Lab 2: Visualizing and Refining Terrain Survey

INTRODUCTION

In this lab assignment we were tasked with visualizing and refining our terrain survey model. Using the same group of three we had during the first lab (myself, Alyssa Krantz and Evan Geurts) we had to go back into the field and recollect our data points using new and better techniques. Before we went out for the second time, we processed our original data. In order to do this we had to take our spreadsheet of data (see lab 1 post) and reformat it into three columns, X, Y, Z. Once the spreadsheet was reformatted to the proper format we were able to upload the document into our geodatabase. The original data had 156 data points. We then performed 5 different data interpolation methods on the data, including IDW, Natural Neighbor, Kriging, Spline, and create a TIN (definitions below). Once the previous tests were run in ArcMap, we were able to import the files into ArcScene and create a 3D model of our terrain. The image below (FIGURE 1) shows the five interpolation tests we ran, I will go into further detail on each of the tests later in the blog when I am talking about the new data we gathered.

Figure 1: The five different data interpolation methods used on our first data set. Each of the five methods will be explained into greater detail later in the lab write-up.

Our next task after processing our first data set was to go back out into the field and redefine our data gathering methods. On Wednesday September 23, the group met at 6pm behind the shed to gather our materials and head down to our study area, which was the same as lab 1 (FIGURE 2).

Figure 2: The study area for this lab was the same as it was for the first lab, under the University of Wisconsin- Eau Claire footbridge on the bank of the Chippewa River

We decided to use one of the 4 foot by 4 foot surveying boxes that was already assembled. We chose this because in our first attempt we spent well over 30 minutes trying to get the box to be leveled. Behind the shed we took the survey box apart so we could easily put it in the back of a car, and then we drove down to the bank of the Chippewa River under the walking bridge. As soon as we parked the car and got it unloaded the rain came and would not let up until we were about an hour into working. Using wing nuts, washers and bolts we assembled the box (FIGURES 3 AND 4).

Figure 3: Alyssa working on assembling the survey box

Figure 4: The box was held together by basic wing nuts, washers and bolts

Once the box was constructed and properly leveled out (FIGURE 5 AND 6) we were ready to start creating our features. We decided to make our features larger because they were barely visible in our first data set when we imported the files into ArcScene to create a 3D model (FIGURES 7 AND 8). We also believed the size of our grid pattern was at fault too, in order to combat that issue we changed our grid pattern from 10cm x 10cm grid boxes to a smaller size of, 5cm x 5cm.

Figure 5: Finished survey box completely assembled

Figure 6: Once the box was assembled we had to make sure it was level with the ground

Figure 7: The finished product with the features 

Figure 8: Our hill and depression were all in one, simulating a volcano with a creator 

The features included in this survey box were the same of those in lab 1. We needed a ridge, hill, depression, plain and a valley. Like previously mentioned, we wanted to make these sand features far larger than our first try. Once the features were all added to the sand box (FIGURE 7) we were ready to begin the next step of collecting data.

For our new method we decided to still use the string to help make the grid, but instead of making an actual grid, we only went one way and were going to use a ruler which we could easily move from tick mark to tick mark as the other axis (FIGURE 9).

Figure 9: A grid was constructed to collect data, one axis was marked using string and the other was marked using a ruler stretched and taped across

We found that using the ruler to measure of the string down to the surface was the best way we could go about getting an accurate elevation, so we stuck with that technique.

At this time it was getting late, the sun had completely set, and we knew it was going to rain tomorrow. So we had no choice but to keep trying to get our measurements in the darkness of night (FIGURE 10). Using flashlights we had brought with we finished collecting our 506 data points and took everything apart. As soon as we finished taking everything apart and brought the box back to my car the rain came and did not let up again until closer to midnight.

Figure 10: Darkness came quickly and we were forced to use flashlights in order to keep working

NEW DATA

Once dry, we typed up the data into a Microsoft Excel spreadsheet in the three column format (FIGURE 11). Again we used an origin in our northern corner of the survey box. This point was classified as (0,0) and every other point along the X and Y axis was 5cm greater in their respected directions. Again, in order to avoid confusion and having to use equations in Excel, we imported all of our Z values in negative numbers, thus not inverting our data.

Figure 11: All the data points were entered into a 3 column Microsoft Excel Spreadsheet 

Once the excel spreadsheet was finished, we imported the file into our ArcMap geodatabase. Setting the X, Y and Z fields to their respected columns we got our 506 data point grid into ArcMap (FIGURE 12). We then could run our five data interpolation tests on our grid, and eventually create a 3D model.

