|
9. Gridded data
Useful information:
(Resizable pop-ups)
MERIS level 2 water products
The MERIS product grid
Image data: A grid of pixels
Geocorrection of remote sensing data
Geographical coordinate systems
Equidistant cylindrical (Lat/Lon) grids
The Universal Transverse Mercator (UTM) grid
Image descriptions:
(Resizable pop-ups)
S1998033~.HDF
MER_RR_2C~ 200402~.N1
usgs_10s8e 40s30e.dat
|
Once the contributing scenes have been masked to leave only valid data, they are ready for the last three steps in the process of creating a composite:
-
selecting a grid that will contain all contributing scenes at the right resolution
-
resampling each image to the composite grid
-
combining the scenes and calculating the average chlorophyll values, writing these to a new, composite image.
Choosing a suitable grid
To select an appropriate grid for the new image you should first consider the following:
-
What coordinate system will you be using?
-
How big is the study area, and where are its boundaries?
-
What is the spatial resolution of the contributing images? (There is little point in making a new image with finer resolution than the original data.)
-
What is the error of the geometric correction for the images? (This is 2km for MERIS and AATSR). Are there reasons for using a finer resolution than this?
-
What resampling method will be used, and how might that affect the choice of spatial resolution?
-
-
The size of the new image in terms of rows and columns. To avoid a gridded image that is too big to be handled comfortably by the software when you come to calculate the composite you may need to reduce the study area, split the image into two or more sections, or perhaps reduce the spatial resolution.
You should be able to determine all these things by carefully considering the images you will be combining.
| |
Question 1.
|
| a) |
What coordinate system would you chose for the 3-day composite of the Benguela region?
|
| b) |
What area should you chose as the image window (i.e. what are longitudes of the left and right edges and the latitudes of the top and bottom edges)?
|
| c) |
What spatial resolution (pixel size in degrees) would you choose, based on your study of the contributing scenes?
|
| d) |
There are three possible resampling methods. Which would you choose in this instance?
If in doubt take a look at the pop-up
'Image resampling methods'.
|
| e) |
The geocorrection error for Meris pixels is 2km. How does this compare with the size of your pixels?
Are there reasons why you might need a finer resolution?
|
| f) |
What would be the size of the composite grid given in columns (DX) and rows (DY). Is this a manageable image size?
|
Resampling scenes to the new grid
The new grid will cover a wider area than any of the three single scenes contributing to it.
It is therefore not sufficient to run 'Resample' and accept the default options. You have to ensure that the spatial resolution and window size
are the same for each of the component scenes, otherwise they cannot be combined to calculate the composite.
-
Activate the scene to be corrected
-
Right-click on the image and select 'Redisplay' to set the null value to 255.
-
From the 'Image' menu choose 'Resample' to open the 'Resample' dialogue.
-
With the Interpolation options on top, set the options:
- Based on: Self
- Method: Nearest neighbour
- Null value: 255
Apply the changes before continuing.
-
Select the 'Window' tab and enter the window edge coordinates you decided on in Question 1. Remember to apply the changes.
-
Select the 'Pixel' tab, set the pixel size to you choice from Question 1 and apply the changes.
-
Select the 'Image' tab and check that the size is what you expected from your calculations in Question 1.
If not, check the 'Window' and 'Pixel' options again.
-
When all settings are correct, click OK to accept and resample the image.
-
Save the image as mer2004020*_a1r.dat
Repeat steps 1-9 for the remaining two scenes.
Creating the composite
To create the composite you need to combine all three scenes into a set, and apply a formula to carry out the necessary calculations.
Open the formula composite_mer_algal1.frm. As you can see the formula follows the usual Bilko format and is divided into the following sections
-
A general description of what the formula does.
-
Constant declarations
-
A statement to calculate the number of valid data points at each pixel location
-
A statement to add all the valid data values at each pixel location and write the result to an interim image
-
A final statement calculating the average from the sum and count of valid data.
| |
Question 2.
Before you continue, spend some time studying the formula and use this to help you answer the following questions:
|
| a) |
Like most Bilko formulae this one works on a set of connected images.
How many images are required in the set for this formula to work? How many of these are new blank images?
|
| b) |
What names are these images given in the constant declarations, and what number should each have in the stacked set?
|
| c) |
Why must the output data be 32-bit float?
|
Once your are happy that you know how the formula works, and how to organise the images to achieve the desired output,
you are ready to calculate the composite.
-
Connect the three scenes you resampled earlier into a stack and add the necessary number of blank images (see answer 2a).
-
Make sure the images in the stack are in the right order.
To change the number of an image in the stack use the 'Selector' to highlight it, then type the number
you want it to have. For example, if you want to change @4 to become @1, select @4 and press 1 on the keyboard.
Then select @5 and press 2 on the keyboard. Having done this you will now find the three images have become
@3,@4,and @5, with the two blanks at @1 and @2. Check that this is actually the case, otherwise
the formula may inadvertently over-write the original images. If that happens all you can do is start again,
as Bilko has no 'undo'.
-
Activate the formula, and choose Options! from the menu bar.
Change the output image type to 32-bit float
(necessary because the formula uses the actual chlorophyll concentrations),
and uncheck the 'Special handling for nulls' box.
-
Apply the formula by pasting it onto the set of connected images.
When you examine the two new image planes ( @1 and @2 ) to see how they have changed, you will notice they are both completely black.
This is because the stretch used is inappropriate for the range of data values in the two new images.
|
