Q2. Describe the new IHS
image that was just created from the standpoint of the color characteristics of
the image itself and the original RGB image that was just transformed. In the
description, state the differences in the patterns of the histograms for the
three bands used in the transformation.
Once the RGB image was transformed to IHS,
the IHS image had way too much contrast. As shown in figure 1(left) is very
hard to distinguish any features on the IHS except for the Mississippi and
Chippewa Rivers. Also shown in figure 1, the color in the enhanced image
(ec_rgb_ihs.img; right) is much less realistic than that of the original RGB
image prior to its transformation.
Figure 2 compares the 3 histograms of the
original eau_claire_2001.img image (top row) to that of the three histograms of
the IHS image (bottom row). The histograms of the IHS image show that the
frequency of the radiometric data is more widely distributed across the
histograms than those of the RGB image. This is to be expected, due to the high
contrast exhibited by the IHS image in figure 1.
Figure 2
Figure 2 shows the histogram data of
both the original RGB image (top row of three) and those of the image after the
RGB to IHS conversion (bottom row of three).
Section 2: IHS to RGB Transformation:
Step 1: Open
the IHS image that was created in section 1 in the Erdas viewer. In this case
it is ec_rgb_ihs.img.
Step 2:
Activate the Raster Processing Tools
on the Erdas interface.
Step 3:
Click on Spectral, followed by IHS to RGB. This will open the RGB to
IHS interface.
a) Ensure that the input file on the IHS
to RGB interface is eau_claire_2000.img.
b) Click on the folder icon next to the Output File section and navigate to the
appropriate output folder. Enter the name of the output file. In this case it
is ec_ihs_rgb.img.
c) Make sure that Intensity=band 1, Hue=band 2,
and Saturation=band 3 on the IHS to RGB interface.
d) Leave all default parameters as they
are on the interface and click OK to
run the model.
Step 4: When
the model has finished running click Dismiss and then close out the dialogue
box.
Q3. Compare and contrast
the newly transformed RGB image and the original RGB image in terms of color
characteristics and histograms.
Figure 3
shows the original RGB image and the image that resulted from the IHS to RGB
transformation (no stretch) that was described in section 2 of this lab. Both
images have the color guns set as follows: red=3, green=2, and blue=1. As far as differences in color, the enhanced
image has more of a brown tint to it that the original RGB image. Also, the
transformed image still does not represent what one would expect to see in the
real world.
However, the
histograms found in figure 4 shows that band 3 (red) is more widely distributed
in the histogram of the original image (top row) than in the one after the IHS
to RGB transformation (bottom row). In contrast, the histogram that represents
band 1 (blue) shows better distribution in the output image, and thus higher
contrast in this band for this image.
Also, it worth noting that the histogram for
band 2 (green) in both the original and enhanced image has not changed at all.
This was confirmed by looking at the statistics for this image which showed no
difference between the two images in this band, except that the median was off
by about +0.333 in the enhanced image.
Figure 3
Figure 3 shows the original RGB image
(left) and the image that resulted from the RGB to IHS transformation (no
stretch; right) displayed in the Erdas viewer.
Figure 4
Figure 4 shows the histogram data of
both the original RGB image (top row of three) and those of the image after the
IHS to RGB conversion (no stretch; bottom row of three).
Section 3: IHS to RGB Transformation
with I & S Stretch:
Repeat the
steps in section 2 to perform an IHS to RGB transformation. However, this time
apply Stretch I & S in the IHS
to RGB interface. Name the new file ec_ihs_rgb2.img
and set color guns to 3, 2, and 1 for R, G, and B, respectively.
Q4. Compare the newly
stretched retransformed RGB image to both the non-stretched and original RGB
images in terms of color patterns, quality and histograms (Make color gun 3, 2,
1).
Compared to
the original RGB image, the IHS_RGB2 image, displayed in figure 5, is not much
different than the first IHS_RGB image with no stretch in terms of color
patterns. Likewise, the image itself resembles the first un-stretched IHS_RGB
image as far as color and quality is concerned. That is, it bears little
resemblance to what one would expect in the real world.
