Exercise 8: Creating a mosaic dataset to contain Landsat imagery

This topic applies to ArcGIS for Desktop Standard and ArcGIS for Desktop Advanced only.

Complexity: Intermediate Data Requirement: ArcGIS Tutorial Data for Desktop Data Path: C:\arcgis\ArcTutor\Raster\Data Goal: Learn how to add the Landsat 7 ETM+ raster type to a mosaic dataset.

In this exercise, you will create a mosaic dataset containing a Landsat 7 ETM+ scene. This exercise will demonstrate the use of the product definition, which can be defined when creating a mosaic dataset. The product definition allows you to customize the mosaic dataset to contain data with a specific number of bands and wavelengths. The product definition controls how the data is added to the mosaic dataset, how it displays by default, and aids in some processing. It is ideal to use a product definition when creating a mosaic dataset to contain satellite imagery.

You will create a Landsat (6 band) mosaic dataset that is useful for analysis since the pixel values are not altered.

This exercise does not rely on any previous exercises.

Before you begin, it is assumed that you have installed the tutorial at C:\arcgis\ArcTutor\Raster\Data. If not, make the appropriate path changes throughout this tutorial so it works properly for you.

The data provided in the Landsat_p114r75 folder comes from Landsat.org/Tropical Rain Forest Information Center, a member of NASA's Federation for Earth Science Information Partners (ESIP) at Michigan State University (http://www.landsat.org/).

Start ArcMap

  1. Start ArcMap by clicking Start > All Programs > ArcGIS > ArcMap 10.1.
  2. Click Cancel on the ArcMap - Getting Started window.

    This window may not open if you've previously opted not to show it.


    You will be setting the default map document's geodatabase later in this exercise.

Create a file geodatabase

If you have already created ImageGDB in another exercise, you can skip this section of steps.

  1. Click the Catalog button Catalog on the Standard toolbar.

    This opens the Catalog window.

  2. In the Location text box, type C:\arcgis\ArcTutor\Raster and press ENTER.

    This location is added to the Catalog tree under the Folders Connection heading.

    If your tutorial data was installed in a different location, alter the path according to your installation location.

  3. Right-click the Raster folder and click New > Folder.
  4. Name the folder Exercises.
  5. Right-click the Exercises folder and click New > File Geodatabase.
  6. Rename the new file geodatabase ImageGDB.

Set the default geodatabase

Each map document has a default geodatabase, which is the home location for the spatial content of your map. This location is used for adding datasets and saving resulting datasets created by various editing and geoprocessing operations.

Learn about the default geodatabase

  1. Right-click the ImageGDB geodatabase in the Catalog window and click Make Default Geodatabase.

Create a new mosaic dataset

Create a mosaic dataset and specify the Landsat product definition. This will create a six-band mosaic dataset, to contain red, green, blue, and infrared wavelength bands.

  1. Right-click the ImageGDB in the Catalog window and click New > Mosaic Dataset.

    This opens the Create Mosaic Dataset tool dialog box.

  2. Type Multispectral in the Mosaic Dataset Name text box.
  3. Click the Coordinate System browse button Spatial Reference Properties.
  4. Expand Projected Coordinate Systems > UTM > WGS 1984 > Southern Hemisphere, click WGS 1984 UTM Zone 50S.prj, then click OK.
  5. Click the Product Definition drop-down arrow and click LANDSAT_6BANDS.
  6. Expand the Product Properties.

    You will see six bands defined and the wavelength ranges associated with each band. When adding satellite imagery, the raster type knows the wavelength ranges associated with each band. It will make sure that the correct wavelengths are loaded into the correct bands. ArcGIS also understands the product definition and will understand which band combination to use when displaying the mosaic dataset.

    Landsat (6 bands) wavelengths

  7. Click OK to run the tool.

    The Multispectral mosaic dataset is created in the geodatabase and added to the table of contents. This is an empty mosaic dataset designed to contain six bands. It is added as a mosaic layer, which is a special group layer. The top level has the name of the mosaic dataset: Multispectral. There are also empty Boundary, Footprint, and Image layers. You will add the Landsat imagery to it in the next steps.

Add rasters to the Multispectral mosaic dataset

  1. Right-click the Multispectral mosaic dataset in the Catalog window and click Add Rasters.

    This opens the Add Rasters To Mosaic Dataset tool dialog box.

  2. In the Raster Type list, choose Landsat 7 ETM+.
  3. Click the Properties button Properties.

    The Raster Type Properties dialog box opens. Here you can enter information to define how the data will be added and processed.

  4. Click the General tab.
  5. Verify the Processing Templates is Multispectral.

    The processing templates define how the data will be added to the mosaic dataset; for example, the template you select creates a multiband image.

  6. Click the Properties tab.

    Here you can see the Percent Clip is the stretch that will be applied to the data and the inputs it will use.

    You can modify these; however, for this exercise you will use most of the defaults.

  7. Click the Functions tab.

    This tab displays the functions that are applied based on the processing template that you chose on the General tab. You chose the Multispectral template.

    Multispectral processing template

    The functions are applied to the imagery from the bottom of the chain to the top. You can see that the Multispectral template creates an MS (multispectral) image, then only applies the Stretch function to the imagery to improve its appearance.

  8. Right-click Stretch Function under the Multispectral template and click Remove.

    This is removed and replaced by the default Identity function. You do this because you want to use this mosaic dataset in analysis, and you don't want the pixel values altered. The Stretch Function helps to display the image but also alters the pixel values.

