Lab 11 - Remotely Sensed Imagery and Color Composites

This lab is a gratefully modified version of lab 10 from Bradley A. Shellito’s Introduction to Geospatial Technologies p752

Learning Objective

This chapter’s lab introduces some of the basics of working with multispectral remotely sensed imagery. The goals to take away from this exercise:

  • Familiarize yourself with the basics of multispectral data manipulation in common geospatial softwares
  • Load various bands into the color guns and examine the results
  • Create and examine different color composites
  • Compare the digital numbers of distinct water and environmental features in a remotely sensed satellite image in order to create basic spectral profiles

Outline:

Submission requirements

Materials (click to download)

Data Name Description
GEOG111_Lab2Questions.docx Handout to turn in
You are answering the questions (laid out in the word doc above and also included in the tutorial below) as you work through the lab. Use full sentences as necessary to answer the prompts and submit it to blackboard. Copy the folder Chapter 10, which contains a Landsat 8 OLI/TIRS satellite image (called clevelandjuly.img) of Cleveland, Ohio, from July 18, 2018. This file shows a subset of a larger Landsat satellite image. We will discuss Landsat imagery in more detail in Chapter 11, but for now, you just need to know that the OLI and TIRS imagery bands refer to the following portions of the electromagnetic (EM) spectrum, in micrometers (µm): * Band 1: Coastal (0.43 to 0.45 µm) * Band 2: Blue (0.45 to 0.51 µm) * Band 3: Green (0.53 to 0.59 µm) * Band 4: Red (0.64 to 0.67 µm) * Band 5: Near infrared (0.85 to 0.88 µm) * Band 6: Shortwave infrared 1 (1.57 to 1.65 µm) * Band 7: Shortwave infrared 2 (2.11 to 2.29 µm) * Band 8: Panchromatic (0.50 to 0.68 µm) * Band 9: Cirrus (1.36 to 1.38 µm) * Band 10: Thermal infrared 1 (10.60 to 11.19 µm) * Band 11: Thermal infrared 2 (11.50 to 12.51 µm)

Keep in mind that Landsat 8 imagery has a 30-meter spatial resolution (except for the panchromatic band, which is 15 meters). Thus, each pixel you will examine covers a 30 meter × 30 meter area on the ground.

Tutorial

Add image to the map

Read in options

Symbology options

  1. Under Channels, you see the three color guns available to you (red, green, and blue). Each color gun can hold one band. (See the “Lab Data” section for a list of the bands that correspond to the various parts of the electromagnetic spectrum.) The number listed next to each color gun in the dialog box represents the band being displayed with that gun.

  2. Display band 5 in the red color gun, band 4 in the green color gun, and band 3 in the blue color gun.

  3. Accept the other defaults for now and click OK.

  4. A new window appears, and the clevelandjuly image begins loading. It might take a minute or two to load completely.

  5. Zoom around the image, paying attention to some of the city, landscape, and water areas.

Question 10.1 What wavelength bands were placed into which color guns? Question 10.2 Why are the colors in the image so strange compared to what you’re normally used to seeing in other imagery (such as Google Earth or Google Maps)? For example, why is most of the landscape red? Question 10.3 In this color composite, what colors are used to display the water, vegetated areas, and urban areas in the image?

  1. Reopen the image, this time using band 4 in the red gun, band 3 in the green gun, and band 2 in the blue gun (referred to as a 4-3-2 combination). When the image reloads, pan and zoom around the image, examining the same areas you just looked at. Question 10.4 What kind of composite did you create in this step? How are the bands displayed in this color composite in relation to their guns? Question 10.5 Why can’t you always use the kind of composite from Question 10.4 when analyzing satellite imagery?
  2. Reopen the image yet again, this time using band 7 in the red gun, band 5 in the green gun, and band 3 in the blue gun (referred to as a 7-5-3 combination). When it reloads, pan and zoom around the image, examining the same areas you just looked at. Question 10.6 Once again, what kind of composite was created in this step? Question 10.7 How are vegetated areas being displayed in this color composite (compared with the arrangement in Question 10.4)? Why are they displayed in this color?

