SRTM_Mask
Back to Tutorials | Back to SRTM Image Processing | Next Page
I should note that each image is different, because of the different land ocean coverage areas. A lot of areas are not even going to have ocean areas at all. We only need to do this step "SRTM_Mask" for ocean bordering tiles. Land-locked tiles can skip to the "SRTM_Elevation" step.
Navigate the menu control "Image->Adjustments->Threshold".
You'll get a popup window that looks like the one in the image below.
This is a touchy control and slightly subjective to the image for this purpose so I'm going to cover the basic principles of selecting the correct threshold level for any image. We'll use this one as the prime example because it is a great example to cover all the necessary points to fully understand the concept.
The concept of the threshold control is pixels that are above the threshold limit are turned pure white. Pixels below the threshold limit are turned pure black.
There is no grey, just like there is no spoon.
We need to drag the threshold slider to the left and find the right value for isolating the shore line as best as possible. The graph above the slider shows an amount of pixels that will be converted over from black to white. In this case all the elevation data that we are trying to isolate from the ocean is off to the far left of the graph. If we do a threshold on that operation it will destroy the boarder around the shore line that we are trying to isolate from the land.
We want to make this process as easy as possible, and this is the best way to isolate that shore line from the noise of the water.
The water has noise because when the space shuttle flew over the world with it's 50 foot boom and the two giant radar dishes, the signals coming out of those dishes are reflected from the surface of the planet. The ripples in the water of lakes, rivers, and oceans do not reflect a smooth signal. So the image gets garbled near the shore and over the water. This includes areas that are swampy, muddy, or have a very high water table.
Below the slider is in the middle, and we can see that the water is still to dark. The shore line especially in the bay is not well defined. Drag the slider to the left as indicated in the image below. Look at the other images and captions to better understand how this function works and what kinds of problems will take shape if the proper value is not found.
Click on the image for a larger view.
Below you can see that I have found the perfect threshold level. The shore line in the bay is still not well defined, but it is optimized at the best possible level without picking up the extra noise from the land data. Looking at the graph will help you find that perfect value, because the graph shows you where that land data begins along the slider. You want to be right up against that, not to far into it by even one value, and not to far away from it, by even one value, and it's not easy to find that perfect solution, but it is possible to get close and optimize it for our needs. That's all you need to do.
Click on the image for a larger view.
Below you can see that I have taken the slider to far to the left, and it needs to come back to the right as indicated in the image. You can see in the lower right especially that some of the land data is now screwing up the mask. This would add work and mix me up later in the process of cleaning this mask, because it is already difficult enough identifying areas that are water, and separating them from land. The last thing we need is more land areas being shown as water.
Click on the image for a larger view.
Below you can see that I have taken this to the extreme far left. Much to far, but the outcome is obvious. Now much of the land has been converted to white and this will totally screw up any objective of isolating the land from the ocean. But you should get a better idea of what we are trying to do here.
Click on the image for a larger view.
Back to Tutorials | Back to SRTM Image Processing | Next Page