a few different single-band displays and see if you can
identify patterns or trends.
You have just altered the legend for an image analysis theme. Note that there
are different options for altering these legends than for vector or grid themes.
Alter band combinations
- With the legend still open, switch back to Multi-band display.
- Alter the band combinations so that bands 7, 4, and 3 are shown in red,
green, and blue, respectively.
This combination (Short-wave infrared composite) shows stressed vegetation
in bright pink/purple.
- Try using different band combinations to see what features appear strongly.
You have just altered the band combinations for the image analysis theme. Textbooks
on remote sensing will describe how using different band combinations highlights
different features.
Create a normalized difference vegetation index (NDVI) grid theme
The Image Analyst has a menu choice for creating NDVI grids.
- Make View 2000 the active document. Make the Landsat_2000.img theme
active.
- From the Image Analysis menu, select Vegetative Index.
- Accept the default band assignments of 4 and 3 for NIR and Visible IR bands.
After a moment the processing will complete and the new image theme will be
added to the view.
The darker the pixel, the lower the NDVI value. The brighter the pixel, the
higher the NDVI value. Higher NDVI values correlate with more vigorously growing
vegetation. How does this compare with the forest stands?
- Identify the stands in some of the characteristically dark and light areas
on the NDVI image.
- For use in further raster analysis, convert the image to a grid by selecting
Theme > Save Image As
Save the grid on your removable drive as ndvi_2000. We won't use the grid
during this exercise, but you may want to make a mental note of this for later
(hint).
- Note that when the grid is added to the view, it is added as an Image Analysis
theme. This means that grids can also be manipulated in the same way as other
images, with the exception of multi-band analysis and manipulation.
You have just created a grid that represents the NDVI values for each pixel.
Find similar areas
- Make View 1998 the active document. Zoom into some heterogeneous
areas near the center of the forest. Alter the outline color of stands to
gray (because the seed polygons are outlined in green).
- Alter the Seed Tool Properties (from the Image Analysis menu)
so that the Seed Radius is 2 pixels. Also make sure Include
Island Polygons is checked.
- Make the image theme active, and then using the Seed tool
,
click on some of the old forest in the Hugo Peak Transect stand. You
will see a patch of pixels outlined in green. These are pixels which are statistically
similar in the displayed bands (note: there is no description in ArcView 3.x help
on the actual statistical ranges used to perform pixel selection).
- To find all areas with similar values, from the Image Analysis menu,
select Find Like Areas.
- In the Find Like Areas dialog, click the New button to make
a new output image to store your classes.
- Uncheck and then re-check the Select Graphics check box. For this
class, enter the Class Name of Mature Forest.
After a few moments ArcView 3.x will select all pixels across the entire image
that are similar.
- Hit the <DELETE> key to remove the seed selection graphic.
- Click on the bright pink pixels outside of what has now been identified
as mature forest.
- Once again, search for like areas. This time call the class New Growth.
- Zoom to the extent of the Stands theme. Do you think this new image
correctly captures the variation in the original LandSat image? If not, try
using different seed areas, or try using different band combinations for creating
the seed areas.
You have just selected a group of pixels that has a similar range of values
in the displayed bands. Altering the band combinations will create different
selections.
Convert an image theme to a series of grids
- Create a new view called Pack Forest RS.
- Add another copy of the landsat_2000.img data set to the view. This
time make sure to add it as a simple Image theme.
- Make the Landsat_2000 theme active.
- From the Theme menu, select Convert to Grid.
- When asked, convert each band to a new grid.
- Convert the first band to a grid called Lan1.
- Continue converting, naming each new grid with the same naming convention
(Lan2, Lan3, etc.).
- When asked, add each grid theme to the view.
- View each band in turn.
You have just created a new grid from each band of this 7-band LandSat TM image.
Each grid has cell values that represent reflectance intensity in specific wavelength
ranges. These individual grids can be used with map calculations to perform
multispectral analysis.
Create an NDVI grid (manual method)
- Invoke the Map Calculator (from the Analysis menu).
