Analytical Data: Module 5

For Module 5 (Analytical Data), we processed various data sets and created infographics using various programs (ArcGIS, Microsoft Word, Microsoft Excel).  This was one of my favorite assignments in this class so far!  We were provided with a data set called the "2018 County Health Ratings National Data", which had various annual statistics.  I chose to compare the two variables, "Percentage of Single Parent Households" and "Percentage of Children in Poverty".  I initially wanted to use the "Violent Crime" category with another variable for this lab but realized that the statistics provided were in "indices" rather than percentages.  I settled on child poverty rates and single parent homes because both had a normalized variable using an overall percentage (per county/state).

With regard to map design, I was struggling with the requirement to put so many things (infographic, bar chart, scatterplot, 2 maps, etc) on the same page, but I think I made it work.  I made many changes once I started to add all of the components to the map.  Some of the original bar charts I created didn't fit well on the page, so I changed the graph orientation. I changed the colors of some of the other graphs so it matched the other graphs as well. One item of note: I located supporting statistics on The Annie B. Casey Foundation website, but could not find the 2018 data set so I used the 2019 statistics instead.

Color Concepts and Choropleth Mapping - Module 4

During Module 4, we learned about color concepts and spent time creating various color ramps manually using step-wise calculations. This was a slow and arduous process to say the least!   We also compared our linear and adjusted progression color ramps with the closest color ramp shown in Color Brewer.  The differences between the linear and adjusted progression maps are quite minor. I found that the Color Brewer color ramp did not seem to have a logical, consistent step pattern, but also had a more divergent color scheme.   

----------------------------------------------------------------------------------------------------------------

Mapping Change with Choropleth Mapping: 

For this map, I chose a six class diverging color ramp from Color Brewer.  I customized my legend in ArcGIS to match the Color Brewer color palette. I created six classes and used the  "Quantile" classification.  

Terrain Visualization: Module 3

 For Module 3, we covered various techniques to visualize terrain data.  We worked with several different data sets this week, specifically, a land cover map depicting tree varieties in Yellowstone National Park. The original data set had several similar tree categories, so I grouped those values into larger categories and relabeled them as Douglas Fir, Lodgepole Pine, and Whitebark Pine.  I utilized the multidirectional hill shade and placed it under my landcover layer.  I set the transparency of the land cover layer to approximately 59%.  After I changed the transparency, I had to modify my original symbology so that the different tree varieties would be immediately recognizable.   Making the land cover layer transparent had muted my original symbology colors.  

 

 

Coordinate Systems - Module 2

For Module 2, we learned about various coordinate systems.  For this lab, I chose the state of Tennessee, for my study area.  Tennessee is approximately 432 miles long from east to west.  It was no surprise to learn that Tennessee had three different UTM zones (15N, 16N, and 17N).  I chose the USA Contiguous Lambert Conformal Conic Projection for this map since this is what is used for the State of Tennessee grid system, according to the Tennessee Department of Transportation.  This is a good choice due to the fact that this projection works well with land with a larger east to west extent.   None of the UTM zones would have worked well since there are three for the entire state.  When creating a map for a study area, it is best practice to use a coordinate system that covers your entire study area. 

 

Map Design and Typography - Module 1

 

For Module 1, we reviewed concepts on map design and typography.  We learned about the five principles of map design which includes: legibility, visual contrast, figure-ground, hierarchical organization, and balance. We were provided with a map of Mexico, along with various features which required labels.  We were required to utilize labels for states, rivers, and major cities.  

Below is the map I created for this assignment.  I was especially excited to learn some new things when I worked on this map. I was unaware of the other options in the Labelling Options/More dropdown (priorities, weights, etc).  Learning about and using this feature really set me up for success when designing this map.  I set up priorities for the different features, making the rivers the least important.  Setting up weights and priorities allows the more important features to be layered over the less important features. I also made a new feature class for Mexico City and made another one for all cities except for Mexico City.  This allowed me to create a separate symbology for Mexico City and create this as a focal point (as the capitol).  I used several different types of labelling options for the different feature types (Labelling/Text Symbols) to help with visual contrast and legibility.  I attempted to balance the map by placing the map elements in the map's white space and unlabeled areas. 