Examining Color Composites and Color Formations

  1. Reopen the image one more time and return to the 5-4-3 combination (that is, band 5 in the red gun, band 4 in the green gun, and band 3 in the blue gun). You can close the other images, as you’ll be working with this one for the rest of the lab.
  2. Zoom and move around the image to find and examine Burke Lakefront Airport, From its shape and the pattern of the runways, you should be able to clearly identify it in the Landsat image. imcenter
  3. Examine the airfield and its surroundings. Question 10.8 Why do the areas in between the runways appear red?
  4. Open a new image, this time with a 4-5-3 combination. Examine Burke Lakefront Airport in this new image and compare it to the one you’ve been working with. Question 10.9 Why do the areas in between the runways now appear bright green?
  5. Open another new image, this time with a 4-3-5 combination. Examine Burke Lakefront Airport in this new image and compare it with the others you’ve been working with. Question 10.10 Why do the areas in between the runways now appear blue?
  6. At this point, keep only the 5-4-3 image open and close the other two.

Examining Specific Digital Numbers and Spectral Profiles

Regardless of how the pixels are displayed in the image, each pixel in each band of the Landsat 8 image has a specific digital number set in the 0–65535 range. By examining those pixel values for each band, you can chart a basic spectral profile of some features in the image.

  1. Zoom in to the area around Cleveland’s waterfront and identify the urban areas. (In the image, these will mostly be the white or cyan regions.)
  2. From the Window pull-down menu, select New Selection Graph. Another (empty) window (called Selection Graph) opens in MultiSpec.
  3. In the image, locate a pixel that’s a good example of an urban or developed area. When the cursor changes to a cross shape, click the pixel once more. Important note: Zoom in so that you are selecting only one pixel with the cursor.
  4. A chart appears in the Selection Graph window, showing graphically the DNs for each band at that particular pixel. (See the “Lab Data” section for a list of the bands that correspond to the various parts of the electromagnetic spectrum.) The band numbers are on the x-axis, and the DNs are on the y-axis. Expand or maximize the chart as necessary to better examine the values. imcenter
  5. The Selection Graph window now shows the data that can be used to compute a simplified version of a spectral profile for an example of the particular urban land use pixel you selected from the image.
  6. For the next question, you need to find a pixel that’s a good example of water and another pixel that’s a good example of vegetation. You also need to translate the data from the chart to a spectral profile for each example. In drawing the profiles from the information on the chart, keep two things in mind: a. First, the values at the bottom of the Selection Graph window represent the numbers of the bands being examined. On the chart below, the actual wavelengths of the bands are plotted, so be very careful to make sure you properly match up each band with its respective wavelength. Note that bands 10 and 11 are not charted because they measure emitted thermal energy rather than reflected energy (as in bands 1 through 9). Note that band 8 (panchromatic) is also not charted. b. Second, the values on the y-axis of the Selection Graph window are DNs, not the percentage of reflection, as seen in a spectral signature diagram. There are a number of factors involved in transforming DNs into percent reflectance values because a DN and the percentage of reflectance don’t have an exact one-to-one ratio, as there are other factors that affect the measurement at the sensor, such as atmospheric effects. However, in this simplified example, you need to chart just the DN for this spectral profile.
  7. Examine the image to find a good example of a water pixel and a good example of a vegetation pixel.
  8. Plot a spectral profile diagram for the water and vegetation pixels you chose on the following diagram. (Remember to plot values as calculated from the DNs.)

imcenter

  1. Examine your two profiles and answer Questions 10.11 and 10.12. Question 10.11 What information can you gain from the spectral profile for water about the ability of water to reflect and absorb energy? That is, what types of energy are most reflected by water, and what types of energy are most absorbed by water? Question 10.12 What information can you gain from the spectral profile for vegetation about the ability of vegetation to reflect and absorb energy? That is, what types of energy are most reflected by vegetation, and what types of energy are most absorbed by vegetation?

  2. Exit MultiSpec by selecting Exit from the File pull-down menu. There’s no need to save any work in this exercise.

TODO

Submission

All you have to turn into blackboard for this week is the final image you created above.