ES 775 Lab 1



This exercise is based on an exceptional Landsat MSS dataset acquired on January 13, 1979. The atmosphere was unusually clear with few clouds visible in the subscene for the Mérida vicinity. FTP the following files to your student workspace. Note that each image file consists of two parts: *.rst and *.rdc. Both are necessary for Idrisi to create a display.

Landsat MSS files for Mérida subscene.
MR1-1.* 1 = green light 0.5 to 0.6 µm
MR1-2.* 2 = red light 0.6 to 0.7 µm
MR1-3.* 3 = shortest infrared 0.7 to 0.8 µm
MR1-4.* 4 = near infrared 0.8 to 1.1 µm

Mérida is located in the tropical Andes Mountains of western Venezuela at a latitude of ~9°N. Elevations in the area extend from less than 1000 m along the Río Chama valley below Mérida to more than 5000 m (16,400 feet) on Pico Bolívar. This great range of elevations over short distances gives rise to dramatic variations in local climate, vegetation, geomorphic processes, and human land use.

Distribution of precipitation is strongly controlled by elevation and prevailing winds. Most rain/snow falls on the southeastern side of the major mountain ridge located to the southeast of Mérida. Annual precipitation amounts exceed 2½ m in this zone; whereas the Río Chama valley generally has less than 1 m annual precipitation. Summer is the rainy season, and winter is relatively dry. Temperature for a given site varies little during the year.

Semideciduous forest predominates at lower elevations on wet slopes, and evergreen (siempre-verdes) forest is found at elevations up to around 4000 m (~13,000 feet). Alpine shrub (páramo) vegetation is developed above 4000 m. Much of the páramo zone is within the Parque Nacional Sierra Nevada. At lower elevations in relatively dry valleys, desert-like vegetation consists of cactus, thistles and thorny plants. Main agricultural crops of the region include potatoes and carrots, which are grown year round, cattle are grazed in high meadows, and trout are raised in fish farms.

City of Mérida, "Mercado Principal" (main market).
View of evergreen forest in tributary valleys above Mérida.
Páramo vegetation on slopes above the village of Apartaderos.
Desert vegetation in the Río Chama valley below Mérida.
All photographs © J.S. Aber.

The Sierra Nevada de Mérida is the product of continuing plate movement along the Bocono fault zone. This boundary separates the South American plate from the Caribbean plate. It is a converging dextral (right-lateral) transform, along which strong compression has thrust up this portion of the Andes Mountains. The Río Chama valley follows the Bocono fault system. The high mountain peaks supported many glaciers during the Ice Age, and typical glacial landforms such as cirques and moraines are common in the páramo region. Small glaciers--Humboldt Glacier--still exist on Pico Bolívar, which is snow covered much of the year.

See Imagen Atlas de Venezuela
(p. 164-179, GSA lab, map case)


Begin by using Idrisi Explorer to set up a project for this exercise. You should be able to access MR1 files in your working folder. Next examine the documentation files for MR1. Use the "Files" tab of Idrisi Explorer to display the files. Click on MR1 to view its metadata below. The metadata file gives basic information about data and file type, and geometrical features of the raster data. Resolution gives the linear dimension of each data cell in the reference units for the raster grid.

1. How many rows and columns does the grid contain? What is the area of each cell in the grid? What ground area is represented by the image?

Now you may minimize the Idrisi Explorer. Next, create a standard false-color composite. Read about the COMPOSITE module in the Help section. This is one of the most important procedures for remote sensing image processing. Open the COMPOSITE module (4th icon from left). Enter MR1-1 for the blue band, MR1-2 for the green band, and MR1-4 for the red band. Enter MR1-124 for the output image. Click on "Linear with saturation points" and select "Create 24-bit composite with stretched values." Give an appropriate title, and click OK.

When the image is ready, it will display automatically in the composite palette. Hit the "end" key to maximize the display size. This composite has a color rendition similar to color-infrared photographs--see following example.

Sample image of the Mérida 124 composite image. Yours should look similar, but without captions and scale bar. Click on the small image to see a full-sized version.

2. Locate and describe the following features. Note such attributes as color, pattern, texture, shadows, etc.

Shadows play an important role in helping to recognize landscape topography, which often reveals geologic features. Shadows can also obscure objects that are not fully illuminated. This can make identification of ground features difficult. It is therefore important to recognize the presence and effects of shadows in Landsat scenes.

3. Where is the apparent position of the sun in relation to this Landsat scene? Explain your answer.

4. Clouds are other features that have strong influence on the appearance of Landsat images. Where are the clouds in this scene? How can you tell them apart from snow-capped mountains?

The color coding depicted on this image is the standard; it resembles color-infrared photographs. You should become quite familiar with this color combination. However, many people find this false-color composite confusing, mainly because active vegetation appears red, maroon, and pink. As your next task, make another false-color composite, as before, but with these color assignments: band 1 = blue, band 4 = green, and band 2 = red. Name your output image MR1-142, and display it as before.

Sample image of the Mérida 142 composite image. Yours should look similar, but without title and scale bar. Click on the small image to see a full-sized version.

