Spectral Signature Analysis and Resource Monitoring

Goals and Background

           The main goal of this lab is to gain experience in the measurement and interpretation of spectral reflectance of a number of earth surfaces. From there, basic monitoring of earth resources using different bands will be completed. Erdas Imagine will be used to  collect, graph and analyze the spectral signatures of the earth surfaces using there spectral signatures. Additionally, the health of vegetation and soil will be explored by using a simple band ratio technique.

Methods

         A personal folder was created prior to the start of lab 8, in order to ensure that the data was saved in the correct spot throughout the lab. Part 1 involved spectral signature analysis. A Landsat ETM+ was used that covered the Eau Claire Area and other regions around Eau Claire to analyze the spectral signatures of certain earth surface features and near surface features. The images used were from the year 2000. The spectral reflectance of 12 materials and man made surfaces will be looked at.

1. Standing Water
2. Moving Water
3. Deciduous forest
4. Evergreen Forest
5. Riparian Vegetation
6. Crops
7. Dry Soil
8. Moist Soil
9. Rock
10. Asphalt Highway
11. Airport Runway
12. Concrete Surface

          Erdas was opened and the image of Eau Claire and surrounding areas was opened. From there under the home tab, the drawing tool was selected and the polygon tool from inside the drawing tool was selected. A polygon was made on Lake Wissota to get the spectral reflectance of standing water. Next, by clicking on raster and then supervised and signature editor, the signature editor is opened. The class one label was switched to standing water, and the rest were changed to there correct label as they were completed. By clicking on the display mean plot window at the top, the user can see what the spectral plot looks like for the area inside the polygon that was made on Lake Wissota. The same process for collecting the spectral signatures for surface 1 is used for surfaces 2 through 12. All of the signatures are going to be displayed in one signature mean plot window, in an effort to get a good idea of how the features vary. The scale chart to fit current signatures button can be used to fit all the different signatures in one frame. The chart background color was changed to white in an effort to increase visibility on the graph, from there the values were recorded and accessed.

        Part 2 involves resource monitoring. This is done by performing a band ratio to view the health of vegetation and soils. The band ratio is normalized by the normalized difference vegetation index (NDVI), and the equation is shown below.

NDVI= (NIR-Red)/(NIR+Red)

       With a fresh Erdas opened, a viewer was added and the second image provided in the lab was added. From there, by clicking on raster, then unsupervised, then NDVI, the indices interface will be opened. A folder name (NVDI) was added to the lab 8 folder in the personal drive and the output was saved there. Be sure the sensor reads 'landsat 7 multispectral,' and that the function is the NDVI. After completion the image is viewed and the vegetation is documented. A map was generated from this tool and it is displayed below by figure 6.

       Section 2 of Part 2 is to perform the same steps as listed above for section 1 of part 2, though the goal now is to monitor the spatial distribution of iron contents in soils with Eau Claire and Chippewa counties. The equation used is displayed below.

Ferrous Mineral= (MIR)/(NIR)

       Be sure to change the select function to ferrous minerals this time so the tool outputs the target variable. A map of the ferrous minerals is displayed below by figure 5. The band values are displayed below for further discussion in the results section.

Band 1 (Blue): 0.45-0.52

Band 2 (Green): 0.52-0.60

Band 3 (Red): 0.63- 0.69

Band 4 (NIR): 0.77-0.90

Band 5 (Short-Wave Infrared): 1.55-1.75

Band 6: (Thermal Infrared): 10.40-12.50

Results

        Figure 1 below shows the first step of the lab, it shows how a polygon was drawn on standing water, and the table to the left shows how the values were displayed. The signature mean plot for the standing water is displayed on the right.

Figure 1

            Figure 2 below shows the signature editor in use in Erdas. It displays the values for the red, green and blue bands.

Figure 2

         Figure 3 below displays the signature mean plot of each of the surface and near surface features in the lab. This is an effective display because it shows the values all next to each other which makes the results meaningful because it is easy to note the differences.

