Example image:
(Resizable pop-up)
Eddies in the Somali Current
(87K)).
Downloads:
Images and tools
Lesson
(HTML pages)
References:
Links to more information
Useful information:
(Resizable pop-ups)
Western Indian Ocean currents
The Somali Current system
AATSR
AATSR Flags
MERIS L2 Flags
|
In this section, you will create a composite using the image you have just resampled (ATS_NR_20040220d.dat),
and a number of other images that have been prepared in advance, using exactly the same method as the one you used in the previous section.
To create a composite several images are connected into a Bilko set and a formula applied to calculate mean image from all valid data at each pixel location.
Opening several images as a set
A set is simply a list of images with the same geographic projection, image, window and pixel size. They may be opened individually and then connected, or they may be opened using a Bilko set document - a text file listing the datasets to be opened, one line per image. To save time, a set of pre-processed images has already been created for you, and kept in a folder called preproc.
In Bilko open the set of images to be processed: ats_nr_20040211_20040228.set.
(Change the file type to SETS (*.set) to see the file).
Accept the default Extract settings - you want the whole area.
When the Redisplay dialog opens:
-
check that the null value is set to 0, and change if necessary,
-
set the default stretch to autolinear,
-
and click All to apply this stretch to all the images in one go.
When the set is open, right-click to open the Zoom and set this to Preserve shape to see the whole scene.
Tab through the images to see the AATSR swatch for each day.
Applying a formula to create the mean image
To create a single image from this stack, you will use the Bilko mean function. This is an intrinsic function, which allows you to calculate the mean of selected images in a Bilko set.
Open the formula document mean_all.frm.
(Change the file type to FORMULAS (*.frm) to see it).
As you can see it is very simple. The output numerical format is not set; it will automatically be the same as the original images. The formula itself is the mean function. It takes two arguments:
-
an index: i,
-
and a selector: i>=1, which in this case selects all the images, from the first (@1) onwards.
The selector can be more sophisticated than this, as you will see in a moment.
To apply the formula, use the same method as for the cloud mask: copy and paste it onto the set of images.
When your composite appears as a new image, there are two further steps to complete:
-
Transfer the metadata from one of the images in the stack to the new composite image.
-
Check that the coordinates display OK
(click on the image and read the location of the cursor in lat/lon). If you see pixel coordinates (just numbers) use View > Coords from the menu bar to see the geographic coordinates.
Save the image as ats_nr_20040211_20040228_all.dat. Close it, and open again, to see the metadata used to scale the colour bar.
Finally Redisplay it using a suitable stretch
(set the null value to 0 again, and Bilko will do the rest), and apply the
modjet1.pal palette if you wish, by opening the palette file when the composite image is active.
Before you continue, take a few moments to study the composite image, and compare it to the single image you created from February 20th.
This is most easily done if you
-
give both images the same palette, for example modjet1.pal, and
-
use the same 'Redisplay' stretch, for example a
Linear stretch, with Min.=29700 and
Max.=30200.
To open the Redisplay and change a stretch, simply right-click on the image.
As you may have noticed this composite is based on both descending (morning) and ascending (evening) orbits. It is common to deal with these separately, as diurnal changes in temperature, and the skin effect may lead to temperature differences between the two overpass times.
|
Question 1.
|
| a) |
How would you modify the mean formula to create a composite of only the descending passes? Use Bilko Help to
(figure (37K)) guide you. (Hint: the images in the set you opened to create the composite are named with asc and dsc to indicate ascending and descending passes respectively.)
|
| b) |
Modify the formula as suggested in your answer to (a), and apply it to the stack to create a composite of only descending passes. Repeat to make a composite of the ascending passes. Apply the same stretch to both.
How do they compare to each other and the composite?
|
| c) |
It is often difficult to assess small difference just by visual comparison. By calculating a difference image you may get more information. How would you write a Bilko formula to do this?
|
| d) |
Apply a linear Redisplay stretch to the difference image, with a negative Min and positive Max of equal absolute value, designed to include all the values in the image. Apply the palette anomaly2_minblack_maxwhite.pal.
How would you describe the difference between the two? Can you think of a reason for this? In your opinion is the difference sufficient to make it necessary to deal with ascending and descending passes separately?
|
|
){kind=link}
){kind=link}
