2.1 Detecting internal waves off Mozambique

The Mozambique shelf has long been recognized as a region with strong internal tides. The area is highly active, with generation and propagation of internal solitary waves. These waves may play a part in transporting sediment, chlorophyll and pollutants, and could have an effect on the life cycle of commercially valued marine animals. Knowledge of the behaviour and characteristic of these waves is therefore important, not just to physical oceanographers, but also to ecologists.

Finding internal waves in SAR images

Before you start working with the satellite images, take a look at the map ot the Mozambique study area (17K)). You may find it useful to refer to this map from time to time, so there will always be a pop-up link to the map on the left sidebar.

If you are new to the phenomenon of internal waves, you should also spend a little time reading about them in the sidebar pop-up What are internal waves? This explains briefly how these waves are generated, and why they may be observed using SAR, which after all only 'sees' the ocean surface.

1. Open the ASAR WSM 1P image dated 28 January 2004 (ASA_WSM_1PNUPA20040128_~.N1), making sure the Extract box of the Open dialogue window is ticked (figure (8K)).
2. Select the folder named Bands from the left frame of the file-structure window (figure (6K)), and open the folder proc data by double-clicking on the folder icon in the right frame of the window.
3. In the 'Extract' dialogue window select a sampling of 8 in both the X and Y dimensions. (figure (5K)) This will give you an overview of the whole image.
4. In the 'Redisplay Image' dialogue set the null value to 0 and select a gaussian stretch from the 'Use' drop-down menu (figure (6K))

In the centre of this image you should be able to identify an area of internal wave activity. This is two darker bands close together, which curve around an angle of about 90^o. If you are unsure about where to look, you may find the area of interest by using the Go to dialogue:

1. In the 'View' menu ( [ALT+V]) change the geographic coordinates to pixel coordinates by unchecking Coords. (figure (2K)).
2. From the 'Edit' menu select 'Go to' ( [CTRL+G]) to open the Go to dialogue. Choose Box Selection from 'Selection Type' drop down menu and set the pixel start position to [x,y] = [350, 250] with a box size of Dx=125 and Dy=125 (figure (2K)). Click 'OK' to go to the selected area (15K)).

Having identified the area of interest, you will now open it at full resolution. Activate the file structure window by clicking on it, and double-click on the proc_dat folder icon to open the Extract dialogue again. This time set the Sample rate to 1. (Do not click OK yet, you have to set the extract 'window' first.) .

Having answered the questions you are now ready to complete the Extract settings (5K) and open the selected area at full resolution. Save the full resolution extract as asawsm_moz28_ext.dat. (Open the 'Save As' dialogue [ALT+F] > [A] ), type the new name into the File name box, and select Bilko.dat from the 'Save as type' drop-down menu. (figure (2K)).

Applying a filter to reduce speckle

As you can see the image is quite grainy (or noisy). This is a characteristic of SAR images known as speckle You can reduce the speckle by applying a mean filter:

1. From the Image menu, select Filter > Options (figure (4K)).
2. When the 'Filter Options' dialogue appears, make sure all the check-boxes are checked (figure (5K)).
3. Select the whole image ( [CTRL +A] , or choose 'Select All' from the 'Edit' menu).
4. From the Image menu, select Filter > Mean (figure (4K)).
5. In the 'Filter Size' dialogue set the rows and columns to 5 (figure (3K)), and press OK to start filtering.

You will notice that the new image that is created has a much smoother appearance, and that the internal wave packet in the centre of the image is more clearly distinguished from the background. Save this new image as asawsm_moz28_5b5_ext.dat.

Bilko also has two SAR specific speckle filters - Lee and Frost. These smooth the image like the mean filter, but avoid blurring or hiding points and lines that are the result of actual changes in surface roughness rather than speckle. These filters may often work better if the filter window is a little larger - say 9x9.

1. Select filter from the Image menu, but this time, choose Frost.
   You will notice that the Frost filter needs more input from you; not just the window size, but also the number of looks.
2. To fill in this box, you need to know how many single radar measurements (looks) have been averaged in a pixel. You will find information about this in the Metadata > SPM file, so open this text file.

1. Apply the Frost filter as suggested by your answer to question 2b.
2. Close the file structure window, so all you have left are three images, the original data, the mean filtered data, and the Frost filtered data.
3. Select the Window menu on the menu bar, and tile the images vertically so you can compare the results of the filtering.
4. Connect the three images into a stack and take a transect across one of the smaller wave features.
5. Open a new transect document to show this transect. The raw data will be red, the mean filtered data green, and the Frost filtered data blue.

Identifying the direction of wave propagation

'What are internal waves? ' explains how the circulatory motion of the water in internal waves lead to the characteristic dark and light bands seen in SAR images. The characteristics of these bands may also provide information about the direction in which the wave packet is travelling. Combined with the use of bathymetric maps there may clues to where the waves are generated.