Figure 6
illustrates the histograms of each band of the enhanced and stretched IHS_RBG2
image. The distribution within the histograms compared to that of the original
RGB image is similar to the differences exhibited between the unstretched
IHS_RGB image and the original. However, in the stretched image the histograms
show that distribution of data within each layer is more widely distributed
than in the unstretched image. Of course the exception to this is the green
band (layer 2) which has not changed significantly with the application of any
enhancement.
Figure 5
Figure 5 shows the IHS to RGB
transformation (ec_IHS_RGB2.img) in the Erdas Imagine viewer. A stretch to the
saturation and intensity was applied.
Figure 6
Figure 6 shows the histogram data for
the enhanced IHS to RGB image (ec_IHS_RGB2.img) after a stretch to the intensity
and saturation was applied.
Part 2: Image Mosaicking:
Mosaicking
is useful in remote sensing when the AOI is extremely large, or it transverses
two satellite scenes.
Step 1: Open
Erdas Imagine and navigate to the desired image, in this case it is eau_claire_2005p26r29 in the Select Layers to Add window. Do not add the image yet.
a) Click on Multiple in the Select
Layers to Add window and then on the Multiple
Images in Virtual Mosaic button.
b) Click on Raster Options and ensure that Background
Transparent is checked. Also, check Fit
to Frame.
Repeat step
1, but this time bring in the image eau_claire2005p25r29.
Click OK when finished. This will load the image presented in figure 7 into the
Erdas Imagine viewer.
Figure 7
Figure 7 illustrates how the image
that resulted by performing step 1 will appear in Erdas Imagine.
Section 1: Image mosaic with the use
of mosaic Express:
Step1: With
the image illustrated in figure 7 in the Erdas Imagine viewer, activate the Raster Tools.
Step 2:
Select Mosaic followed by Mosaic Express from the resulting drop
menu. Doing this will result in the appearance of a Mosaic Express window.
a) Under the Input tab, click on the folder icon.
b) Next, select the file eau_claire2005p25r29. This will be the
top layer.
c) Repeat part c, but load the image eau_claire2005p26r29.
d) Click on Next. Continue to do so, leaving all parameters at their defaults,
until the Output Dialogue is
reached.
e) Click on the folder icon next to Root Name and navigate to the
appropriate output folder.
f) Name the output file eau_claire2005msx.img.
g) Leave all parameter at their defaults
and click Finish to run the model.
Step 3: When
the model finishes running, click on Dismiss and close out the window.
Q5. Describe the nature of
color in your output image, in other words, is there a smooth color transition
between one image and the other especially at the boundaries?
As shown in
figure 8, the colors in the two images that were merged via Mosaic Express do
not match each other. The result is a distinct boundary in the images.
Figure 8
Figure 8 shows the mosaic image that was created
by running Mosaic Express in Erdas Imagine as it appears in the viewer.
Section 2: Image Mosaic using
MosaicPro:
Step 1: Open
the same two images in Erdas Image as in section 1; follow the same steps when
opening them (eau_claire1995p25r29.img
and eau_claire1995p26r29.img).
Step 2: Click
on Mosaic in the Raster tools and
select MosaicPro. This will open the
MosaicPro window.
a) Click on the Add Images icon
to open the Add Images dialogue box.
b) Highlight eau_claire1995p25r29.img, but DO
NOT add it.
c) Click on the Image Area Options tab in the Add
Images dialogue box and then select Compute
Active Area. Click OK.
d) Repeat steps a, b, and c to add eau_claire1995p26r29.img.
e) Make sure that the eau_claire1995p25r29.img is the bottom
image in the MosaicPro Window.
f) Click on Color Corrections
in the MosaicPro tool bar. This will open the
Color Corrections dialogue box.
g) Check the Use Histogram Matching option. This will activate a Set button;
click on Set and then select Overlap
Areas and click OK.
h) Click on Process in the MosaicPro interface followed by Run Mosaic.
i)
Navigate
to the appropriate output folder and name the new image eau_claire1995msp.img. Click OK to run the model.
j)
Once
MosaicPro is complete, click Dismiss
and then close the window.
