I'd like to show you some of the basics of working with satellite imagery ArcMap. Okay, here I am inside of ArcMap and in my ArcCatalog pane I've got a geodatabase called Vancouver. And inside that, I've got the different bands for any image that was taken by Landsat 7 for the city of Vancouver. So band1, 2, 3, 4 and so on. So if I just take one of those bands, let's say band 4 which is the near infrared, and drag it onto the ArcMap data view, that will load the image for me. So you may look at that and say I guess that must be Vancouver at night. [LAUGH] No it's not. So one thing to be aware of with satellite imagery is that when they have a sensor that they design to orbit the Earth and take images of all different parts of the Earth. They have to be able to create a sensor that's sensitive enough to be able to sense anything that's really dark to really light. So that could be from a really black, dark lake to bright snow and ice in the Arctic or something like that. And so what the ends up with meaning, in terms of what we're seeing here, is that your average image for an average location is not going to use the entire range of possible values from, in this case, from 0 to 255 because it's an 8 bit image. So a lot of the values that you're going to get, in other words, the way that the sensor is calibrated, the values are going to be relatively low. Because they're leaving room, if you want to think of it that way, for really bright things like snow and ice. So that they have that range to be able to sense those things. So things that are not snow and ice, that are sort of more typical, like pavement or water, whatever, are going to be lower in that range between 0 and 255. And so when you load the image in, it's going to look as though it's relatively dark. Fortunately, there's a way that we can enhance this image so that we can see it better. I go over to my roster in the table of contents and right-click and select Properties and go to the Symbology tab. You'll see that we have values from 0 to 255, so that's all that the possible values. And we have a Color Ramp from black to white. And if we scroll down here, we can, if we move down here, we can see Stretch. And then there's different types of stretches that are possible. So Percent Clip, Standard Deviation, Histogram Equalize. These are just different ways of being able to enhance the image. If you've ever used, whatever, Photoshop, Instagram, things like that, you think of them as filters, or ways of being able to enhance an image. This is just kind of a fancier version of that. So one that often works pretty well is Standard Deviations. So I'm not going to explain the details of this right now but essentially what it's doing is taking the range of cell values that are in the original raw image that I've got here. And it's stretching them out so that they do take up the entire range from 0 to 255. There's different methods that can be used to do that, but essentially that's the idea. So if I click OK, Now you can see that we have a much better looking version of our satellite image. We can actually see stuff now. So the important thing to remember is that what I just did there is only affecting the display of the image. It's not affecting the actual original cell values for the data file. And we can use something like the identify tool to click on a cell in our image, and explore what the cell values are. And so what we can see here is that we have two values that are listed, there's a Stretched value and a Pixel value. And so the Pixel value is the actual value of the cell that is stored in the geodatabase. The Stretched value is the version that's been created by doing this enhancement. So that's just for display purposes. That's just telling us that that's what it's pretending it is, let's say, in order to be able to show it better so that we can see it better. So the actual pixel value is 8. Now this is band4, which is the near infrared band. And I know that water absorbs near infrared light quite well. And so that would make sense that the cell value would be quite low. So out of a possible range of 0 to 255, having the cell value of 8 is pretty low. That's what I would expect to see. If I click on another area, like this bright spot down here, that looks like it's probably a farmer's field of some kind. That is a much higher pixel value of 145. It's the original cell value, so much higher than the water value because vegetation tends to reflect near infrared light quite well. So we would expect to see high values there. So I won't go through too many of these but I do really encourage you to use the identify tool to be able to explore cell values for different types of land cover in different bands. I think it's a great way to really start to understand what types of land material or land cover will reflect high or low in this band versus that band. To try to familiarize yourself with what the general kind of characteristics of these land cover types are. So far, we've just been looking at one image, so that's got values from 0 to 255. It's going from back to white. It's kind of grayscale image. But what if I want to look at different bands in combination and assign them different colors, in order to give me a better way of interpreting them? Well, we can do that using something called a band composite. Okay, so let's search for a tool that will give us a band composite. If I type that in. And sure enough there's a tool called Composite Bands. Let's open that. And here's the dialogue box for it. It's pretty straightforward, so I'll just click on the browse to my geodatabase. Vancouver. And I can create a composite of just a couple of bands or three or four or five. Why not do all of them? So this creates a different file. This is what I'm going to do is create a different file with all of these bands added to it. And so I'm going to save this is an .img file, that's an image file. It's a format that was first created for the software ERDAS IMAGINE, which is remote sensing software. But that ArcMap is able to work with. And I find that sometimes it's easier to work with this img format than to work with a geodatabase, believe it or not, depending what I'm doing. Or you may want to have different options available. So I'm going to show you how to create an img file from the bands that you have in your geodatabase. I'll click OK. And you can see down in the lower-right there that it's thinking about it. It's doing the processing that's required to read all of that data from each of the bands and combine it into a new file. So there we have our new file format. And you'll notice right away that we have a different looking image. Because what's happening is we now have three bands available that we can use in combination to be able to visualise our data. If you look more closely, you'll see that there's a red, green, and blue. Those are the colors that are used by the screen in order to display my image. So those are the three that I have to work with for any particular bands that I want to be able to display. So if I left-click on the red box, you'll see that there's a list that comes up of all of the layers that are available to me. Now I happen to know that because I assigned them in the right order, that Layer_1 is the equivalent of band1, Layer_2 is band2 and so on. And so if I wanted to create a natural color image, in other words, make red, the red band look red, the green band look green, and so on. I can say okay, so red I want to be band3. Green, I want to be band2. Blue, I want to be band1. And so that's creating a natural color image. So blue to blue, green to green, red to red. And this is basically simulating what you would see with your naked eye if you were floating above Vancouver or if you just used a camera and took a photo. However, we're not limited to only looking at it from a natural color point of view. We can also do what's known as a false color image. When we do this, we can actually visualize bands that are invisible to the naked eye. So for example, if I assigned the near infrared, which is band4 to red, and band3, which is red, to green. And green to blue. So this is a classic color combination. This is known as a false color infrared color combination, which with Landsat 7 numbering that's bands 4, 3, and 2. So near infrared, red, and green are being assigned to red, green, and blue. And what this does is it mimics or simulates the effect or the look that you would get from an old time film version of infrared imagery. So when they use to use film in cameras that was sensitive to near infrared light, this is the way it would look. And I think it's just become tradition or convention that this is one of the go to color combinations that people often use because we're used to it and it works quite well. It's quite effective. There's lots of other ones you could use, as well, but this is a good one to start with. So you can see from this that vegetation becomes much more prominent because vegetation reflects the near infrared light quite well. And I've assigned near infrared to red, then things in the image that reflect a lot of near infrared, like vegetation, will look bright red. And so it's a great way of being able to differentiate things like neighborhoods with a lot of vegetation in them versus those that don't, or ski hills, or farmland. So you can really get a sense of what is there by using these color combinations. We can use other combinations thought, so for example, I could say let's go 4, 5, 3, and that gives me a different combination with a different look. None of these is necessarily right or wrong. There are guides you can find online to combinations that are useful for particular purposes. So for vegetation, agriculture, or geology, those kinds of things. But I encourage you to experiment with them yourself. It's another way to really start to become more fluent in interpreting what it is that you're looking at from the data that you've got. If you're not sure about the spacial resolution of your data, you can always right-click on it. Go to Properties, go to Source, and you'll see here, for example, it says Cell Size (X,Y) 30, 30. So that means that these cells are 30 meters by 30 meters.
