IDRISI to analyze Landsat 5 data of
Creek National Wildlife Refuge
Band 4 of Squaw
Creek showing features in and around the refuge
ES 775 Advanced Image Processing
processing of remotely sensed data is a useful means of obtaining information
about the Earth’s surface. The purpose of this project is to demonstrate
the capabilities of IDRISI and
to discuss how the information obtained from it can be used to further
the understanding of an area of interest. In this case, the area
of focus is the Squaw Creek
National Wildlife Refuge in Mound City, Missouri.
Creek National Wildlife Refuge is a protected area within the Missouri
River Valley. The habitat is made up of some of the last remaining
of its kind and consists of agricultural fields, bottomland forest, loess
bluff hills, wetlands, and wet prairie grasslands. It sustains a
variety of wildlife including threatened and state endangered species such
as the eastern massasauga rattlesnake that resides in the wet prairie.
The wildlife here depends on the refuge management to maintain or manipulate
the habitat as necessary. The use of satellite imagery is one way
in which more information can be gained about the land and then used in
order to improve management practices.
an image from the Landsat 5 satellite from July 02, 1997 is being used.
This satellite carries a Thematic Mapper (TM) sensor system, an optical-mechanic
sensor that operates in 7 bands of visible, reflective-infrared, middle-infrared,
and thermal infrared regions of the electromagnetic spectrum. Each
of the spectral bands represents a specific wavelength and demonstrates
separate characteristics. The following chart shows each of the bands,
their wavelength, and the applications for each of them (Jenson, 2000).
increased penetration of water bodies, supports analyses of landuse, soil,
and vegetation characteristics
to green reflectance of healthy vegetation
for vegetation discrimination, soil boundary, and geological-boundary delineations
to the amount of vegetation biomass present, useful for crop ID and emphasizes
soil/crop and land/water contrasts
to the turgidity of water in plants, useful in crop drought studies/ plant
vigor investigations, and discriminates between clouds, snow, and ice
amount of infrared radiant flux emitted from surfaces, useful for locating
geothermal activity, vegetation classification, vegetation stress analysis
and soil moisture studies
geologic rock formations
satellite imagery of Squaw Creek
Landsat satellite image can be obtained from the U. S. Geological Survey
(USGS) in a raster format that will work in IDRISI. Therefore, no
conversions are necessary. The images are organized by path and rows
across the Earth’s surface. A window including only the refuge boundaries
can be extracted from the rest of the image.
Squaw Creek Refuge lies within Path 27 and Row 32 in northwest Missouri.
First, a composite that will show distinct features can be made in order
to determine what portion of the image needs to be extracted. A 345
composite of the refuge works well because roads, pools, fields, and cities
all can easily be distinguished. Then, zoom into the refuge and determine
the four corners of the area that needs to be extracted. The four
corners determined here are:
|Upper left column
|Upper left row
|Lower right column
|Lower right row
extract the window, go to Reformat--> Window. Use the corners
determined above and give file names to the new, extracted images.
The new images now isolate the refuge and can be analyzed in a variety
of ways using IDRISI.
A 345 composite
of Squaw Creek National Wildlife Refuge
extracted from a
Landsat 5 satellite image.
images are made when selecting three different bands of the Landsat image.
Each composite image emphasizes different features in the image.
Some will bring out features of vegetation, while others bring out man-made
features such as roads or cities, and others may highlight water bodies.
Below are several composites with a description of what is shown in the
The standard false color composite is made up of bands 2, 3, and 4.
Vegetation is shown in shades of pink and red. Darker shades represent
thicker, more active vegetation while the lighter shades are generally
or and less dense. This image shows that much of the refuge is covered
in vegetation, including most wetland units. It was taken in July
when the pools/wetland units have dried up or drained for the summer.
Notice the darker shade of red that represents the forest compared to the
lighter shades of the prairie or wetland units. The agricultural
fields are primarily bare and are shown in shades of blue.
image is a natural color composite made up of bands 1, 2, and 7.
