ES 775 Lab 3

Ikonos and Ground Truth

James S. Aber


Introduction

Ikonos was the first commercial earth-resources satellite launched in 1999. Several similar commercial satellites have followed in the years since. It provides high-resolution multispectral imagery. Individual blue, green, red, and near-infrared bands have 4-meter resolution (linear cell size), and the panchromatic band (green+red+near-infrared) has 1-meter cell size. This spatial resolution rivals conventional aerial photography. Imagery may be ordered from a global archive, or users may specify acquisition for certain dates, locations and cloud-cover conditions.

Ikonos from Satellite Imaging Corporation.

For this exercise, you will work with multispectral data for a portion of Cheyenne Bottoms in central Kansas. Cheyenne Bottoms is the premier wetland of Kansas; it is considered to be the most important wetland site for shorebird migration in the central United States. Cheyenne Bottoms occupies approximately 64 square miles (166 km²) in Barton County near Great Bend and Hoisington. It is managed in part by the Kansas Department of Wildlife and Parks (CBWA) and partly by the Nature Conservancy. The dataset for this exercise covers Nature Conservancy land in the northwestern portion of the bottoms.

ES 775 title image.

Go to the FTP datasets/advanced/lab03/ folder for this exercise. Download Idrisi files for Ikonos multispectral bands: CB_BLUE, CB_GREEN, CB_RED AND CB_NIR. This dataset was acquired in July 2003 during a typical dry summer period. The images have been resampled into UTM zone 14 grid system.

Exercise – Ikonos

Open Idrisi Explorer and examine the metadata for one of the images. Check information about the raster image.

1. Provide the following spatial details about the raster grid: number of rows and columns, linear size (resolution) for each cell, and total ground area (in km²).

Now create two standard color composites: natural color (blue, green, red) and false-color (green, red, near-infrared). Use the color composite function, as you have done in previous exercises. Select the appropriate bands, and provide a suitable output file name and image title. Your resulting images should appear similar to the following examples.

Note: The output files are large, three times the file size of input files.
Make sure you have sufficient disk space available.

Natural-color composite and false-color composite.
Annotation of geographic features (left).

Describe the appearance of the following features in each of the composites, paying attention to color, shadows, and patterns. Note: you need to zoom and pan, in order to display full resolution for selected features.

2. City of Hoisington.

3. Sewage treatment plant southeast of Hoisington (3 black pools).

4. Marshes in Nature Conservancy land.

5. State wildlife area (CBWA).

6. On this basis what are the smallest features you are able to identify visually, and how are you able to recognize these features? Hint: zoom in to see details.

Much of this scene represents wetland conditions, for which the distributions of active vegetation and water bodies are key environmental indicators. In order to highlight these elements, use the red and near-infrared bands to create a vegetation index for the scene. Open the VEGINDEX module (under Image Processing, Transformation). Select NDVI (normalized difference vegetation index), enter the appropriate bands, and provide a suitable output image name. Click OK, and then examine the resulting image and its numerical values.

7. What is the range of numerical values for NDVI? What are the data and file types? How large is the NDVI image (in bytes) compared to the input files?

NDVI files consist of real numbers, which require much more storage space than do byte-binary numbers. It is customary to convert raw NDVI values into a byte-binary scale, as follows.

Do this now, employing a series of temporary files, which you can delete later. Display your final image using the "ndvi256" palette. Examine the image and look at values for different types of land cover.

8. Give NDVI values (0-255 scale) typically associated with the following features.

From your analysis, it should be clear that water bodies have NDVI values of zero. RECLASS the NDVI image so that water bodies have a new value = 1, and all other features have a value = 2. Name the output file "water" and display it with the "qualcoolw4" palette.

9. How effective was this technique for separating water bodies from all other types of land cover? Explain you answer.

10. What is the surface area (in km²) of the water class (1) in this image? What percentage of the total area does this represent?

Exercise – Ground truth

Idrisi has a photo layer capability for linking ground-truth photographs to base maps or satellite images. See Idrisi Tutorial, exercise 1-7. Read about photo layer in the Idrisi Help section on digitizing. The objective of this section is to create photo layers using an Ikonos panchromatic image as the base.

Now download the file NC_CB_PAN into your student working folder. The original panchromatic band had 1 meter spatial resolution. The image dataset was windowed to the specific study area of Nature Conservancy marshes, and cell resolution was contracted to 2 meters to reduce file size.

11. What parts of the spectrum are included in the pan band?

12. What are the smallest objects or features that you can visually identify in this image?

There are only a few ways to improve the appearance of a single-band, gray-tone image, such as this, namely with the STRETCH module (under Image Processing, Enhancement). Experiment with various stretch options to enhance this panchromatic image. Save your prefered version as the base image for the photo layer.

Ikonos panchromatic stretched sample image of the Nature Conservancy study area. For the photo layer, the "nature trail" site has been used repeatedly over the years for small-format aerial photography. Kansas highway 4 runs across the top of the scene along with a railroad.

