In the previous lab the class used a sandbox as their study area which was 114 x 114 cm filled with sand. Each groups first task was to create a variety of different types of terrain such as; a ridge, hill, depression, valley, and plain. Our group created a grid system of 19 x 19, where each point is 6 cm apart from one another. Once the grid was created measurements of each unit was taken where the top of the sandbox is the zero elevation level, after the data points were collected they were then transferred onto an excel spreadsheet and ready for use in ArcMap. Data normalization is used to reduce data redundancy and improve data integrity, this skill was very important in this lab because it allowed the class to easily find any errors in the data and fix them. Once the data is normalized each group has collected accurate and usable data, which will allow them to create 3-D models of their landscape. The data points collected show the difference in elevation from point to point, this allowed each group to create their landscape through the use of interpolation. Interpolation is a tool found in ArcMap, this tool predicts the values for cells or the data points collected in a raster. The different interpolation methods used in part II of this activity are; IDW, Kriging, Natural neighbor, Spline, and the creation of a TIN raster.
Methods:
The first step for this project was to create a folder specifically for all of the data collected and maps created. Next is to create a geodatabase within the folder that was just created, after the geodatabase is created the excel file with all of the data is imported into our geodatabase making sure that all of the data entered is set as numeric in excel. The excel file containing all X,Y, and Z data is opened up in ArcMap, then the different interpolation methods are implemented. The first is the IDW (Inverse Distance Weighted) method, this specific method averages all of the data within one cell of the grid system. The closer the data points are to the center of the cell in question the larger weight it has on the averaging process of each individual cell. The Kriging method estimates the surface of the landscape within each individual cell and from a scattered set of points with z-values. This procedure produces a prediction of what the surface looks like from the data collected from the study area, while providing a measure of accuracy for each surface prediction given. Natural neighbor is the next method, this method uses the z-value that was provided for each individual cell, finds the closest other data points neighboring the query point and creates a surface where all areas are smooth except for the locations with input values. The Spline method estimates the input values for each individual cell using a mathematical function that minimizes the curve on the surface. Creating a overall smooth surface that passes exactly through the data points collected making the depiction of the study area very accurate. The final method is the TIN (Triangular Irregular Network) method, this method connects all of the data points by triangulating a set of vertices. This forms a series of triangles that are all connected by these vertices creating a surface that displays elevation. Once all of the each of the interpolation methods was used on ArcMap, the data was then moved onto ArcScene where the image was shown in 3-D format. In order to create a map for each of the interpolation methods, each of the 3-D images needed to be saved as a layer file so then they can be opened on ArcMap and the legends from ArcScene can be used in the final maps. Next each 3-D image was exported as a 2-D JPEG, still displaying the image in 3-D. The orientation used helps each viewer see most if not all different types of terrain represented in the study area. The scale is very important because it allows someone to tell how much elevation is present in a certain area.
Results/Discussion:
The different interpolation methods show multiple advantages and disadvantages with representing the elevation data from the study area. Some of the methods proved to show the data in similar ways, but others were better at representing certain types of terrain. Below is figure 1 which are the results for the IDW interpolation method.
 |
| Figure 1: IDW Interpolation of Sandbox Study Area |
As one can see from the image above the areas that are at the zero elevation level are represented by the color green, while the areas that are above that level are blue and the areas that are below are yellow. The disadvantages to this method are that in the top right corner, the depressions aren't represented to the correct depth each of them smaller then they are supposed to be. There is the same problem with the valley on the left side of the map, it also doesn't represent it do the correct depth, too shallow. What can be taken from this image is that the IDM interpolation method doesn't represent area's below the zero elevation accurately, while the area's above that elevation level are represented accurately. Below is figure 2, representing the Spline interpolation method of the sandbox study area.
 |
| Figure 2: Kriging Interpolation of Sandbox Study Area |
After looking at figure 2 one can see that the Kriging method is probably the worst at representing elevation. The valley and depressions aren't represented correctly when being compared to the study area, they are far too shallow and not deep enough. While the ridge and hill are far too close to the ground and not high enough, once again not representing these features correctly when being compared to their real life features in the study area. Figure 2 is supposed to be representing a 3-D image based on the data collected in this activity but yet it almost looks as if it is still 2-D. The method still allows somebody to be able to see where the features are located because of the colors but it doesn't correctly show the elevation levels that it's supposed to. The map seems to be showing that the valley is the same depth as most of the study area which is not the case because the only other area that should be showing close depth values to the valley would be the depressions. Figure 3 is below, in this figure is the natural neighbors interpolation method.
 |
| Figure 3: Natural Neighbor Interpolation for Sandbox Study Area |
This method for interpolation does an especially good job at representing the valley and depressions, the valley is to the far left of figure 3 and is colored blue. The three depressions are at the top left of figure 3 and are also colored blue. The main issue with using this method is that all levels that are supposedly below the zero elevation level are all listed at the same depth through this method. This can lead to problems, because of this people can misread figure 3 and misunderstand how deep some of the features are supposed to be and how some aren't as deep as they should be. The next method is represented by figure 4, this is the spline interpolation method.
 |
| Figure 4: Spline Interpolation of Sandbox Study Area |
This method is the best at representing the landscape that was created in the sandbox, the spline interpolation does the best job at representing the difference between what is supposed to be below the zero elevation level and what is supposed to be above. Also does a very good job at representing a gradual increase or decrease in the landscape for example in the valley, the walls of the valley are changing color as they go down meaning that the depth is increasing. The final method is represented in figure 5, not really a interpolation method but the Triangular Irregular Network or TIN does a very good job at representing different elevation and depth levels for valleys, depressions, hills, and ridges. While looking at figure 5, somebody can easily see which colors are the highest (white, gray, and brown) and which ones represent the lowest points in the landscape (green, beige, and light blue), but in the end the best at representing the entire landscape as a whole is the spline interpolation method.
 |
| Figure 5: TIN Raster of Sandbox Study Area |
Revisit Survey:
When remaking the survey with the landscape that was created, group 2 went back to the sandbox and created more data points towards the back right of the study area, where the depressions and the ridge are located. In order to do this, the group decided to change their points from 6x6 in that area to 3x6 which will provide more detailed digital imagery of that area. After looking at figure 6 below, one can see that the area where the depressions and ridge are located, there is much more detail then before in figure 4. Not only does it show a more accurate depth for the depressions but more accurate height for the ridge. The colors are continuously changing in those areas now because of the more data points taken to turn that area into a more accurate depiction of the study area.
 |
| Figure 6: Spline Interpolation of Revised Sandbox Study Area |
Conclusion:
This survey is mainly about data normalization, which is very similar to many other field based surveys that have been performed to obtain data. If somebody were to do this activity without normalizing the data it would be very time consuming because of all the extra data points that would needed to be collected. This specific project is different from any other project because in this project instead of mapping an already existing landscape, the groups were able to design their own and map it themselves. It is not always this easy to create such a detailed grid based survey because this project was based on such a small scale. Compared to other projects where someone might have to collect data for a study area spanning hundreds of acres, this would be much harder to create a grid for something that large. Elevation isn't the only type of data that these interpolation methods can be used on, other types of data such as climate or temperatures. For example by creating a grid system over a large area and taking temperature data from each grid unit, a map of the gradual climate change throughout that study area can be made through the use of interpolation.
No comments:
Post a Comment