Figure 12: The 506 data points were entered into ArcMap and were ready to have data interpolation tools ran on the set

Below are the definitions and products of the 5 data interpolation tests we ran on our data,

Definitions from webhelp.esri.com

Inverse Distance Weighted (IDW): A method of interpolation that estimates cell values by averaging the values of sample data points in the neighborhood of each processing cell. The closer a point is to the center of the cell being estimated, the more influence, or weight, it has in the averaging process (Figure 13).

Figure 13: Inverse Distance Weighted data interpolation ran on our set of data. This method uses averages from nearby points in order to fill in the "gaps" between  data points.

Natural Neighbor: Finds the closest subset of input samples to a query point and applies weights to them based on proportionate areas in order to interpolate a value (Sibson, 1981). It is also known as Sibson or “area-stealing” interpolation. Its basic properties are that it's local, using only a subset of samples that surround a query point, and that interpolated heights are guaranteed to be within the range of the samples used (Figure 14).

Figure 14: Using the data interpolation method of Natural Neighbor

Kriging: Kriging assumes that the distance or direction between sample points reflects a spatial correlation that can be used to explain variation in the surface. Kriging fits a mathematical function to a specified number of points, or all points within a specified radius, to determine the output value for each location (Figure 15).

Figure 15: Kriging Data interpolation method

Spline: The Spline method is an interpolation method that estimates values using a mathematical function that minimizes overall surface curvature, resulting in a smooth surface that passes exactly through the input points (Figure 16).

Figure 16: Spline data interpolation method

Triangular Irregular Networks (TIN): are a form of vector based digital geographic data and are constructed by triangulating a set of vertices (points). The vertices are connected with a series of edges to form a network of triangles (Figure 17).

Figure 17: Triangular Irregular Networks (TIN) model of our survey area

DISCUSSION

 

Of the five maps created, each of them have their own pros and cons. The first map, Figure 13, used the Inverse Distance Weighted (IDW) data interpolation method. This method averages the value for the in between sections of the points. In this particular project I do not see any pros in using the IDW data interpolation method. As for cons, when looking at the image above it made it look like our ridge was not so much one smooth mountain ridge but instead made it look like it had several peaks. All of the features had a jagged/pointy aspect to it.

The second data interpolation method used, was Natural Neighbor. Natural Neighbor data interpolation is a good tool to use if we are working with smooth data. Natural Neighbor passes through all of our input data points, but the space between the points is treated as a smooth section. Unfortunately, several of our features lost their sharpness to this method. Natural Neighbor worked well on the plain section because we did not have any abrupt changes to the landscape.

The third method I used was Kriging data interpolation. This is my personal favorite one to use. Kriging uses far advanced math equations that I would not begin to know how to explain, but to sum it up, Kriging uses the surrounding points to spatially relate the data points to fill in the gaps. This results in a much smoother, realistic looking land formation. Kriging was the first data interpolation method to be able to slightly pick up our depression we built in the center of our hill.

The fourth method used was Spline. Spline data interpolation works similar to Natural Neighbor. Just like Natural Neighbor, Spline uses the exact data points and then uses mathematical equations to minimize curvature of the features. For the purpose of this lab, minimizing the curvature of the features will create an unrealistic looking landscape, and much like Figure 16 shows, our land features are very pointy and fake looking.

The final method used was Triangular Irregular Networks or TIN. In my opinion, TIN is the best method to use to show the features, and based upon the output, when compared to the rest of the images, the TIN one looks most like the actual sand box. TIN creates a series of triangles from the data points in order to create the land scape. Although this method lacks in the real smooth look as say the Kriging had, it makes up for it in being able to show precise features of the landscape.
 

Overall, this project touched on multiple beginning aspects of surveying and critical thinking. Other areas this lab introduced us to/jog our memories on, was using ArcMap and ArcScene to create 3D models of landscapes. We were able to recollect data in a better fashion and thus create a better 3D map of said data.

During our time working on lab 2 we did come across a couple issues. The first being the three of us all have such busy schedules, it was difficult to find a time we all could meet and work together on the data collection. The one time we could meet, which was when we did this caused many more issues. We got about half way through setting up our grid when it became way too dark to be able to see anything, luckily we brought flashlights with. Then while collecting the rain started, the rain caused our grid strings to become loose and also made it so our tape would not stick any more. Our last major issue we had was using the flashlights on our sand landscape. The shadow the light would create on the landscape made it very difficult to see where the ruler was and if it was touching the ground or not.

CONCLUSION

Although we came across several issues with losing light and Mother Nature, the overall project was a success. We achieved our goals of gathering a new survey data set, taking that data and turning it into an excel file, importing the file into ArcMap, running data interpolation tools on our collected data and lastly creating a 3D model of the data.

Hopefully next time we have to go in the field to collect data it won’t be rainy or pitch black out. 

SOURCES

webhelp.esri.com