Aerial photographs of Cheyenne Bottoms.

Small-format aerial photographs (SFAP) images are selected to represent dry (drought) and wet (flood) conditions. Individual SFAP images are identified and described in the following tables. The approximate view direction (azimuth) is given for each oblique image.

Drought conditions during 2006
Browse image Name
Season
Azimuth Features
dry01
spring
45°
Dry mudflats and meadows. Delta of
Deception Creek above scene center.
dry02
autumn
50°
Dry mudflats and meadows.
Nature trail, lower right.
dry03
autumn
130°
Dry meadows in foreground.
CBWA in far background.
dry04
spring
140°
Dry mudflats and meadows.
CBWA in far background.
dry05
spring
225°
Dry mudflats and meadows.
Blood Creek in background.
dry06
spring
300°
Dry mudflats and meadows.
Hoisington in far background.
dry07
autumn
315°
Dry mudflats and meadows.
Kite flyers at bottom of view.
Hoisington in far background.
dry08
autumn
360°
Dry mudflats have been plowed.
Tractor is mowing vegetation thatch.

Flood conditions in 2007 & 2008
Browse image Name
Year
Azimuth Features
wet01
2007
45°
Flooded pools and meadows.
Delta of Deception Creek upper center.
wet02
2008
60°
Flooded pools and meadows.
Delta of Deception Creek upper left.
wet03
2007
90°
Flooded pools and meadows.
Kite flyers at right edge.
wet04
2008
140°
Flooded pools and meadows.
CBWA in far background.
wet05
2007
270°
Flooded pools and meadows.
Blood Creek in left background.
wet06
2007
300°
Flooded pools and meadows.
Hoisington in far background.
wet07
2008
320°
Flooded pools and meadows.
Hoisington in left background.
wet08
2008
360°
Flooded pools and meadows.

Download the SFAP images via FTP into your project working folder. You are now ready to create photo text layers using on-screen digitizing. Read and follow procedures in the Idrisi Help section on digitizing photo layers. Note in particular the procedure for editing or changing existing photo-layer vector files.

You will make two photo text layers—one for drought conditions and the other for flood conditions. First display your stretched Ikonos image in gray tones and maximize the display window (end key). Digitize a point on the nature trail as the position for the text to appear on the Ikonos base map.

Continue by practicing for the flood photo display. Entering the caption can be tricky for multiple photos. It's best to write out the whole caption as a text file, as in the following partial example for the wet (flood) photo layer. Then copy and paste the text into the digitize-options caption box.

Nature Trail <wet01.jpg>{45}<wet02.jpg>{60}<wet03.jpg>{90}<wet04.jpg>{140} ...

In order to activate the photo layer function, click "Feature Properties" button in the Composer window. Now move the cursor to a text label and click to show the associated SFAP images. Examine the SFAP images displayed using each photo layer.

Screenshot of the wet (flood) photo layer displayed on the Ikonos base image. Notice how the small-format aerial photos appear around the nature trail in the appropriate viewing (azimuth) directions. Individual photos can be highlighted, enlarged, or repositioned, as was done for this example.

Notes: You can display only one photo layer at a time on the base image. You must "remove" one photo layer before you can "add" the other. Some of the SFAP images may overlap each other; click on one to move it away or resize it. Unfortunately the composer does not allow saving the complete map composition with the photo layer displayed. The image above was created by taking a digital photograph of the computer monitor.


Now compare the SFAP images with the panchromatic and multispectral Ikonos images. The Ikonos dataset was acquired in July 2003 and represents typical low-water conditions in which small pools of water occupy marshes, mudflats are partly moist, and stream channels also contain water. In other words, the Ikonos image depicts conditions in between completely dry (drought) and full pools (flood).

Many types of land cover are represented in the SFAP and Ikonos imagery: clean water, muddy water, moist mudflat, dry mudflat, emergent wetland vegetation, wet meadow, dry meadow, etc. The images depict these types of land cover in complicated mosaic patterns that change through time.

13. How effective is the Ikonos dataset for displaying, classifying, and interpreting different kinds of land cover at Cheyenne Bottoms? Explain your answer.

14. How could the effectiveness of Ikonos, Landsat, or other types of satellite imagery be increased for land-cover classification at Cheyenne Bottoms?

As your final task, create a photo-layer composition including the Ikonos panchromatic image (with scale bar) and the dry (drought) photo text layer. Reposition and resize individual photos to provide a good view of each, but without completely covering the Ikonos base image (see example above).

Collect a screenshot of your photo-layer composition. The easiest way to do this is the keyboard Print Screen (PrntScrn) function. Paste the print-screen image into Paint for further processing; crop off unnecessary margins and save an image to turn in.

Turn in:


Return to course schedule.
ES 775 © J.S. Aber (2013).