This image has a much more naturalistic appearance, because vegetation appears green. It still is a false-color composite, however, based on visible and near-infrared bands of data. Now use the Map Composer (box on right) to put special features on the display. Click on "Map properties" in order to add the following features:

Place the scale bar in lower right corner of image and north arrow in upper left corner. Save your final map composition as MR1-COMP. Also save a "bmp" file, then use Paint to convert the bmp file from 24-bit to 256 colors (use "save as" function). File size is reduced considerably, although some colors may be changed slightly. Further reduction in file size may be accomplished by converting to "gif" or "jpg" format. Turn in your image file with your answers.

5. What declination value did you use for the north arrow?

6. What file size (in bytes) is the 24-bit bmp image? What size is the 256-color bmp image? How do these numbers relate to image size (number of pixels)?
Now you will experiment with band ratios. In order to make ratios, it is first necessary to correct for the effects of haze in the raw data. Haze is generally a factor for the visible bands (1 and 2), but not for the infrared bands (3 and 4). This is because shorter wavelengths of visible light are scattered more in the atmosphere than are longer infrared wavelengths.

The amount of scattering also depends on atmospheric conditions at the time of image acquisition, elevation of the image site, and other factors. The amount of haze is indicated by the minimum value for a band of data. Under ideal conditions, no haze results in a minimum value of zero. Higher minimum values indicate the effect of atmospheric haze in the measurements.

Use the Metadata module to examine the minimum values for bands 1-4 of the dataset. You will find that all bands have a minimum value of zero.

7. Explain why there is no apparent haze effect in this dataset.

On this basis, you do not need to perform haze correction for any band. In order to make band ratios, neither the numerator nor denominator may have a zero value--impossible to divide by zero. It will be necessary to adjust band values so that the minimum is 1. Use the SCALAR module (under GIS Analysis--Mathematical Operators). Input files are MR1-1 etc. and output files are MR1-1Z etc. Click on "Add" for the operation, and indicate the scalar value as 1. Output documentation with an appropriate title, and give "scalar +1" as the value units. Upon running scalar, the corrected image will appear with a legend of values. Compare the min/max values for the original files and the corrected files.

You are now ready to make a band ratio. Use the OVERLAY module (4th icon from right). Select MR1-3Z as the first image and MR1-1Z as the second. Name the output image MR1-31R. Click on "First/Second" as the overlay operation. Output documentation with an appropriate title, and give "3/1 ratio" as the value units, and click OK. Your image will display automatically with the Idrisi 256 palette (default). Note: autoscaling is utilized for an image of real data type.

Note the legend to see the numerical range of values and color coding, and use "Feature properties" to examine values for individual cells in the image. Compare this image to your MR1-COMP in order to recognize landscape features. The 3/1 ratio represents near-infrared/green. It is a good general-purpose display of vegetation, water bodies, human land use, etc.

8. What kinds of landscape features are depicted by ratio values > 4, values between 2 and 3, and values < 1?

9. Describe the appearance of clouds and shadows on the ratio image in comparison to the composite image.

Use the Windows file manager now to look at file sizes. Compare the ratio image files with the original image files. It is obvious that real-binary format takes much more disk storage space than byte-binary format. To save storage space and to facilitate further processing, you will convert MR1-31R to byte binary format.

Two steps are involved. First use SCALAR to multiply the ratio file by a constant value. Name the output file TEMP1. The resulting file should have a maximum value around 250, but not > 255. Then use the CONVERT module (under Reformat) to change from real to byte-binary format. Click on "Rounding" for integer conversion type (default), and name the output file MR1-31B. Examine its metadata to see what happened to the numerical values.

10. What constant value did you use for the scalar multiplication?

Ratio images (in byte-binary format) may be used with other bands to build false-color composite images; the results can be dramatic. Make a false-color composite with the following band/ratio combination: band 2 = blue, band 4 = green, and ratio 3/1 = red. Select linear stretch with saturation, and create an 24-bit composite, as before.

Note: use MR1-2Z, MR1-4Z and MR1-31B as the input images for the composite.

Name the image MR1-RAT. This type of composite has the advantage that it includes information from all four bands of the original dataset.

11. Describe the appearance of features in the ratio-composite image, as for question #2, paying particular attention to vegetation cover.

Sample image of the Mérida special composite image. Yours should look similar, but without title and scale bar. Click on the small image to see a full-sized version.

As a final task, read about the CLUSTER module (under Image Processing, Hard Classifiers). This is a basic technique to identify and classify image cells that have similar numerical values in all bands of the dataset. The idea is to recognize natural groupings or clusters based on band values, in other words spectral signatures. In order to run CLUSTER, select the "Number of files" as 4, and enter the four original file names, MR1-1, etc. Name the output image MR1-CLUS, accept other default values, and click OK.

The resulting clustered image will display with a legend and the qualitative palette (random colors). You should see 14 clusters. This image is clearly quite different from the others you have made. For example the clustered image lacks shadows and looks "flat" compared with the other images. Compare your various images, in order to recognize different features on the clustered image. Each class on the clustered image represents a specific combination of band values that relate to features on the ground--vegetation, soil, human structures, water, etc.

Sample image of the Mérida cluster analysis. Yours should look similar, but without title and scale bar. Click on the small image to see a full-sized version.

This approach is called unsupervised classification, as the identity of classes is not specified in advance. The nature of each class, thus, is not known. The features represented by each class can be determined only through examination of the original imagery and knowledge of ground cover.

12. Identify the clusters that represent evergreen forest (southeast corner of scene) in sunlit situations. What clusters represent forest in shadowed positions?

Turn in:

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