1.9 Saving high precision images

Pixel values on the status bar Saving 32-bit floating point data

TUTORIAL

1. Image files

Images:

ATS_10arcmin_ DSST_0301.hdf Description

Download image (5.7 MB)

Useful information:

Image data: A grid of pixels

Image file types supported by Bilko

Numerical data formats

Null values and missing data

In the previous section you will have noticed some of the differences between opening 8-bit images and images containing other numerical formats, such as 32-bit floating point data. Here you will look the differences between 8-bit and other data in more detail, and learn how to save such high precision data without loosing information.

Note: To continue you should still have the AATSR image ATS_10arcmin_DSST_0301.hdf open, and also at least one of the 8-bit ATSR images you worked on earlier.

Pixel values on the status bar

If you move your cursor around the new image, you will be able to read the temperature of each pixel from the Bilko status bar. You should notice one important difference from the 8-bit images you were working with earlier.

Move the cursor around in all three images in turn, and make a note of the pixel values shown on the status bar. Then use the sidebar information on numerical data formats, and Bilko's context sensitive help program to help you answer the questions below (select from the toolbar and point).

Question 1
a) How does the pixel information on the status bar differ for the AATSR image?
b) How would you explain this difference?

Saving 32-bit floating point data

Before you continue create a new image in the same way as you did for the January 1997 image (see T1.5 ).

use the Go To dialog to select a sub-area corresponding to the one you chose for the previous two images. ( figure (5K)),
Copy the selection and use it to create a new Image doucment, making sure the the Apply stretches check box is not checked.
Save the new image in the usual way ([CTRL+S] ). You will notice that this time the Save As dialog defaults to the file type Bilko.DAT. Accept the default choices and press OK.

Question 2

You may want to review the sidebar information on Image file types supported by Bilko before attempting to answer the questions below.

Answers:
(Resizable
pop-ups)

Answer 1

Answer 2

Back up to:
Q1 Q2

a)	Why do you think it was important that the Apply stretches checkbox was left unchecked when you created a new image document from the AATSR selection?
b)	Can you think of a reason why the default choice for saving the AATSR extract was Bilko.DAT, not GIF?
c)	What other file type could you have chosen to save this image extract without losing information?

The considerations that apply when saving real values (32-bit floats) also apply when saving 16-bit and 32-bit integer data. If you choose a file type that can only handle 8-bit integers (e.g. GIF, PCX and BMP) you will at best loose the accuracy and precision of the original data. At worst you will produce an image where the pixel values can no longer be related to the geophysical values, because the scaling factor and offset are unknown.

Next: Opening Envisat scenes