Network Analysis

Goal and Objectives: 

The goal of this assignment was to lean how to use and perform network analysis. I used model builder to automate the process for us.  Using model builder helped my understanding to show the flow of work that goes into network analysis and what steps need to happen at what time in the process to get to the end goal.  I applied network analysis to the determine the cost of frac sand mining across counties by the increased truck traffic from moving the sand from mine to rail terminal.  This method helped to understand the impact of frac sand mines that is not just the impact of the mine itself, but the entire county.

Methods:

I had to prepare the feature classes to use the network analysis tool. I am using the mine data that I revived from the Wisconsin DNR.  We have been using this data for a few weeks now.  Some of the mines are not currently active therefore not all of the mines are used.  The mines that I am looking at ship their sand from the trucks to the rail terminals. Using a python script helped too speed up the process which selected all of the active mines without rail loading stations within 1.5km of railway.  I was left with 44 mines.   

To start the second part of this exercise I was given a geodatabase that had rail terminals in Wisconsin and had to use these for my analysis. I also used a street network data set from ESRI street map which was used during the network analysis to learn how to map routes and find the closest facility using incidences and facilities. 

Figure 1: This is a model showing how I came to the conclusion of the graphs below.

Once I was finished with that steep I was to use model builder and create a flow chart (Figure 1). This flow chart was the tool to figure out the distance and cost associated with the roads due to the distance of the trucks going from the sand mines to the rail terminals and back.  First i had to use the Make Closest Facility Layer tool and the add location tool.  I used the mines as the incidents and the rail terminals as the facilities which allowed me to set up my network analysis.  I then had to solve by routes to determine the closest station to each mine. Then, I had to use the select data  and Copy Features tools to create a feature class from the routes that I solved for from the network analysis. I then had to use the Project tool to change it to NAD 1983 HARN Transverse Mercator feet to get the correct numbers to use in my calculations. I then had to use the Summary Statistics tool which allowed me to create a table with the route distances broken down by county.  I then had to create two different fields and use two different calculations.  The first field I had to create was miles from the total shape length.  I then had to create a field called cost for the amount it cost to repair the roads.  The two equations I used were multiplying the foot distance by 0.00018939 to get meters.  I then took the meters number and multiplied it by 2*50*[total_Miles]*0.22.  The numbers for cost and trips are hypothetical.

Figure 2: Final Map showing the closest routes from mine to rail terminal along with the cost of maintenance per county.

Results:

The results I got from the data from the first part can be shown below.  This shows the different routes and gives a better understanding of the model we previously used.  

Figure 3: This graph is showing the distance  trucks would need to drive within each county to get from the mine to the terminal.

Figure 4: This graph is showing the annual cost each county will have to spend on its roads due to the heavier traffic from the trucks delivering sand to the terminals.

Discussion:

I was actually very surprised by the price of maintenance.  I thought it was going to be much higher than I arrived at, but this may be due to things not being calculated.  I also feel that there area some other variables that may happen while the trucks are on the road that were not accounted for, for example, break downs, accidents, and extra trips.  To change any of the numbers we could also start to add the facilities outside of Wisconsin. 

One problem I thought of is that are there any counties that do not allow the trucks to drive on certain roads. I feel like this is something that should be looked into to get a better  detailed and more accurate representation, because I know that the trucks may not be allowed to always go the fastest way if its through a neighborhood or something of the like.  

Conclusion:

I have found that the network analysis tool is one of my favorite tools to use.  I could see myself using this tool in a future job, and I feel like it has some of the largest uses.  Its not just to find the fastest route, but to determine where things can be put or which way to go.  Being able to figure out the price for maintenance on the roads is beneficial, and the figure could be brought to a meeting about opening up a sand mine in a different county determining its affects on the surrounding roads.  I feel like this would be great for the people to see who live in the surrounding area.  