  9. Click the General tab.
  10. Click Save As.
  11. Navigate to the Exercises folder, type Landsat_NoStretch in the File name text box, then click Save.

    By doing this, you have saved the change you made to the raster type. Therefore, if you were to add more data to the mosaic dataset you can load this raster type file to be sure to apply the same settings.

  12. Click OK.
  13. Verify the Input is Workspace.

    You only have one image scene to add, so you could point to the single file needed; however, if you're adding multiple images organized into a folder, you want to use the Workspace option.

  14. Click the Input browse button Browse.
  15. Navigate to C:\arcgis\ArcTutor\Raster\Data, click Landsat_p114r75, then click Add.
  16. Check Update Overviews.
  17. Click Advanced Options to expand the list of parameters.
  18. Check Build Raster Pyramids.
  19. Check Calculate Statistics.

    It is generally recommended that you build pyramids and calculate statistics on the data that is added to a mosaic dataset. Statistics improve the display, and pyramids reduce the number of overviews that are generated. By checking these check boxes, the operations are run prior to calculating the cell sizes or overviews. If pyramids or statistics exist for a raster dataset, they are not regenerated.

  20. Check Build Thumbnails.
  21. Type 6 band Landsat imagery for analysis in the Operation Description text box.

    This parameter allows you to add descriptive text to the log that is maintained by the mosaic dataset.

  22. Click OK to run the tool.

    The Landsat 7 ETM+ satellite image is added to the mosaic dataset. The footprints are created for each raster dataset, and the boundary is generated for the entire mosaic dataset. The overviews are then generated for the entire mosaic dataset.

  23. You may need to click the Full Extent button Full Extent to view the mosaic dataset.

Examine the tables

There are four main tables that are maintained with all mosaic datasets: the Attribute table, Logs table, Raster Type table, and Levels table.

  1. Right-click Multispectral in the table of contents and click Open > Raster Type Table.

    There are two rows in this table. The first row (OBJECTID = 1) identifies the Landsat 7 ETM+ raster type that you used to add the data to the mosaic dataset. The operation description you added is displayed in the Description column. The second row (OBJECTID = 2) identifies that there are overviews generated for this mosaic dataset and the time at which they were created.

  2. Right-click Multispectral in the table of contents and click Open > Attribute Table.

    You should see three rows in the table. The name of the rasters or their source datasets are listed in the Name column. If you scroll across the table, you will see the Tag column. The MS tag represents the multispectral image.

    Attribute table
    MinPS and MaxPS indicate the pixel resolutions where the images will be displayed. There are only two overview images. These can be identified by the Ov_ prefix in their name and by looking at the Category field. When you view the image at its full extent, the bottom overview image is displayed, and as you zoom in, requiring a higher pixel size, you see the image above it in the table. Next you see the source imagery identified by the MS tag.

    If you continue scrolling across the table you will see some fields with information specific to the satellite imagery, such as the acquisition date. These fields are populated from images and will be different for each dataset that is added. There is no data for the overviews since they don't represent the source images and are often generated from more than one image.

  3. Close the table window.

Examine the image

One reason for creating a mosaic dataset containing the raw source images, without applying a stretch or pan sharpening, is to be able to take advantage of the information in the pixel values, either by a pixel analysis or just using a visual analysis.

Imagery is alway exciting because of the timely information it provides. There is an ongoing fire in this image which you are going to examine using a visual analysis.

  1. Click the Go To XY button Go To XY on the Tools toolbar.
  2. Click the Units button Choose Units and click Meters.
  3. Type 482600 in the X text box and 7657168 in the Y text box.
  4. Click the Zoom To button on the Go To XY dialog box.
    Go To XY dialog box
  5. Right-click the Image layer in the table of contents and click Properties.
  6. Click the Symbology tab.

    You will enhance this image to make the fire features more visible.

  7. Click the Stretch Type drop-down arrow and click Percent Clip.
  8. In the min text box type 0.25, and in the max text box type 1.
  9. Click OK to close the dialog box and apply the stretch.

    In the image you should be able to detect the smoke trail coming from the fire, see a few active flames, and the areas that have already burnt.

    Natural color stretched
  10. By modifying the stretch and band combinations you can make the burn scar and actively burning perimeters more visible. Start by right-clicking the Image layer in the table of contents and clicking Properties.
  11. Click the Symbology tab.

    You will enhance this image to make the burn scars more visible.

  12. Click the drop-down arrows next to each band and set the Red channel to NearInfrared_2, the Green channel to NearInfrared_1, and the Blue channel to Red.
    Band combination

    By removing the blue band the smoke will not be as visible, and by altering the other bands the effects on the soils and vegetation will be enhanced.

  13. Click Apply on the Layer Properties dialog box.
    Enhanced burn scar
  14. Click the Stretch Type drop-down arrow and click Minimum-Maximum.

    This will alter the image to highlight the burning edges.

    Enhanced burning edges
  15. Click OK to close the dialog box.

Recognizing that this kind of information is in an image (due to the pixel values and various wavelength bands) is just one type of analysis. You could continue working with this data to process the image and determine the size of the burn scar. Or if this was only one image in a series over time you could potentially determine the starting location of the fire and its extent and regrowth patterns.

You have completed creating this mosaic dataset. Steps in other exercises include modifying the properties of the mosaic dataset, such as the default mosaic methods, or allowed fields. It is recommended that you examine the mosaic dataset properties before you share the mosaic dataset.

Related Topics