Lab 6 Profiles of DEMs and change detection by using the DEMs Introduction This lab will introduce you to change detection by subtraction between two images. You will subtract two Digital Elevation Model (DEM) images and then derive volumetric changes between the two images. The four DEMs covering the south side of the Wrightsville Beach, Wilmington, NC (Fig. 1) were acquired by the light detection and ranging (LIDAR) sensor in 1997, 1998, 1999, and 2000. Between 1997 and 1998 data acquisition, a hurricane made landfall in the area, and between 1998 and 1999, another hurricane made landfall. There was no hurricane affecting the study area between 1999 and 2000 data acquisitions. To detect changes of images, all the images must be: (a) georeferenced to a common grid (or coordinate system), and (b) the images must have the same resolution. In this case, four DEMs are geo-referenced and have the same cell resolution. Differencing method is used to derive the difference between two DEM s at a time. Once achieved, the output can further be defined into areas of positive change (deposition) and areas of negative change (erosion). The resulting image can then be analyzed for patterns of erosion and deposition. Fig. 1a, 1998 DOQQ shows the Wrightsville Beach, NC. Fig. 1b, Study area is shown in details. View the DEM images Open four viewers and display layer 1 (1997), layer 2 (1998), layer3 (1999), and layer 4 (2000) in gray scale. You may need to resize the viewer windows to view the entire study area. Click on the i icon to answer questions 1 and 2. 1
Spatial and surface profiles of a DEM A spatial profile shows a cross section of the DEM along a transect or line. A surface profile shows the 3-D view of the DEM. Both profiles help understand a DEM. To get a spatial profile of a displayed DEM, i) Display a DEM layer (you can use 1997, 1998, 1999, or 2000 DEM). ii) In the top menu bar, find Panchromatic/Spatial Profile/Spatial Profile to open the profile window (Fig. 2). Within the window, the x-axis is the line for the transect, and y-axis shows the height along the transect. Fig. 2. iii) Click on the "Create a new profile line icon to activate it (Fig. 2), move back (into) the viewer window, pressmove mouse button to draw a line over the image, and double-click the mouse button to end the line. A transect is shown in the viewer window (Fig. 3), and height profile alone the line is shown in the spatial profile window (Fig. 2). Moving the line within the viewer window, different height profiles will be created. iv) Once done, click on the "File/Close" or x at the upper right corner (Fig. 2). Fig. 3. 2
To get a surface profile i) Display a DEM layer (you can use 1997, 1998, 1999, or 2000 DEM) if you don t have one. ii) In the top menu bar, find Panchromatic/Spatial Profile/Surface Profile to open the profile window (Fig. 4). iii) iv) Click on the "Create Box in Viewer" icon, move into the viewer window where the DEM is displayed, and press-move mouse button to draw a box (Fig. 5); the surface (height) profile of the boxed area are displayed (Fig. 4). Moving and/or resizing the box within the viewer window allow you to view surface profiles at different locations. Once done, click on the "File/Close". Fig. 4. Fig. 5. Change Detection by Spatial Modeler The spatial modeler in the IMAGINE allows you to create and execute many operations. For this lab, you will need to model the difference of the two DEM data at a time. Click on the "Toolbox/Model Maker " to open a New_Model window (Fig. 6), and its tool palette (Fig. 7). Use the tools to create the input and output raster layers, function, and link the graphics together. (Note: depending on the order of the creation of the graphics, you may have different order or "n".) Also, you need to navigate the spatial modeler to the folder where you get your input data, and where you output your differenced image. i) To add the input data: double-click on the raster layer icon (Fig. 6), and load 4layers_dems_sout1.img as an input image. Again, layers 1, 2, 3, and 4 are DEMs in 1997, 1998, 1999, and 2000, respectively. ii) To define the difference function: In the function definition box, click on the 4layers_dems_south1[2].img (the current data) from the input list, click on the - (minus) from operator and number panel, and then click on the 4layers_dems_south1[1].img (the previous data) (Fig. 8). Since you use the recent data minus the previous data, the result will be a) if the different is negative at a location or pixel, there will be erosion at that location or pixel, i.e., 3
there were more sands in the previous time than the current time at that location or pixel, b) if the difference is positive at a location or pixel, there will be deposition at that location or pixel. iii) To output the differenced data: Name the output image as 98_97_diff.img (Fig. 9). Make sure to select the "Delete If Exists" and select the data type as the Float Single. Click on OK to finish (Fig. 9). iv) Save the model as 98_97_diff.gmd for future use. Run your model by clicking the "Run" (lighting) icon (Fig. 6). Fig. 7. Fig. 6. Fig. 8. 4
Fig. 9. Open up another Viewer window and display the difference image 98_97_diff.img that shows erosion and deposition within the study area. Make sure you click on the "Pseudo Color" options so that you can change the colors of pixels later (Fig. 10). Now let's determine where the erosion and deposition occurred. In the viewer window (Fig. 10) showing 98_97_diff.img, right-click on the "98_97_diff.img" and select Display Attribute Table to open the attributes (Fig. 11). The table detailed the height changes at each pixels sorted from the lowest value to highest value. The value is in meters. Scroll down and/or up the table and look at the values, negative = erosion and positive = deposition. Fig. 10 5
To select all negative values: 1) click on the 0 under column Row, and ii) scroll down to the last negative value, shift-middle-click to select all the negative values. (The middle is the wheel of your mouse.) Then move the cursor into the cell under column Color. Right-click on the color, select a color (e.g., red) and release the mouse button (Fig. 11). Repeat this process for all the positive values. Since we did not exclude the ocean from the DEMs, a value nearest to zero (either positive or negative) can be used to represent ocean. In my example, the value is -0.0248516, and I color-coded it as blue (for ocean). (One can exclude the ocean by using an AOI; we skip that in this lab.) Use the green color to highlight the areas where deposition occurred (Fig. 11). Now, in the viewer window, deposition occurs at green colored areas, and erosion at red colored areas. Fig. 11. 6
Compute the volumes of deposition and erosion There are at least two ways you can compute the volumes of deposition and erosion. i) You can use the volumetric computation functions in the Imagine s/w, which is a little bit complicated. ii) You can use the MS/Excel to compute them. a) Copy and paste the data into the Excel Click on Value to select the value column and shift-click on Histogram to select the 2 nd column. Then right-click to open a drop down menu and select Edit/Copy or click on the copy icon directly. Move your cursor into an opened MS/Excel file. Within cell A1, right-click; you have copied and pasted the data into the Excel. b) Compute the volumes of erosion and deposition To compute the volume per row in the 3 rd column: (1 st column) * (2 nd column) * (pixel size x) * (pixel size y). Once you finish the 1 st row, you need to drag your calculation to the end of the file to finish the computations for all rows of the 3 rd column. To compute the total erosion: add all the negative values of the 3 rd column. To compute the total deposition: add all the positive values in the 3 rd column. Answer question 3. Questions 1. (0.5 point) What are the pixel sizes of x and y? 2. (1.5 points) What are the minimums, medians, maximums, modes, means, and standard deviations of the 1997, 1998, 1999, and 2000 DEM data, respectively? 3. (1 point) For DEM difference 1998 1997, what are the total volumes of erosion and deposition for the site? If the total volume of deposition or erosion is used as an indicator for beach deposition or erosion, then which one, deposition or erosion is the dominant process? 4. (2 points) Two difference images of DEMs (1998 1997 required, and 1999 1998 or 2000 1999). Acknowledgment This lab handout was supported by an AmericaView grant to the East Carolina University at Greenville, North Carolina, 27858, USA. 7