Southwestern+Canada+(H10,+V3)

JC

This will be the page where I talk about the negative vegetation trends in Southwestern Canada, more specifically in the southern regions of British Columbia. More will be added as I have more information available.

Here is a map of the most current of my data

The first of the images below is of the spring, followed by fall, and then summer. I decided to use NDVI because their seems to be a negative trending in vegetation here. There is a lot of water and evergreen in this area. I thought that since these were green all year round it would be a good way to see how these are changing throughout the year. As you can see in the images below you can clearly see a clear difference between the evergreen on the right side of the image and the areas of grassland which turns to snow and ice.

Comments from Kirsten: next time please make jpg's to upload! File - Save File as (on the image window).



Here is a pdf version of my 4th lab. There was to much to just put on the page, so I just attached a pdf doc.



For lab 5 I am looking at the NDVI of this area and comparing this to the albedo. As you can see from the below images, where the albedo increases the NDVI decreases. This is because snow and ice has a very low value in NDVI and it has a very high value in the albedo. Looking at the problem areas over a period of time we will see that the NDVI values will decrease overall and the albedo values will increase as the time series progresses.



With these examples you can see how the NDVI and Albedo are opposites. To make sure that it is the Albedo causing the change I am also going to look at the LAI (Leaf Area Index) to makes sure that it isn't just plant dying that is causing this drop in NDVI. At this time I haven't worked with the LAI data. (I just thought to do this earlier today)

Final Paper for the MODIS section- Jacob Cantrell

LANDSAT

Lab 8

Lab 9. These are the images I made using some of the Convolution and Morphology tools:

There are the 6 PC bands:





Principle components by band number. 1: average 2:brings out things that weren’t shown in the first image 3: brings out more additional features that aren’t seen in the first two 4: Shows noise 5: more noise and you can kind of see the vertical bands in the data that they eye cannot see in a normal image 6: Another noise band but shows horizontal bands in the image that are invisible to the naked eye.



Band Eigenvalues Sum of Eigenvalues % of Variability

The image on the left is the non corrected True Color image. On the right hand side is the PC inverse image. As you can see the largest difference is from the water. This is to be expected since it will have the highest amount of reflectance for the area. We can also now see the difference in types of vegetation.
 * 1 || 987869.1548 || 1342709.155 || 0.735728 ||
 * 2 || 232450.762 || 1342709.155 || 0.173121 ||
 * 3 || 110514.0539 || 1342709.155 || 0.082307 ||
 * 4 || 7109.087329 || 1342709.155 || 0.005295 ||
 * 5 || 3072.483617 || 1342709.155 || 0.002288 ||
 * 6 || 1693.613229 || 1342709.155 || 0.001261 ||

This is the second part of the Lab_Report2. The first picture is the change detection for my area using the two-color multiview image of band 4. As you can see there are two distinct types of change detection. A positive change shows in the sea foam green color. The negative changes show in red. The white-ish colors are the places that have a lesser amount of change.

This second picture is of the PC using band 2 with it's histogram. This shows the change in the land using a greyscale. The darker colors are negative trends, the lighter color are positive trends and the middle colors are little or no change.

This last picture shows the change in NDVI. This one I had trouble with because at first I have only a white image. Using the Linear 2% enhancement. Using this, the below is was my result. This image shows where the NDVI changes. This means the pink areas are the places where there is a change. The black areas are the places that there was no change.



All three of these methods are very informative. They all have different pros and cons. The NDVI for example shows the degree of change better than the other two. The two color shows exactly where the changes occur the best and which was the area is trending. the PC shows the exact areas where there was change and shows the degree of change both positive and negative.

Lab 11 and 12.









Question 1: My lowest user accuracy was in my urban class. I had an accuracy of 86.62 for the urban class. I think that this accuracy is still very acceptable for most cases. The areas that have urban growth in my image also have a lot of green areas. This means that there will be some overlap when I am making my ROI’s. I think having 272/314 pixels correct is enough to generate a 99.63 percent Prod. Accuracy for the urban class. For the second confusion matrix, my lowest accuracy is 79.19%. Again the lowest percent produces a higher accuracy of Prod. Accuracy. This percent gave a 100% prod. Accuracy. I think that this makes the lowish percentile acceptable.