GEOG 372: Introduction to Remote Sensing

October 3rd and 4th, 2007

 

Laboratory 5

Contrast Stretching and DN to Reflectance Conversion in ENVI

 

No late labs can be accepted. Please type the answers just below the questions and hand the printouts at the beginning of the next lab: October 10th (0102) and 11th (0101). Also do not forget to mention your MAJOR on your labs

 

The purpose of this exercise is to explore contrast stretching and conversion of digital numbers (DN) to reflectance in remote sensing data. The data being used is from a Landsat ETM+ scene acquired on October 5th 2001.

 

The purpose of this exercise is to explore contrast stretching and conversion of digital numbers (DN) to reflectance in remote sensing data. The data being used is from a Landsat ETM+ scene acquired on October 5th 2001.

 

Part 1 Histograms and Contrast Stretching

 

- Open ENVI 4.2. On the ENVI toolbar go to File – Open Image File and navigate to the U:g372\372-fall07\lab5data folder. Open three files named L71015033_03320011005_B20, L71015033_03320011005_B30, and L71015033_03320011005_B40. Do not open the files ending with .lut or .sta extensions. These three files correspond to just bands 2, 3, and 4 of a complete Landsat scene and can be identified by the last part of the file name (_B20, _B30, _B40, etc.). Band 2 senses the green portion of visible light, Band 3 senses the red portion of visible light, and band 4 is in the near infrared.

 

 

- Load band 4 as a single band in grayscale. In the toolbar located in your image window, go to Enhance – Interactive Stretching. The window that shows up includes two histograms of your data. The first is your original data, and the second is the stretched data that you see in your image window. The x axis refers to the pixel values, and the y axis indicates the amount of pixels located at each data value.

 

IMPORTANT: In remote sensing data, the raw data is composed of digital numbers (DNs). These numbers can range between 0 and 255 corresponding to different shades of gray, with 0 being black and 255 being pure white.

 

The histograms you see now are of the entire image you see in the scroll window. This histogram does not show the kind of detail we want because it includes a huge amount of pixels with a value of 0 (notice the tall vertical line located at 0).

 

Using the Cursor Location/Value tool, move the cursor around the Scroll image box, and look at the numbers where it says Data: at the bottom of your Cursor Location/Value window.

 

1. What is the source of all of the 0 value pixels?

 

In order to see more detail in the histograms, we are going to look at histograms from subsets of the data file.

 

- From the Tools menu in your image box, open the Pixel Locator and type in the following image coordinates 4000, 4500.

- Now, on the toolbar in the box containing your histograms, go to Histogram Source and click on Image. This changes the histogram from including the entire image to just calculating the histogram from the data you see displayed in the main image window.

 

2. What is the range of data values on the input histogram? How is this different from the output histogram?

 

Again, using your Cursor Location/Value tool, move around the image window, taking note of the different data values.

 

3. What is the reason for the 2 peaks in your histogram?

 

Now go to the following location: 3130, 1100.

 

4. Why is there only one peak in the histogram?

 

Now go to 6900, 6000.

 

5. What do the 3 peaks in your histogram correspond to?

 

Now go to your histograms window. Under the Defaults menu, starting with [Image] Linear, click on each of the histogram stretch types. Also, move to different areas on your image, such as urban areas, agricultural areas, and water. Notice how the input and output histograms can be radically different depending on your stretch type.

 

Previously, we mentioned that when using the Cursor Location/Value tool to look at the numbers where it says Data: at the bottom of your Cursor Location/Value window. These numbers correspond to the values in your input histogram. The numbers on the top of the Cursor Location/Value window where it says for example (Scrn: R:106 G:106 B:106) correspond to the output histogram.

 

6. Why are the numbers the same for R: G: and B:? What if you were looking at a color image?

 

7. Which stretch type gives you the most detail in urbanized areas? What about over the Chesapeake Bay?

 

Now open band 3 as a grayscale image in a new display. Following the same steps as above, open the histogram box, navigate to 4000, 4500, and change the histogram source to image.

 

8. How and why is this histogram different from the band 4 histogram?

 

9. Go to the other 2 locations and describe the difference between these histograms and the ones for band 4.

 

Part 2 DN to Reflectance Conversion

 

In part 1 of the lab, you explored the histograms of raw remote sensing data. This data is in DN format. For part 2, we will convert this raw data to reflectance in ENVI. Conversion of DN to reflectance is a fairly simple procedure in ENVI for Landsat data. Given the correct data documentation, ENVI can locate the proper parameters needed for this conversion automatically from the internet.

 

10. How is DN different from reflectance? Why is it important to perform this data conversion?

To perform the conversion, go to the main ENVI toolbar. Under Basic Tools, go to Preprocessing – Calibration Utilities – Landsat TM.

- In the TM Calibration Input File window, select the band 4 file and click OK.

- Under TM Calibration Parameters, make sure the calibration type selected is reflectance, then click Get Calibration Parameters from Web.

 

11. What is the value for Sun Elevation in degrees? What are the Scale min and max values?

 

 - In the bottom of the window, output result to File and click Choose.

 - Navigate to your personal folder and give the file a name such as Band4Reflectance

 - Click OK.

 

Once the reflectance image has been computed, it will appear in the available bands list as a new file. Load it in a new grayscale display, and open the histogram box.

Using the same process as part one, go to 4000, 4500 and change the Histogram Source to Image.

 

12. How have data values changed as a result of conversion to reflectance? What are your maximum and minimum values in the input histogram now?

 

 - Repeat the procedure for conversion to reflectance for band 3. Once this is done, open this new file in a new display and open the histograms at 4000, 4500, remembering to change the Histogram Source to Image.

 

13. What are the maximum and minimum reflectance values for this band?

 

Notice that in both bands, the maximum reflectance is nowhere near 1.0 or 100%, but in the output histogram, the reflectance data has been stretched over the full range of grayscale values from 0-255.

 

14. Why is it important that these reflectance data values be stretched over the full 0-255 range?

 

 - Link the displays for the 2 reflectance converted bands and using the Cursor Location/Value tool, move around the image and compare reflectance values in each band.

 

15. What is the approximate range of reflectance values for vegetation for each of the 2 bands? Water? Urban areas?