Figure 3

         Figure 4 below displays the distribution of ferrous minerals in Eau Claire and Chippewa Counties. The distribution of more minerals is certainly centered to the west. This would make sense because there is less tree cover, and soil that has been more eroded and had time to form minerals. 

Figure 4

         Figure 5 is a map that displays the areas of heavy vegetation in the areas of Eau Claire and Chippewa Counties. The entire eastern side is covered in thick vegetation, mainly in the northeast corner. Whereas in a line to the southwest of lake Wissota there is a good amount of land that does not have vegetation on it.

Figure 5

Sources

Satellite image is from Earth Resources Observation and Science Center, United States Geological Survey.

Photogrammetry

Goal and Background

         The purpose of this lab was to gather an understanding of photogrammetric tasks that can be preformed on aerial photographs and satellite images. Also the skills of how to obtain and interpret spectral reflectance, and to also understand why certain earth features react the way they do when taken in an aerial photograph. Additionally, the mathematics behind photographic scales are explored, including area and perimeter. Finally, the complex process of performing orthorectificaiton on satellite images was done, which is a very applicable skill to have, as it ensure the images are geospatially accurate.

Methods

        Before beginning lab 7, a specific lab 7 folder was created inside the Q drive to ensure that all data was saved in the same spot and so the professor can go through and view all the work done throughout. The first step of the lab was to determine the distance between two points that would be used to calculate the scale of the image. The calculations are discussed in the next sentence. 1. The distance is  2.7 inches on the monitor, 8824.47 feet in real life. 2.7 inches/ 8824.47 feet x 12 inches. 2.7 inches equals 104,869.64 inches and one inch equals 39,210.9778 inches. The scale is 1:40,000.

       Next, the altitude of where an image was taken was given and the goal was to determine the scale, the elevation of Eau Claire county was also given to add in the calculations and they are shown below. 

            S= (f)/(H/h)

f= 152mm

H= 20,000 feet

h= 796

S= (152mm)/(20,000ft-796ft)

S= (5.98in)/ (240,000in – 9552 in)

S= (1 inch)/( 38536.45 in)

Scale is 1:39000

        The next step of the lab was to digitize an area around a pond to find the size of it. The measure perimeters and areas tool was used to draw the polygon and the area was then calculated. The pond was 93.67 acres. 

       The relief displacement of a smoke stack in an aerial image is going to be determined in the next part of the lab. The displacement between the principal point and the top of the stack is .354 inches. The tower should be moved .354 inches towards the principal point in order to account for this discrepancy. 

       The next part of the lab sought to generate a three dimensional image using an elevation model. The goal is to evaluate relief displacement and how it affects aerial images. The previous steps done were completed in order to be able to do this. The anaglyph tool was used in Erdas to made an image that could be viewed with Polaroid glasses. The vertical exaggeration was set to 1 and the rest of the defaults were accepted. The model was ran and saved in the lab 7 folder and the results were viewed in Erdas. The elevation features of Eau Claire are clearly evident after running this tool. 1. The elevation features in Eau Claire are very prominent and they show up very clearly. For example the hill on the UW campus and near Mt. Simon. 

        The third and final part of the lab involved the orthorectification of satellite images, this process is very lengthy and it was over 3/4 of the lab and took many hours to complete. The first step of the lab is to create a new project. An image of Palm Spring California was opened in Erdas and the photogrammetry project manager was used to create a new block. The polynomial based pushbroom was used and the SPOT pushbroom was selected. 

        From there the horizontal reference sources were done. The projection chooser was opened and the projection type was changed to UTM, from there the spheroid name was selected as Clarle 1866. The datum name was changed to NAD27(CONUS) and the UTM zone was changed to 11. No changes to the vertical section were needed.

        Next, the GCP's were placed to make sure the images were spatially accurate. The start measurement tool was clicked and the tool was activated. The classic point measurement tool was used and the second image was added. Screen captures of the desired x and y coordinates were provided to ensure data quality. The points were added by clicking on the ortho image that was in the left view, the create point icon was clicked on and the values were very close to the x and y provided. Next the corresponding point was found on the spot_pan image in the right viewer. The create point was used again and the coordinates were checked. The next 9 GCPs were collected the same way and they were saved. 

