Tip:
Highlight text to annotate it
X
[MUSIC PLAYING]
This video will show you how to compute the normalized
difference vegetation index or NDVI using ArcGIS.
NDVI can be computed from multi spectral remotely sensed
data using the formula of the NIR infrared ban minus the red
band divided by the NIR infared band
plus the red band.
This yields values that range from negative 1 to positive 1
with negative 1 values indicating low vegetation
content and values closer to one indicating
high vegetation content.
For the demonstration, we're going to be working with the
Landsat 8 scene of Western Vermont.
There's Lake Champlain.
There's the Green Mountains over on the
east covered by clouds.
And there's Burlington, Vermont,
Vermont's largest city.
I created this image by stacking together bands one
through seven from Landsat 8.
Now let's create a color infared composite by going to
the layer symbology and displaying the NIR infrared
band to the red color gun, the red band to the green color
gun, and the green band to the blue color gun creating a
five, four, three composite.
There are a number of ways to compute NDVI in ArcGIS, but
the quickest and easiest is to go to the Windows menu and
activate the Image Analysis Tools.
The first thing to do in the Image Analysis Tools is to
click on the button to open the Image Analysis Options.
Once in the Image Analysis Options window, click on the
tab for NDVI.
We're going to uncheck the Use Wavelength button because we
don't have wavelength metadata.
And we're going to check the box for Scientific Output
because we want our values to range from negative 1 to 1.
We're then going to adjust the band combinations.
Band five in this particular Landsat image corresponds to
the NIR infrared band.
And band four corresponds to the red band.
So we'll set band five and four as the infrared band and
red band respectively and then click OK to
activate those settings.
Now we're ready to compute NDVI.
First, confirm that the image layer we want to use is
selected, and then we'll scroll down and under
Processing and click on the NDVI icon, which
looks like a leaf.
Using the Image Analysis functions, we can compute NDVI
in a matter of seconds.
But it's important to note that the NDVI layer that we've
created is only temporary.
To improve the display of the NDVI layer, we're going to go
into our Layer Properties, and under the Symbology tab,
adjust the color ramp.
We'll use a red to green color ramp where green represents
high vegetation content and red low vegetation content.
Let's examine our output.
Lakes and urbanized areas that have very low amounts of
vegetation have corresponding low NDVI values.
Clouds also have low NDVI values.
If we look in the shadowed areas of the clouds, we can
see that NDVI values are impacted.
Any within the shadowed areas, the NDVI values tend to be
lower than similar vegetation in non-shadowed areas.
Zooming into downtown Burlington gives us a chance
to look at NDVI values in an urbanized area.
We see that the city center, which has very few large
patches of vegetation, has low NDVI values, but surrounding
areas, even residential areas with large tree canopy, do
have higher NDVI values.
Now let's move south to the landscape that's dominated
more by agricultural land use.
Those fields with active, healthy crops
have high NDVI values.
NDVI values tend to be low in those fields where the crops
have either been recently harvested exposing the bare
soil or the crops are in poor health.
Finally, we'll take a look at some wetland vegetation
adjacent to a river.
We see a mix of NDVI values.
The lower NDVI values in those areas where we have
predominantly water and the higher NDVI values are those
areas where the vegetation is obscuring the water.
To make the NDVI layer a permanent raster, we'll go
back into the Image Analysis Tools and click
on the Export icon.
This launches the Export Raster Data window.
The first step is going to be to specify the location for
this new raster file.
After we've reset the location, we'll go through and
give it a file name.
In this particular case, I've used the .tif extension to
export the raster as a .geotiff .
Once we're set, we'll click Save and a new raster layer is
produced which we're adding to our ArcMap document.
I'm going to remove the temporary NDVI raster and go
into the symbology for our newly added raster and give it
a red green color ramp.
Now we'll take a look at how you can use layer symbology to
determine a threshold to differentiate vegetated areas
from unvegetated areas.
Within the layer symbology, I'm going to select the
classified option for raster display changing the number of
classes to two and applying a red green color gram so that
red means not vegetated and green means vegetated.
Clicking on the Classify button will
bring up the histogram.
And I can use the slider to attempt to determine the
appropriate threshold value for non-vegetated and
vegetated pixels.
In addition to using the slider bar, I also have the
option to enter the break value manually by typing it in
under Break Values.
We can now use this threshold value in the raster calculator
to create a new raster layer that consists of vegetation
and non-vegetation.
Opening up the raster calculator, we're going to use
our NDVI layer in the expression and simply say
greater than 0.25, which we determine to be
the threshold value.
We're going to output this to a new raster layer in GeoTIFF
format, so we're adding the .tif extension at the end.
Clicking OK will execute the raster calculator operation.
The pixels in our resulting raster layer have values of
zero and one.
Zero means the NDVI was less than or equal to 0.25 and one
means the NDVI value was greater than 0.25.
To evaluate how well we did with this threshold, we can
use the Effects Toolbar and the Swipe tool to swipe our
binary layer over the original image dataset.
[MUSIC PLAYING]