It shows objects in their natural color and can be helpful in its interpretation.Vegetation
is shown in shades of green while fields are in browns. Roads and
cities are shown in white or blue.
false-color composite image is made up using bands 3, 4, and 5. This
image seems to best represent areason the refuge that contain water.
Water bodies are shown in dark blues. Due to the time of year, the
areas of water are not extensive, but small patches of water can be seen
in some parts on the refuge.
cities, and bare fields are represented best in this 124 false-color composite
image. In this image these areas include Mound City, highways, and
bare fields which are located in the southwest corner.
image can be created to analyze the vegetation on the refuge. NDVI
stands for the Normalized Difference Vegetation Index. This is an
index derived from reflectance measurements in the red and infrared portions
of the electromagnetic spectrum. It describes the relative amount
of green biomass from one area to the next. To create an NDVI image,
bands 3 (red) and 4 (near-infrared) are used. First go to the
Overlay module to enter in band 4 as the first image and band 3 as
the second image. Then select the overlay option: First – Second
/ First + Second and use “raw ndvi” as the value units. The NDVI
values range from –0.76 to 0.82 where vegetation is shown in green while
other features are shown in yellow, red, and blue (Figure 1).
NDVI values are fractional real numbers ranging from –1.0 which means no
vegetation, to +1.0 which refers to the maximum vegetation. These
raw NDVI values are changed to the byte-binary data type by using Scalar
add 1 (Figure 2). Scalar is then used to multiply the second
image (Figure 2) by 125. This third image (Figure 3) is then used
to Convert the image from real-binary to byte-binary with rounding.
The final NDVI image (Figure 4) can be viewed with the NDVI palette and
then used for vegetation analysis.
The final NDVI image shows neutral
values (yellows) representing bare ground, lower values (browns) represent
bodies of water, and higher values (green) are active vegetation.
This image makes it very clear the amount of vegetation covering the refuge
and the lack of water during the summer months.
analysis can be done with the TASSCAP (tasselled cap) module.
TASSCAP can be found under Program --> Analysis --> Image Processing.
When using a Landsat TM image, 3 images will be derived. They include
a greenness image, a brightness image, and a moistness image. The
brightness image refers to soil brightness, the greenness (or Green Vegetation
Index – GVI) refers to green vegetation cover or vegetation biomass above
ground, and the moistness image refers to soil moisture.
One image that
shows interesting results is the moistness image. The wet prairie
definitely stands out apart from the surrounding areas of wetlands and
forest habitat. To further look at wet prairie habitat, the ISOCLUST
module can be used. It can be found under Analysis --> Image Processing
--> Hard Classifiers. Using ISOCLUST will separate different
features by their spectral reflectance.
ISOCLUST image combines like spectral signatures together into 16 clusters.
So, similar objects and features are represented by the same color.
In this image the wet prairie is represented in 2 different colors, cluster
4 and cluster 7. The wet prairie habitat can be further isolated
by creating a Boolean Image.
Boolean Image of
the Wet Prairie
A Boolean image
is an image in which the attribute of any cell is the integer 1 or 0.
EDIT to make an attribute file. Enter in the clusters
4 and 7 to assign them a value of 1 as shown below. All other features
will then be assigned to 0 and will appear blacked out in the image.
Save this as an attribute file to be used in the next step.
Now run ASSIGN
under Analysis --> Database Query. Choose the ISOCLUST image
as the feature definition image and choose the attribute file that was
just created. The image now isolates the wet prairie habitat.
majority of the wet prairie habitat can be seen in the center of the refuge.
A benefit of creating a Boolean image that isolates this habitat is to
see how much of the same type of habitat can be found surrounding the refuge
This could be useful
information when applied to studying the state endangered eastern massasauga
rattlesnake discussed before. This species primarily uses the wet
prairie habitat. By finding surrounding patches of wet prairie habitat,
another population of the snakes could be found living off the refuge boundaries.
There are many
possibilities to explore satellite imagery using IDRISI software.
From creating composites that best depict the land features to manipulating
the imagery to highlight specific habitat, much information can be gained
about the Earth. In the case of the Squaw Creek Refuge, analyzing
satellite imagery can greatly aid in management techniques and simply gaining
more information about the habitat.