        Next, the second image was added to the block file and the GCP's were collected for that image. The type and usage were set for each of the control points and the tie points were collected. The point measurement tool was again used and the classic point measurement tool was used. The same process described above was again used to collect GCPs.  

         Then, the automatic tie point collection was completed. This is a necessary process to do before the orthorectification process of the two images in the block. The tie point collection process measures the image coordinate positions of ground points appearing on the overlapping areas of the two images. After that, the images were triangulated and the images were orthorectified by clicking on the start ortho resampling process icon. The DEM for palm springs was used as the DTM source and the output cell size for x and y was changed to 10. The file was saved in the lab 7 folder created a the beginning of the lab and the resampling technique was bilinear interpolation. Also be sure to use the current cell sizes. Once the images are done being orthorectified they can be view to take note of the discrepancies between the images. The images can be viewed by clicking on the plus sign next to the ortho folder, repeat the process with a second viewer. Sync the views and zoom in and then use the swipe tool to see the quality of the output. 

Results

        A stereoscopic image is effective in showing things like elevation of the terrain as well as man made features that are higher than the ground around them. When viewed with the 3D places the building show up very clearly, they almost appear how it would look in reality. In comparison the 2D images that are standard do not show the elevations of the features.

Figure 1

      
         The resolution is definitely better in the right image of figure one compared to left image. The left image shows the elevations much better, it is harder to notice elevation changes of the landscape on the right image. Though the right image depicts the heights of the buildings and other structures much better. This could be because it appears that the right image was taken close NADIR, which means the camera was nearly perpendicular with the area of interest. This can be noted when looking at the smoke stack to the west of towers halls. 

Figure 2

                The distance between A an B is 2.7 inches on the monitor, 8824.47 feet in real life. 2.7 inches/ 8824.47 feet (x) 12 inches. 2.7 inches equals 104,869.64 inches and one inch equals 39,210.9778 inches. The scale is 1:40,000. It is important to view the image at the size of the screen, if the image is re sized a different size of scale would be interpreted. 

Figure 3

       Figure 3 shows the step when the image is ready for triangulation.

Figure 4

      Figure 4 above shows the process that was used to put in the GCPs, the example show is from when the second image was added. The highlighted row shows how the GCP was added to the second image.

Figure 5

    

           Figure 5 above shows much how the images are not matched up as well as one would hope. The changes in color from the white ridge there are apparent and easily notable. The swipe tool was used in Erdas after linking the views in order to get a good idea of the quality of this output. 

Figure 6

       Figure 6 shows another issue between the images. It is clear that the images are not perfectly aligned. Though if the images are viewed at full extent there is not a noticeable difference between them. 

Figure 7

          Figure 7 above shows what the two images looked like after the orthorectification. Upon zooming in close on where the pictures overlap, they are not that spatially accurate, this is talked about in more detail above and displayed by figures 5 and 6. 1. The degree of accuracy between the two orthorectified images is a little disappointing. There is a sort of stair step effect that can be noted, the images do not overlap perfectly as shown by the image below. The ridge that appears as a white line in the middle of the image shows that they are not spatial accurate. 

       After completion of this lab, there is a number of photogrammetric tasks that could be duplicated. This lab was very strenuous and long and there was a few errors that occurred throughout, aside from the data corruption issues that occurred this was a very beneficial lab that has many real world applications.


Sources

National Agriculture Imagery Program (NAIP) images are from United States Department of Agriculture, 2005. 

Digital Elevation Model (DEM) for Eau Claire, WI is from United States Department of

Agriculture Natural Resources Conservation Service, 2010. 

Lidar-derived surface model (DSM) for sections of Eau Claire and Chippewa are from

Eau Claire County and Chippewa County governments respectively. 

Spot satellite images are from Erdas Imagine, 2009. 

Digital elevation model (DEM) for Palm Spring, CA is from Erdas Imagine, 2009.  

National Aerial Photography Program (NAPP) 2 meter images are from Erdas Imagine, 2009.