- Format a calculation with the NDVI function. Feel free to cut-and-paste
from the browser:
([Lan4] - [Lan3]).Float / ([Lan4] + [Lan3]).Float
This performs the NDVI calculation manually, and also forces the output to
be stored in floating-point value, to maintain the decimal places for the
output grid. If you do not use the .Float request, your values will
not be calculated correctly.
- When the calculation completes,
- Classify the new grid into 30 equal-interval classes.
- Use the Browns to Blue-greens dichromatic Color Ramp.
- Sort in the Value field in Descending order. Vigorous vegetation
will appear in the darkest browns.
You have just performed the calculation of NDVI on a series of grids representing
different bands from a LandSat image. This results in a single grid whose value
represents the NDVI value for each pixel. While it is easier with the GUI tool,
this method shows how the calculation works on a more programmatic level.
Perform Tasseled Cap analysis
Greenness
- Open the Map Calculator and enter the Tasseled Cap greenness function:
( [Lan1] * -0.2848 + [Lan2] * -0.2435 + [Lan3] * -0.5436
+ [Lan4] * 0.7243 + [Lan5] * 0.0840 + [Lan6] * -0.1800)
|
Note: the 2000 scene has been corrected
for atmospheric effects. This removes the 6th band from the original
7 bands. For this reason, the 6th band in this image is used in the
Tasseled Cap analyses, rather than the 7th, as described in the original
article (Crist, E., P., Cicone, R., C. (1984). A Physically-Based
Transformation of Thematic Mapper Data - the TM Tasseled Cap. IEEE Transactions
on Geoscience and Remote Sensing, GE-22, 256-263. ) If you are using
a LandSat image with 7 bands, use band 7 rather than band 6 in the calculation,
e.g.,
( [Lan1] * -0.2848 + [Lan2] * -0.2435 + [Lan3] * -0.5436 + [Lan4] *
0.7243 + [Lan5] * 0.0840 + [Lan7] * -0.1800) |
I suggest using cut-and-paste from the browser to your Map Calculation
dialog. You can select Edit > Copy from the browsers's menu. Then
place the cursor in the Map Calculation expression area and use the <CTRL-V>
keystroke combination to paste.
Note: for the 1998 and 2000 LandSat images, use band 6 rather than band 7
in the Tasseled Cap analyses.
- When the new grid is added to the view, alter the legend to use the Chartreuse
monochromatic Color Ramp.
Now vegetation that is "greener" will actually appear a darker green
in the view. What do you think "greenness" means? How would you
support your idea based on other data sources? You may also want to load the
Boundary theme for reference.
You have just created a single grid that represents Tasseled Cap greenness
of the original LandSat image. The greenness value is theoretically proportional
to the amount of green vegetation.
Wetness
- Using the Map Calculator, create a calculation for the Tasseled Cap wetness
index:
([Lan1] * 0.1509 + [Lan2] * 0.1973 + [Lan3] * 0.3279
+ [Lan4] * 0.3406 + [Lan5] * -0.7112 + [Lan6] * -0.4572)
|
Note: the 2000 scene has been corrected
for atmospheric effects. This removes the 6th band from the original
7 bands. For this reason, the 6th band in this image is used in the
Tasseled Cap analyses, rather than the 7th, as described in the original
article (Crist, E., P., Cicone, R., C. (1984). A Physically-Based
Transformation of Thematic Mapper Data - the TM Tasseled Cap. IEEE Transactions
on Geoscience and Remote Sensing, GE-22, 256-263. ) If you are using
a LandSat image with 7 bands, use band 7 rather than band 6 in the calculation,
e.g.,
([Lan1] * 0.1509 + [Lan2] * 0.1973 + [Lan3] * 0.3279 + [Lan4]
* 0.3406 + [Lan5] * -0.7112 + [Lan7] * -0.4572) |
Note: for the 1998 and 2000 LandSat images, use band 6 rather than band 7
in the Tasseled Cap analyses.
- When the new grid is added, change its legend to use the Cyan Monochromatic
Color Ramp.
Now pixels that with vegetation that is more "wet" are shaded a
deeper cyan. What do you think "wetness" means? How would you support
your idea based on other data sources?
You have just created a single grid that represents Tasseled Cap wetness of
the original LandSat image. The wetness value is theoretically proportional
to the amount of moisture in the soil or vegetation.
Syllabus