Lab 8

Goal and Background

          The purpose of Lab 8 was to learn the prerequisites for correct image classification, that is, to gain experience in measuring and interpreting the spectral signatures of various Earth surfaces an materials. Specifically, spectral signatures will be collected, graphed, and analyzed whether or not they have spectral separability.

Methods

          In Erdas Imagine, bring in the 2000 Landsat ETM+ of Eau Claire, WI into the viewer. Spectral signatures are going to be collected from twelve different materials.

1. Standing Water (Figure 1)

2. Moving Water (Figure 2)

3. Deciduous Forest (Figure 3)

4. Evergreen Forest (Figure 4)

5. Riparian Vegetation (Figure 5)

6. Crops (Figure 6)

7. Dry Soil (uncultivated) (Figure 7)

8. Moist Soil (uncultivated) (Figure 8)

9. Rock (Figure 9)

10. Asphalt Highway (Figure 10)

11. Airport Runway (Figure 11)

12. Concrete Surface (Figure 12)

          In the Home tab, select Drawing to activate the area of interest tools. Select the polygon tool. In the viewer, digitize an area within Lake Wissota, double clicking to stop digitizing. Then in the Raster tab, select Supervised group and then Signature Editor to open the Signature Editor window. 

In the Signature Editor window, select Create New Signature(s) From AOI. This will add the area that was just digitized into the editor labeled Class 1. Rename it as Standing Water to better represent what signature was collected. Select Display Mean Plot Window to display the spectral plot of the Standing Water signature (Figure 1). Collect the spectral signatures for the rest of the materials. Link viewer to Google Earth to identify features more easily in the image. Scale Chart to Fit Current Signatures in the Signature Mean Plot Window to view the entire signature for some. After all the signatures are collected, in the Signature Mean Plot window, select Switch Between Single and Multiple Signature Mode, which will show all twelve signatures collected within the same window (Figure 13).

          In the second part of Lab 8, simple band ratios were performed to monitor the health of the vegetation and soil. For vegetation health, the Normalized Difference Vegetation Index (NDVI) was performed, using this ratio: NDVI = ((NIR-Red)/(NIR+Red)). Bring in the 200 Eau Claire and Chippewa counties image into the viewer. In the Raster tab, select the Unsupervised group and then the NDVI tool. This will open the indices window. Make sure that the image in the viewer is in the input. Send the output to a personal folder. Make sure the Sensor reads Landsat 7 Multispectral. Highlight NDVI in Select Function. Run the model and create a map of the output in ArcMap (Figure 14).

          For soil health, the Ferrous Mineral Ratio will be implemented to monitor the distribution of iron content, using this ratio Ferrous Mineral = MIR/NIR. With the original 2000 Eau Claire/Chippewa image in the viewer, select the Unsupervised group and then the Indices tool within the Raster tab. Send the output to a personal folder. Again, make sure that the sensor reads Landsat 7 Multispectral. Under Select Function, highlight Ferrous Minerals. Run the model an create a map of the output in ArcMap (Figure 15).

Results

          Figure 1-12 show the Signature Mean Plot of all the materials digitized in Erdas Imagine. Figure 13 shows all the signatures in one window, where the spectral differences between all the signatures are more evident. Band 6 appears to display the most spectral differences amongst the signatures. However, the two water signatures would need to be viewed in Band 5 in order to distinguish them, according to Figure 13.

Figure 1: Mean Spectral Signature of standing water. 

Figure 2: Signature Mean Plot for Moving Water.

Figure 3: Signature Mean Plot of Deciduous Forest.

Figure 4: Signature Mean Plot of Evergreen Forest. 

Figure 5: Signature Mean Plot of Riparian Forest.

Figure 6: Signature Mean Plot of Crop.

Figure 7: Signature Mean Plot of Dry Soil.

Figure 8: Signature Mean Plot of Moist Soil.

Figure 9: Signature Mean Plot of Rock.

Figure 10: Signature Mean Plot of Asphalt Highway.

Figure 11: Signature Mean Plot of Airport Runway.

Figure 12: Signature Mean Plot of Concrete Surface.

Figure 13: Signature Mean Plot for all collected signatures.

           Figure 14 shows a map of the vegetation presence in the Eau Claire and Chippewa counties. The values range from water (is not vegetation) to high vegetation. 

Figure 14: Map of the vegetation presence in the Eau Claire and Chippewa counties, generated by NDVI function.

         Figure 14 shows a map of the ferrous mineral content in the Eau Claire and Chippewa counties. The values range from vegetation (ferrous minerals were nor calculated) to high ferrous minerals. High ferrous mineral content appears concentrated to crop fields and low ferrous content along rivers and bodies of water.

Figure 15: Map of the ferrous mineral presence in the Eau Claire and Chippewa counties, generated by Ferrous Minerals function

Sources

Earth Resources Observation and Science Center, United States Geological Survey (2000) Satellite Images. Reston, VA.