Q6. Compare the output
mosaic image using the MosaicPro (MP) and that obtained earlier using the
Mosaic Express (ME). In your discussion, state the reason(s) for the
differences in the image quality.
Figure 9 shows a comparison of two mosaicked
images as viewed in the Erdas 2013 viewer. The image on the left in figure 9
shows the image in which MosaicExpress was applied to join the two images
(eau_claire199msx.imge). The right-hand side of figure 9 illustrates an image
that was processed using MosaicPro (eau_claire1995msp.img). From he
Looking at figure 9, it is clear that MP
creates a much more seamless image than does the ME image. This is because the
MP image process is more precise. For instance the “color corrections” option
in MP allows the radiometric data in the two input images to be synchronized. In
this case “histogram matching” was used, which balanced the color differences
in the two images.
However,
even though MP created an image that was superior in quality to the ME image,
the boundary-line of the two images can be seen in the MP image. This Fact is
true especially when closely examining the image (i.e. zooming in).
Figure 9
Figure 9 compares the images produced
using MosaicExpress (left) and MosaicPro (right).
Part 3: Band Ratioing:
This section of the lab will use a ratio
transformation to create a normalized difference vegetation index (NDVI) of the
Eau Claire area. This type of index can help an analyst distinguish vegetation
from other surface features.
Step 1:
Bring the desired image into the Erdas 2013 viewer; in this case it is eau_claire_2011.img.
Step 2:
Select Unsupervised, followed by NDVI in the Raster tools menu of the
Erdas viewer. Doing this will open an indices interface.
a) Make sure that the input image is eau_claire_2011.img.
b) Click on the folder icon next to the Output File portion of the Indices
interface and navigate to the appropriate output location.
c) Name the new image eau_claire2011ndvi.img.
d) Make sure the Sensor section of the indices interface reads ‘Landsat TM.’ Landsat TM 4 and 5 should have the same bands.
e) Under Function in the indices interface, highlight NDVI.
f) Click OK to run the model.
g) Dismiss at the
completion of the run and then close out the window.
Q7. What will you expect to find in areas that are very
white in the NDVI image?
Figure 10 shows the NDVI (eau_claire_2011ndvi.img
) that was created using the steps found in part 3 of this lab with an inset
viewer over the city of Eau Claire. White areas throughout the entire image
(i.e. not just those within the inset viewer) indicate high reflectance and
show those in which vegetation is prominent
Q8. Comment on the presence
or absence of vegetation in areas that are medium gray and black.
Areas in the image in figure 10 that are
black or medium gray indicate areas that lack vegetation completely or those
that are sparsely vegetated (e.g. they have no or very little reflectance). For
instance, the Mississippi, St. Croix, and Chippewa Rivers all appear black on
the main map in figure 10.
The inset viewer in figure 10 shows the city
of Eau Claire which is generally medium gray. This is expected due to the lack
of vegetation within the city limits compared to the more rural areas
surrounding it. This is because rural areas in this region contain forested and
agricultural land which will show greater reflectance an NDVI image.
Figure 10
Figure 10 shows an NDVI image of
West-central Wisconsin and eastern Minnesota as it appeared in the Erdas 2013
viewer (eau_claire 2011ndvi.img). The inset-viewer in the center right-hand
portion of the screenshot shows the city of Eau Claire in greater detail.
Part 4: Spatial and Spectral Image
Enhancement:
Section 1a: Spatial Enhancement—Low
Pass Filter:
Step 1: Open
the appropriate file in the Erdas 2013 viewer; in this case it is chicago_tm1995_b3.img.
Q 9. This image demonstrates
some amount of high frequency which needs to be suppressed. What is a high
frequency image?
Figure 11a
shows chicago_tm1995_b3.img as it was displayed in the Erdas Viewer after
performing step 1 above. The image is high frequency because its brightness
values (BV) change substantially over a short distance. This is shown by chicago_tm1995_b3.img’s
histogram, which is depicted I figure 11b.
