Wednesday, October 18, 2017

Special Topics in GIS Lab 7

This week we worked with TINs and DEMs. For those of you new to elevation models the TIN is a network of triangulated elevation points that are laid out in a grid with slope, aspect, and elevation represented. The DEM is a digital elevation model that basically assigns an elevation to a grid area based on the resolution of the area. This week we practiced creating elevation models in ArcGIS and analyzing them. We started with the TIN and found many ways to adjust the symbology of the data to represent exactly view we need from the product. TINs were fairly easy to work with and visualize especially when converted to a 3D image in ArcScene. Finally working with the DEMs we created a slope analysis for a ski resort and were able to display the areas with ideal slope for medium skill level skiers. This document could show the resort the best places to form the next run and what skill level to label it. Below is a screenshot of the DEM analysis in ArcScene. You can see the categorized slope areas along with their aspect and overall elevation.

Tuesday, October 10, 2017

Special Topics in GIS Lab 6

This week in class we focused on Location-Allocation analysis. It involved determining the best solution based on selected criteria for matching locations with central hubs. A good example is many customers in different locations around the U.S. needing service from package handlers like UPS that has central hubs in different areas. The analysis would provide a solution to which customers should be serviced by which hubs.

This week we focused on a solution that matched customers to a distribution center. After the analysis was run we compared the solution to an overlay of the market areas. Some of the customers were being serviced by distribution centers that were not responsible for their market area. To fix the outliers we simply needed to re-designate some of the market areas. The image below shows the new market areas. It only differs slightly from the original market areas due to the small number of outliers. It is interesting to see the simple analysis that we performed this week. It has so many possible applications and could save companies a lot of money by increasing efficiency.

Wednesday, October 4, 2017

Special Topics in GIS Lab 5

This week in class we learned a lot about solving Vehicle Routing Problems using ArcGIS. I really enjoyed the week because it involved getting down into the settings of the software to really see what the program is capable of. The goal was to analyse data given for a company that needed to deliver orders to different customers in the south Florida region. The orders were located at a central distribution facility and there were 22 trucks and drivers available to deliver. In the beginning we restricted ourselves to 14 trucks to try and save on costs, but it resulted in many orders not getting delivered and some being delivered late. After the addition of two more routes all packages were delivered and only one was slightly late (2 minutes). The addition of the two extra routes greatly improved customer service and increased revenue. An image including the delivery zones and routes surrounding the central distribution center shows where all of the drop-off sites are located.

The system took in all of the information provided by the user and followed strict constraints to produce the solution that would decrease distance and time costs while stopping at the maximum amount of drop-off sites.

Wednesday, September 27, 2017

Special Topics in GIS Lab 4

This week we are working with networks. In the lab we created a network and practiced adjusting functionality to see how it would display certain routes. The basic network was just a set of edges and junctions (roads and intersections). Then we added a restriction layer that would not allow certain turns onto some streets from others or if there were traffic lights. Finally we used speed limit data to determine the speed a vehicle would travel on each road based on the time of day. This speed limit also varied depending on which direction you were driving on the certain road.
In every step of the lab we performed a route analysis. With the basic network the route it simply plotted the shortest distance to get through all of the stops from beginning to end. Then the route was run with the restrictions turned on which increased the total distance and time of the route. This is due to the fact that some turns were not allowed so it had to calculate a different route. Finally with the traffic information provided the route was calculated with data for 8:00am on a Monday morning and returned that it would take 122 minutes and cover 103 kilometeres which was again an increase from the route run with restrictions.

Wednesday, September 20, 2017

Special Topics in GIS Lab 3

The lab for week three involved accuracy statistics of completeness and positional accuracy. The actual lesson and reading focused on point based accuracy analysis versus line based accuracy analysis for line features. Lab two involved our accuracy analysis on the point based approach to analyzing the line features if you want to check that post out. This week the analysis involved the completeness statistic and two layers of roads in southern Oregon. The road layers were placed on a grid system where the layers were intersected to "cookie cut" them. Then the total lengths of the roads were added up in each grid for the two layers. Then a comparison could be made by looking at which layer had more length of road in each grid. This percentage was then displayed in a choropleth map to show the layer comparison.
The overall goal of the accuracy assessment in this lab was to see which layer was more complete by looking at the total lengths of roads in the county and then analyzing them grid by grid. From the product you can see that the TIGER layer seemed to be more complete in the urban areas and the Centerline layer was more complete in the rural areas.

Monday, September 11, 2017

Special Topics in GIS Lab 2

For this lab we looked at the city streets of Albuquerque, New Mexico. The image above is a screenshot of the test points selected for the accuracy analysis. To obtain the accuracy statistics for the layer we had to create a network data set for intersection points within the street layer. Once the points were created a random generator was used to select a number of random points throughout the study area. The selected points were then narrowed down based on the criteria for an appropriate test point. Finally twenty-four points remained. Next the points were compared to more accurate orthophotos. Reference points were placed on the correct or "true" location that the selected points were trying to represent. The distance between these points was used to then calculate the accuracy statistics based on the NSSDA guidelines. The accuracy statement derived for the above layer is:

 Horizontal Positional Accuracy: Tested 243.08 feet horizontal accuracy at 95% confidence level

The depicted layer was found to be the lesser accurate of the two layers tested and did not represent the true locations as well as the other layer.

Wednesday, September 6, 2017

Special Topics In GIS Lab 1

This week it was nice to get back to the grind by working with ArcMap and using the toolbox. We created buffer regions around an average location of fifty GPS points taken by the same device in the same location. The goal was to determine the precision and overall accuracy of the data or device itself. The results can be seen in the following map. I attempted to keep the layout simple and only show the needed data to understand the precision and accuracy aspects.
The point that was made this week in class is that accuracy and precision are two different things. Precision is measured as the distance a point sample is from the average of all of the point samples. The accuracy is measured as the distance the samples are from the reference point or the "True" location. The reference point measured 3.78 m from the average location of the way points taken by the GPS. This shows that the GPS location was fairly accurate. 3.78 meters is about the standard width or length of a bedroom in a normal house. For measurement tools from space that seems fair to call the average way point accurate. The horizontal precision of 4.85 meters seems in line with the accuracy of the average. I would question the results if the horizontal precision was much larger than the horizontal accuracy. Vertically the accuracy is 5.96 meters and the vertical precision is 5.8 meters. Again the numbers are fairly similar and small in scale. In my military experience these numbers would be fine for striking a large building, but if I was looking to strike a vehicle I would have a high probability of missing.