Geocoding Frac Sand Mines

Goals and Objectives:

The goal of this lab was to geocode a set of data that was given to us by the DNR.  This data was not formatted correctly, making it tedious to format each one. We used this data to gain the understanding of how normalization is key to providing data. It also shows the importance of accuracy of data.  

Methods:

Before anything could be done in ArcMap I had to normalize the data that was given to me (Figure 1).   I had to add new columns to the data to separate out different parts from different columns to make it work with the geocoder. By doing this, I was able to successfully able to use the geocoder for a few of the addresses. Sadly, most of mine were only in Public Land Survey System (PLSS) notation making it difficult to use the geocoder, therefore I had to manually locate the mines myself by using the description given to me in the address on the excel file. 

Once all of the addresses were normalized (Figure 2) I was able to upload the table into ArcMap and use the geocoding tool. I had to set up the geocoding tool by selecting which columns went to what on the geocoder. For example, on the geocoder addresses went with addresses. Esri's  The geocoder then matched the addresses using the "World Geocoding Service." The addresses did not come out exactly accurate.  This meant that I had to go into each point and assure they were in the correct spot with the sand mine.  I noticed in the once that I was specifically assigned to geocode that a few did not have the sand mine in the ESRI base map that I was using.  I then had to find it on google and locate it back on the base map using the roads around it.  This made it slightly more difficult, but using the PLSS helped to narrow it down even farther. I used the address inspector to unmatch the points and choose the correct point by using the "Pick Address From Map" tool.

Finally, I was then able to compare the addition of my colleagues results, my results and the actual locations of the mines. To do this I was able to use the "merge tool," which we recently learned in a demo, to combine the data together. Next, I used the "Near" tool to calculate the closest distance of my points compared to my colleagues points. 

Results:

When I first went to use the geocoder I had some of the addresses not showing up, put in the middle of the town, and not located anywhere near a street.  This was not the result I was hoping for, but expected it because I had very few with actually addresses and needed find the location of the PLSS to find the sand mines. 

Figure 3: A map showing the Merged Geocoding and the actual mine locations.

I then used the Distance to Point tool, which allowed myself to figure out the distance between my mines locations and my classmates (figure 4).  This was interesting because I thought the numbers were going to be much higher than they actually were, but ended up being much closer.  

Figure 4: The distance results between my mine locations and my classmates.

Discussion: 

There are so many reasons why there is such a different in the distance between the geocoded points.  In a class reading by Lo they are listed out in a table. I do not believe there to be any gross errors in this data because they are all assumed as sand mine locations.  However, there may be systematic errors found within the data because not all of the mines were shown on the base map making it harder to be as accurate as possible.  There even may be some random errors which could occur from just making a mistake or possibly just moving a point that was not supposed to be moved. This lab however was a great lesson in learning how the different errors can really change the outcome of the data; and how important it is to have accurate data when making real life decisions that are based on data that I have made/used.  

The Inherent errors in the data are actually a large source of the errors in the geocodings.  These are errors in digitizing the data. The fact that each dataset was created differently by each of my classmates really throws the data for a loop.  If it was only one person then only that person has to worry about the format, but with a whole class the format is constantly changing. Operation errors were also a large part of this lab because many students may have used the data incorrectly or made a mistake when using the data. 

To figure out which points are correct compared to the ones that are not correct I would want a full list of addresses of each mine, but without this information we could go over the data as a group to figure out which of the points are correct, and which ones are not.  having the latitude and longitudinal data is also very helpful, making it easier to find the location on the map.  

Conclusion:

Being able to normalize data, understand the data, and accurately use the data is extremely beneficial when geocoding. Have a set of rules would beneficial to the data; this would make it easier to use and having the guidelines would allow a smaller chance of error. 

References Cited:

Lo, C., & Yeung, A. (2003). Data Quality and Data Standards. In Concepts and Techniques in Geographic Information Systems (pp. 104-132). Pearson Prentice Hall.