Figure 11
a)
b)
Figure 11 shows the high frequency
image, chicago_tm1995_b3.img (a) and its histogram (b).
Step 2:
Activate the Raster tools.
Step 3:
Click on Spatial, followed by Convolution in the Raster menu. Doing
this will activate the Convolution
interface.
a) Ensure that the input file is chicago_tm1995_b3.img.
b) Under Kernel Selection, select 5x5
Low Pass.
c) Click on the folder icon next to the Output File section of the Convolution interface.
d) Navigate to the appropriate output
folder. Name the output file chicago_tm1995_b3low.img.
e) Leave all other parameters in the
Convolution interface as they are and click OK to run the model.
Step 4:
Dismiss once the model finishes running and close out the box.
Q10. Outline the differences
between the original image and the 5x5 Low Pass filtered image you just
created.
Figure 12
shows spatially enhanced chicago_tm1995_b3low.img (right) compared to the
original chicago_tm1995_b3.img (left). As shown, the new image is much smoother
than the original when both are viewed in Erdas 2013 at the same extent (1:1152587).
This smoothness, which resulted from changes done to the brightness values, is
especially when comparing the space circumvented by the red ellipse on the new
image to that on the original.
However,
when zooming in with the two views synchronized, the spatially enhanced image
becomes blurry more quickly than the original one.
Figure 12
Figure 12 shows the original chicago_tm1995_b3.img
(left) and chicago_tm1995_b3low.img (right), on which a 5x5 low-pass
convolution filter was applied.
Section 1b: Spatial Enhancement—High
Pass Filter:
Q11. What
is a low frequency image?
A low
frequency image is one in which the brightness values (BV) of pixels change
little over a given distance (i.e. it has low contrast).
Step 1: Open
the desired image in the Erdas 2013 viewer; in this case it is sierra_leone2002b3.img.
Step 2:
Using the methods in Section 1a,
apply a 5x5 High-Pass convolution
filter on sierra_leone2002b3.img.
However, save the new file as sierra_leone2002high.img.
Q12. Outline the differences between the original
image and the 5x5 High Pass filtered image you just created.
Shown in
figure 13, the convolved image (right) has much better contrast than the
original image (left). Viewed from the “Fit to Screen” extent (1:995207), what
appear to be roads (white) show up much better in the convolved image than in
the original.
However,
when zooming in (images synchronized) sierra_leone2002b3.img appears to have a
lot of noise. For example, at an extent of 1:86394 the convolved images begins
to take on a salt and pepper appearance.
Figure 13
Figure 13 shows the original sierra_leone2002b3.img
(left) and sierra_leone2002high.img (right) on which a 5x5 high-pass
convolution filter was applied.
Section 1c: Spatial Enhancement—Edge
Enhancement:
Step 1: Open
the desired image in Erdas Imagine 2013; in this case it is sierra_leone1991.img.
Step 2: Open
the Convolution window in the same way as in section 1 a and b.
a) Make sure that the Input image is sierra_leone1991.img.
b) Highlight 3x3 Laplacian Edge Detection under Kernel Type.
c) Click on Fill.
d) Uncheck Normalize the Kernel.
e) Click on the folder icon next to the
output image and navigate to the appropriate folder. Name the new image sierra_leone1991edge.img.
f) Leave the other parameters the as
they are and click OK to run the
model.
Q13. What is
a Laplacian convolution filter?
A Laplacian
convolution filter (LCF) is a linear edge enhancement method approximates the
second derivative of two adjacent pixels. This is done to make the borders
between two features in an image more prominent. In other words, an LCF
attempts to make surface features in an image more easily distinguishable from
one another.
Q14. Outline the differences
between the original image and the Laplacian edge detection image you just
created.
Figure 14
shows the original image (left) and the convolved sierra_leone1991edge.img
(left) in the Erdas 2013 viewer at the same extent (1:944817). Although the
overall brightness of sierra_leone1991edge.img appears to be less than that in
the original, surface features such as roads and rivers are more easily
discernable in the image on the right than in the original.
