ES 771 AVHRR Vegetation

EASTERN NORTH DAKOTA

Introduction

This exercise is based on six AVHRR LAC scenes extracted from the Northern Great Plains AVHRR data set on CD-ROM. The scenes cover part of eastern North Dakota corresponding to the Devils Lake and New Rockford 1:250,000 map sheets (47-49°N and 98-100°W). The six scenes were acquired during 1987 and 1988 as follows.

Scene Date
ND1 June, 87
ND2 August, 87
ND3 April, 88
ND4 May, 88
ND5 June, 88
ND6 August, 88

For each scene, you have band 1 and band 2 images: ND1-1, ND1-2, ND2-1, etc. Set up a "main working folder" for this exercise on your computer. Copy these files (via FTP) into your main working folder, where you will conduct the exercise.

The area of this scene is covered mainly by cropland, both dry and irrigated, grassland (pasture), numerous lakes, and extensive wetlands. Forest cover is limited to steep hills and valleys. The entire region was repeatedly glaciated during the Pleistocene, and thick glacial sediment (till) mantles the surface. Noteable physiographic features include: Turtle Mountain (NW), Missouri Coteau (SW), Devils Lake basins (center), and deep spillway valleys, such as Sheyenne valley. Examine the topographic maps (on campus) to become familiar with the geography.

See ground views of North Dakota.
Review exercise 2 on Devils Lake.

Note: AVHRR images are quite different from Landsat in terms of spatial resolution (cell size) and wide field of view. Keep this in mind for the following exercise.

Exercise

Use Metadata to examine ND1 documentation files.

1. How many rows and columns of cells does the scene contain? What is the ground area of each cell (in km²), and what ground area does the image cover (in km²)?

Now display ND1-1 and ND1-2 as two windows. Choose title, no legend, grey scale, and autoscale. Maximize the size of each display (end key), and position the two displays side by side for easy visual comparison.

2. Why is band 1 generally dark? Explain based on spectral conditions of ground features.

3. Why is band 2 generally bright? Explain based on spectral conditions of ground features.

Next make a similar two-window display for ND2-1 and ND2-2. ND2 is generally darker for both bands than ND1.

4. How could you account for the rippled appearance of ND2? Hint: think about atmospheric conditions.

5. Explain why Devils Lake is so bright in both bands of ND2. Hint: AVHRR has a wide field of view which is subject to special lighting conditions.

AVHRR images have proven valuable for documenting general vegetation conditions over large land areas. Take the ND1 bands to create a simple vegetation index based on the 2/1 band ratio. This is an infrared/red ratio. Use the VEGINDEX module (under Image Processing, Transformation) to create this ratio. Choose the Ratio option, and call the output file ND1-RAT. The image will display automatically with the NDVI palette, legend, and autoscale.

ND1 sample image. The band 2/1 ratio is depicted to show active vegetation in the scene. Dark green indicates the greatest amount of active vegetation. Light green, yellow and tan indicate intermediate vegetation cover, and water bodies are dark brown/black. Your image should appear similar, but without captions.

6. Describe the general distribution of vegetation displayed in the scene.

7. Give the ranges of typical band-ratio values for the following features, and describe the kind of vegetation and ground conditions that probably exist.

8. What are the data and file types of ND1-RAT? What is the file size (in bytes) of ND1-RAT? How do these parameters compare with ND1-1 (or ND1-2)?

Real numbers (with fractional parts) require considerably more disk storage space than do byte-binary (or integer) numbers. This may cause a space problem with large images. Byte-binary format is limited, however, to 256 whole integer values (0-255).

Next prepare 2/1 band-ratio images for the five other scenes following the same procedures and naming scheme as you used with ND1-RAT.

9. Use Metadata to check the maximum and minimum values for each band-ratio image. Give these values--fill in table below.

Image Max Value Min Value
ND1-RAT
ND2-RAT
ND3-RAT
ND4-RAT
ND5-RAT
ND6-RAT

10. On this basis, which image apparently has the strongest vegetation signal, in other words the highest maximum value?

Idrisi Andes can display a series of images as separate windows on the screen. This allows for quick visual comparison. The band-ratio images should be stretched (4th icon from left) for display. The manner of stretching is important; stretch each ratio image the same way, using: linear stretch, specify lower bound = 0.5, specify upper bound = 4.4, and 256 display levels. Name the stretched images ND1-RATS, ND2-RATS, etc.

Now display each stretched image with no title, no legend, grey scale, no autoscaling. Arrange the windows so you can see each in overlapping fashion.

11. Why were 0.5 and 4.4 used as minimum and maximum values for stretching each image?

12. Describe apparent changing seasonal vegetation conditions during the spring and summer of 1988.

13. Can you discern any differences in vegetation between June 1987 and June 1988 growing seasons? If so, describe and offer an explanation.

The infrared/red ratio can be used for more than just vegetation. This ratio also identifies water bodies, regardless of water depth, turbidity, or sun glint. Reexamine ND1-RAT (not the stretched version) paying particular attention to pixel values for Devils Lake and the surrounding shore/wetlands.

You will now use ND1-RAT to create a classified image that shows only large lakes and shore/wetlands of the scene. Read about and use RECLASS (2nd icon from right) to accomplish this. Reclassify ND1-RAT with the following ID assignments: lakes = 1, shore/wetlands = 2, all other areas = 3. Name the reclassified image ND1-CLAS. Accept the warning to convert output file to integer values. Your image will display automatically with garish colors of the qualitative palette.

14. Describe how you created the ND1-CLAS image.

15. How much of the scene is covered by lakes; how much by wetlands? Hint: use AREA (tabular output) to find km² in each class.

To complete this exercise, make an image composition based on ND1-CLAS. The composition should include an appropriate title, subtitle (your name & date), scale bar, and legend with captions. Create a custom palette in which the background is light brown, lakes are dark blue, and shore/wetland is pale green (see Symbol Workshop, third icon from left). Name your final composition DV-LAKE and make a digital image file (jpg) to turn in.

Note: Creating customized legend captions is tricky! Open metadata for ND1-CLAS. Click on the "Categories" line, then click on the [...] button to right side. This will present a blank table for Code and Category entries. Notice the small buttons on right side. These allow you to insert, remove, and add lines. Add two lines, then fill in codes (1-3) and captions for each code. Click OK, then be sure to save (Ctrl + S) the modified metadata. You will have to redisplay the image for the new legend to take effect.

Turn in

ES 771 schedule or ES 771 homepage.

© Notice: ES 771 is presented for the use and benefit of students enrolled at Emporia State University. Any other use of text, imagery or curriculum materials is prohibited without permission of the instructor, J.S. Aber.