Posts

GIS 5935: Lab 7 - DEMs and TINs

Image
This week started our exploration of terrain and elevation models, mainly TIN (Triangulated Irregular Network) and DEM (Digital Elevation Models). The TIN is a type of vector model to represent a 3D surface, whereas a DEM is a raster dataset.

To begin the lab, we explored the basic differences between DEMs and TINs by both draping a radar image over a TIN and creating a ski run suitability map using a DEM. We then learned how to create TINs and all the different ways you can symbolize and visualize the data.  In part D of the lab, we compared contours created from both TINs and DEMs. Overall, what is unique of the TIN model as opposed to the DEM is that the TIN allows for input of data after the model and triangles are drawn, this is very beneficial in the long run because with a DEM, you would have to start completely over with analysis. An example of this is shown below:
By later adding in a lake to the TIN model, hard edges were created showing more detail and information in the an…

GIS 5935L Lab 6 -- Location Allocation Analysis

Image
This week finished up our final lab in the network series and was about network allocation.

To begin, we went through the ArcGIS tutorial files where we learned about maximizing attendance and targeting market share and other ways network analyst can be used for analysis.

We then went through a scenario pertaining to the "minimize impedance" problem where we adjusted market areas to distribution centers using location allocation.  This is done to optimize or take a systematic approach to dividing distribution center services to each market area.

The analysis involved a lot of joins and manipulation of feature classes within the geodatabase, and it's really interesting how much customization is possible within network analyst. By using the minimize impedance problem, 13 market areas were changed to more efficient or optimized distribution centers. The before and after analysis is shown below as well as which market areas were changed.


GIS5935: Lab 5 -- Vehicle Routing Problem

Image
This week continued the series on networks, and we learned about the vehicle routing problem analysis.  I never realized the capabilities of this analysis and can't wait to put it into use through work!

First we went through the network analyst tutorials where we learned about the different specifications and personalizations you can put into your analysis which can account for anything such as which side of the road to pull up to to allowing for breaks and incorporating the cost of these breaks as well as "specialties" each route or vehicle can accommodate.

For the actual lab portion of the assignment we used the VRP to carry out analysis for a trucking company in South Florida.  We created optimized routes for a day's worth of pickups, while staying true to the company's goal of providing continuity between drivers, customers and service areas.

The new route allowed all orders to be delivered with only 1 with a time violation.  This greatly increases the custom…

GIS 5935: Lab 4 - Building Networks

Image
This week was about building networks and making the travel times more accurate by incorporating a variety of attributes such as traffic, time, one way roads, and restricted turns.

To begin, we started out by building a basic road network only including one way road restrictions. We then created a route through 19 facilities (or stops) the time it took approximately 97 minutes of travel time to complete the entire route.

The second step was adding in restricted turns.  These were turns that were identified by the network builder as being restricted, and when creating a new route, these were restricted in the parameters.  This resulted in approximately 106 total minutes of travel time.  The summary of the route is shown below.
After just incorporating restricted turns, we then used historical traffic data. This also resulted in around 106 total minutes of travel time (summary screenshot shown below).  This did not model turn restrictions, so is in comparison to the first scenario. 

GIS5935: Lab 3 - Road Network Completeness

Image
Building on last week's lab, and even the week before in our data quality section, we learned how to evaluate maps for completeness.

Specifically, we took two road networks (TIGER data from U.S. Census), as well as Jackson County street center line data.  We were to compare these two networks and decide which is likely to be more "complete".

To begin, we first calculated the total length of both road networks. According to this basic analysis, TIGER road length was longer (11382.7km) compared to Jackson County centerlines data (10805.8 km), with this we can assume that TIGER data is more complete.
After this, we went into a little more depth and evaluated the completeness in terms of grids.  To do this, I intersected the road network with the grid for both networks, and calculated the length in km.  I then summarized the grid code field to obtain a summary of the length of the roads based on each grid codes. With some excel manipulations, I was able to determine the perc…

GIS 5935: Lab 2 -- Determining the quality of road networks

Image
This lab was about testing for the accuracy of street maps, and really got into the importance of accuracy and shed light on how reliable different data sources are.

For the lab, we had to determine the accuracy of a city wide road network of Albuquerque NM, as well as street network data from StreetMap USA. To do this, I first used the Sampling Design Tool addon that selected out 100 random junctions from the ABQ dataset.  I went through these test points and found 20 "good" intersections that were legible and had an associated USA StreetMap point that was legible (image below). I made sure these were spread out throughout the map as to ensure little bias. After this, I then went and digitized reference points of where I thought the true location of the intersection was based on aerial photography which was provided.
screen capture of the 20 test points and digitized reference points

After getting the coordinates of each point I was able to use the attached worksheet to calc…

GIS 5935: Lab1 -- Calculating Metrics (Spatial Data Quality)

Image
And so begins the start of the fall semester with special topics in GIS. The first few modules of the class covers spatial data quality, and the very first lab is all about calculating metrics and measuring accuracy and precision.
In the first part of the lab, we were to see how precise and accurate data from 50 points tagged on the same GPS at the exact same location were.  Below is a visual map of the precision with the 50th, 68th, and 95th percentiles. 
The waypoints were in completely different locations which was shocking as they were theoretically supposed to be at the same exact location.  We first calculated an average location and elevation from all the points and determined how precise the data was by how close the measured values were from the observed. In this case, precision was measured using the average and comparing the difference between all of the points.  After this, we took a look at accuracy. In this situation, we measured accuracy by having a true reference poin…