Figure 14
Figure 14 shows the original image,
sierra_leone1991.img (left) and the convolved sierra_leone1991edge.img (right).
Section 2: Spectral Enhancement:
This section
of the lab will illustrate how to improve on the visual appearance of two
images. This will be done using two methods of liner contrast stretch to
improve them visually.
Also,
histogram equalization will be performed to one of the images.
Section 2a: Min.-Max. Contrast
Stretch:
Step1: Load
the desired image into the Erdas 2013 viewer. In this case it is eau_claire19913b.img.
Step 2:
Activate the Panchromatic tools in
the Erdas menu bar.
Step 3:
Click on General Contrast, then
select General Contrast again from
the resulting dropdown tab. This will open the Contrast Adjust interface.
a) Click on Method and select Gaussian.
b) Click on Apply.
The
resulting image with the Gaussian linear stretch is shown below in figure 15.
Figure 15
Figure 15 shows eau_claire19913b.img
after a Gaussian stretch was applied.
Step 4:
Clear eau_claire19913b.img and do not save.
Section 2b: Piecewise Contrast
Stretch:
Step 1: Load
the desired image into the Erdas 2013 viewer. In this case it is eau_claire1991b5.img.
Step 2:
Click on General Contrast, followed
by Piecewise Contrast in the Panchromatic tool menu.
Step 3: Use
the figure below to set the parameters in the resulting Contrast Tool dialogue
box. These values were taken by moving the crosshairs over the histogram and
taking note of the resulting values.
Step 4:
Click on the Middle Range button of the dialogue box and repeat the same
process as above.
Step 5:
Increase the dynamic range of brightness for the final mode to 180 and apply it
to the image.
Q15. Compare the appearance
of the piecewise contrast stretched image with the original image.
Figure 16
shows the original unaltered image (eau_claire1991b5.img; right) and the
spectrally enhanced image on the left. Notice that surface features in the
enhanced image are much more easily distinguished than in the washed-out
original.
Figure 16
Figure 16 shows the original image (eau_claire1991b5.img)
on the right and the same image after a piecewise contrast stretch was applied.
Section 2c: Histogram Equalization:
Step 1: Load
the desired image into the Erdas 2013 viewer. In this case it is l5026029_0292011b30.img.
Step 2:
Activate the Raster Tools and select
Radiometric, followed by Histogram Equalization.
Step 3:
Click on the folder icon next to the output file section of the dialogue box
that resulted from step 2 and navigate to the appropriate folder. Name the new
file ec_2011_b3_he.img.
Step 4:
Accept all default parameters and run the model by clicking OK.
Q16. Outline the differences
you observed between your input image and your Histogram Equalized image, and
also their respective histograms.
Figure 17a
shows the original image on the left (l5026029_0292011b30.img) and the image
that resulted after histogram equalization was applied on the right (ec_2011_b3_he.img).
Notice that the contrast in the enhanced image is much better than in the
original.
This change in contrast is also reflected by
the respective histograms of the images, with the original displayed on the
left side of figure 17b and the enhanced on the right. The frequency of
brightness values of the enhanced image is much more distributed across its
histogram than in the original.
Figure 17
a)
b)
Figure 17a shows the original image
on the left (l5026029_0292011b30.img) and the image enhanced using histogram
equalization on the right (ec_2011_b3_he.img). The respective histograms of
these images are shown in figure 17b.
Part 5: Binary Change Detection
(Image Differencing):
In this part
of the lab image differencing will be used to detect changes in brightness
values (BV) between two images of Eau Claire and neighboring counties. Both
images were take around the same time, August, though the older one was taken
in 1991 and the newer in 2011.
Section 1a: Creating a Difference
Image:
Step 1: Open
Erdas 2013 with two viewers. In each window open the appropriate images; in
this case they are: ec_envs1991.img and
ec_envs2011.img in view one and two, respectively.
Step 2:
Synchronize the viewers and zoom in and out/pan around to observe any
differences between the two images.
Step 3:
Activate the Raster processing
tools.
Step 4:
Click on Functions in the raster tool menu, followed by Two Image Functions. This will open the Two Input Operators interface.
a) Ensure that the following is correct
within the interface:
1)
Input File 1: ec_envs2011.img
2)
Input File 2: ec_envs1991.img
b) Navigate to the appropriate output
folder via the folder icon next to the Output
section of the interface. Name the output file ec_envs91_11.img.
c) Under the Output Options section of the interface, change the Operator from (+) to (-).
d) Next, click on Layer beneath both input files and change them from All to 4 on each.
e) Click OK to run image differencing.
Step 4:
Dismiss at the end of the run and close out the window.
The images
produced using the steps above is displayed below in figure 18. In this case
the image, on the right, is shown next to one the input images that was used to
create it (ec_envs1991.img; left). Only the NIR band 4 was used to create the
output image in order to simplify its processing.
Figure 18
Figure 17 shows one of the input
images, ec_envs_1991 (left) compared to the newly created ec-envrs_91_11.img
(right). Both images have viewers inserted in approximately the same area.
Section 1b: Estimating the Threshold
of Change:
Step 1: Open
the Image Metadata interface and
click oh Histogram.
Step 2:
Observe the distribution of the histogram and the range of BVs. Determining the
cutoff point of the histogram for the threshold will be determined using the
rule of thumb equation:
[1] MEAN
+1.5(STANDARD DEVIATION)
Step 3: Move
the cursor to the center of the histogram (i.e. in the middle of the bell) and
take note of the value obtained in doing so.
Step 4: Go
to the General section of the
Metadata interface and take note of both the mean and standard deviation.
Step 5: Add
the results obtained from Steps 3 and 4: the resulting sum is the upper limit
for the change/no change threshold.
Step 6:
Repeat the above steps to obtain the lower limit of the change/no change
threshold; this should be a negative value.
Section 2a: Mapping Change Pixels in
a Difference Image Using Spatial Modeler:
This section
of the lab will map changes that occurred within Eau Claire and neighboring
counties between August 1991 and August 2011. Below, equation 2 shows how the
difference will be contrived mathematically using the spatial modeler program:
[2]
ΔBVijk = ΔBVijk(1)- ΔBVijk(2)+C
Where:
1)
ΔBVijk is the change in
pixel values.
2)
ΔBVijk(1) is the BVs of
the 2011 image.
3)
ΔBVijk(2) is the BVs of
the 1991 image.
4)
C is a constant, 127 in this case.
5)
i is the line number.
6)
j is the column number.
7)
k is a single band of Landsat TM.
Step 1:
Select Model Maker from the ToolBox menu in the Erdas 2013 viewer. This
should open the Model Maker interface. Figure 19 displays a diagram in order to
help identify the tools used to run the program.
Figure 19
Figure 19 shows some of the the tools
used in the Model Maker interface.
Step 2:
First, click on the selection tool (A), followed by a Raster Object (B), and
finally in the model panel (E). Refer to figure 20 as to how model
maker should appear after steps 2 through 6 are completed.
Step 3:
Repeat step 2 to add another Raster Object
to the model maker interface.
Step 4:
Repeat the processes in steps 2 and 3, but this time place a Function (C) on the screen. This function should be placed below the two
raster objects.
Step 5: Add
a third raster object to the model maker interface.
Step 6: Use the
Connector Arrow (D) to connect all
the objects in the model maker window. Again, when finished the resulting
interface should look like figure 20.
Figure 20
Figure 20 illustrates how the model
maker interface should appear after steps 2-6 were completed.
Step 7:
click on the raster object in the upper left of the model maker interface and
click on it using the left mouse button. This should open a raster interface.
a) Under Input, bring in ec_envs_b4.img image.
Step 8: Repeat
step 7, but this time click on the right-hand
raster object and choose ec_envs_1994b4.img.
Step 9:
Select the function object and put the following on the bottom of the Define
Function interface:
$n1_ec_envs_2011_b4-$n2_ec_envs_1991_b4+127
Step 10:
Name the output image ec_91-11chg_b.img and save it in the appropriate folder.
Step 11: Run
the model by clicking on the tool that is represented by arrow F in figure 19.The
modified image that was just created should look like the one in figure 21
below.
Figure 21
Figure 21 illustrates how the
differenced image ec_91-11chg_b.img) created in part 5, section 2, steps 1-11 should look when brought into the ERDAS
viewer.
Section 2b:
In order to obtain an image with only the BVs
that changed between 1991 and 2011, another model will need to be created using
Model Maker. However, before this can be done a new change threshold will need
to be determined. Equation 3 will be used to do this:
[3] Mean + 3(Standard Deviation)
Step 1: Open
the image Metadata and select Histogram for image ec_91-11chg_b.img.
a) Once done observing the histogram
information click on the general tab and take note of the Mean and Standard Deviation.
b) Use Equation 3 above to calculate the new change-/no-change threshold.
Step 2: Open
Model Maker and create an Input Raster Object, a Function Object, and an Output Raster Object. Figure 22
illustrates how this model should appear in Model Maker.
Step 3:
Connect all the objects in Model Maker with arrows, as in Section 1.
Figure 22 
Figure 22 shows how the model,
created in steps 2 and 3 above, should appear in the Model Maker interface.
Step 4: Using
Section 2a, Step 7 as a guideline, insert ec_91-11chg_b.img
into the Input Raster Object.
Step 5: Open
the Function Object, as in Step 9 of Section 2a.
a) Change the function from Analysis to Conditional.
b) Click the EITHER IF OR function. This should now appear in the script area of
the Function Define interface.
c) After b has been done the script area
of the Function Define interface should read:
EITHER 1 IF ($n1_ec1>CHANGE-/NO-CHANGE THRESHOLD VALUE) OR 0 OTHERWISE
**Note: the
change-/no-change value should be the one that was obtained in Step 1 of this
section. **
Step 5:
Label the new output image ec_91-11bvis.img
and save it to the appropriate output folder.
Step 6: Run
the model. An Error means that the script will need to be written properly.
Step 7: Open
ec_91-11bvis.img in Erdas; it should look like the one in figure 23 below.
Figure 23
Figure 23 shows how ec_91-11bvis.img,
created using steps 1-7 in section 2b above, should look once it is brought
into the Erdas 213 viewer.
Step 8: Use ArcMap to create a map of the changes
that occurred in the AOI between 1991 and 2011. This should be done overlaying ec_91-11bvis.img
onto ec_envs1991b4.img in ArcMap. Figure 24 shows how the resulting map should
appear.
Q22. Describe the spatial distribution
areas that changed over the 20 year period. Are these areas close to urban
centers or not?
Most of the changes appear that occurred
between 1991 and 2011 appear to be in rural areas, or just outside of city
limits. For instance, some significant changes are circled below in figure 19,
which shows a map that illustrates the changes between 1991 and 2011. This
circled area lies just outside the Menomonie/Cedar Falls area and appears to be
increased cropland when zooming in using the Erdas viewer. This would make
sense since the differencing of the two original images (ec_envs_1991b4.img
subtracted from ec_envs_2011b4.img) are both NIR bands which will exhibit
greater brightness values for vegetated areas.
Also, zooming in on the Erdas images in other
areas on the differenced map (ec_91-11chg_b.img) show that these areas
represent cropland as well.
Figure 24 
Figure 24 shows a map created in in
ArcMap of the changes that occurred in BVs over the AOI between 1991 and 2011.
This was done using the images (ec_91-11bvis.img and ec_envs1991b4.img) created
in section 2b, part 5 of this lab.
References
Wilson, C.
(2013). Remote Sensing of the Environment. Lab 5. Fall 2013. Image Mosaic and
Miscellaneous Image Functions
2 (pdf).
.
(1999).
Erdas Field Guide Fifth Edition Revised and Expanded (pdf). Retrieved from:
Also note,
Dr. Cyril Wilson’s (UWEC) Lab 5 of the same name was used as a template for
this blog. All words in italics are his, verbatim.