Module 6: Scale Effect and Spatial Data Aggregation

During this week's module, we learned about scale effect and spatial data aggregation. Map data with a large scale typically has a lot of detailed information in it, but small-scale maps usually do not.  We also learned about spatial data aggregation and the Modified Area Unit Problem, which is a statistical bias present when conducting spatial data analysis, which can affect statistical results.

We were provided with US Congressional Districts and asked to determine if gerrymandering had occurred.  Gerrymandering is the process of changing voting district boundaries for political advantage. I learned about the purpose of the Polsby Popper score, which is calculated using the following formula:  PP= 4πA/P²  , Where A= Area ; P=Perimeter.  If the resulting number is closer to 1, that means that the area is compact.  If the resulting number is low (or closer to zero), this generally indicates that the area is non-compact. 

I calculated the Polsby Popper scores for all US Congressional Districts, which indicated that North Carolina’s District 12 had the least compactness score.  It appears that the district is spread out, and not contiguous with neighboring districts.

North Carolina - District 12 (Least Compactness Score)

 

Module 5 - Interpolation

 

During this module, we used various interpolation methods to create a visual representation of water quality in the Tampa Bay.  We utilized Spline (Regular/Tension), Inverse Distance Weighting, and Thiessen methods.  The biggest lesson I learned during this module was that the sample point set and surface can affect what interpolation method works best! 

The Thiessen method finds the sample values of the "nearest neighbor" and gives a proportionate weight to the value.  With Inverse Distance Weighting, weighted values are assigned according to distance.   With the Spline Technique, a mathematical equation is used based on surface curvature. 

Below is a screenshot of the Inverse Distance Weighting Interpolation Technique: 

(Note: The data provided for this exercise was intentionally modified for instructional purposes and is not an accurate representation of the water quality in Tampa Bay.)

Module 4: Surfaces - TINs and DEMs

For this module, we discussed creating, editing, and analyzing TINs (Triangular Irregular Networks) and DEMs (Digital Elevation Models).  One of the lessons I learned that this week is that the GIS Analyst must determine what techniques work best for the given study area. For example, data collection methods, study area terrain, and elevation are a few factors that impact which techniques produce the most accurate results. In this case below, the contours noted for the study area appear to be more accurate with the DEM.  

        TIN: 

            DEM: 

Module 3: Data Quality - Assessment

During Module 3, we worked on an assignment which focused on assessing the completeness of two      road networks in Jackson County, Oregon.  We were provided with TIGER roads and Centerline road      files for analysis.   TIGER or Topologically Integrated Geographic Encoding and Referencing has been an important part of the United States' spatial data framework since the 1980s.  TIGER has had accuracy challenges over the years, but has been steadily improving.  For example, in the "Positional Accuracy of TIGER 2000 and 2009 Road Networks" research article authored by Zandbergen,  Ignizio et al., TIGER 2009 was more accurate when it came to urban locations by a factor of two, when compared to rural areas.  

When conducting an initial assessment of the length of both road networks for the Module 3 Lab, it appeared that TIGER roads are more complete.  We were also provided with a grid for the entire        county and asked to determine how many segments from TIGER Roads and Centerlines are contained within each grid.  Then we were required to determine the length differences between the two road sets for each grid polygon.  Although I figured out how to this inside of ArcGIS, I utilized Excel to manually validate my calculations. My calculations showed that the Center_Lines actually had more segments per grid, than TIGER roads. 

 

Module 2: Data Quality Standards

For Module 2, we learned about data quality standards for spatial data accuracy.  The Positional Accuracy Handbook (1999) helped me learn about national standards of spatial data   accuracy and how to report a formal accuracy statement as required by the National Standard for Spatial Data Accuracy (NSSDA). Furthermore, we learned about the seven steps to apply these standards to our work.  I applied those steps to this assignment as follows:

1)      determined that we would be testing horizontal accuracy of the data sets.

2)     I created 22 test points within the study area  (various intersections in Albuquerque, NM)

3)     I created a point feature class for StreetMapUSA_Sample and created a point for the intersection closest to the test point data.  I also created a point feature class for ABQ_Streets_Sample and created points for intersections closest to the test points.  

4)     I populated all XY values for all points I created for ABQ_Streets_Sample and StreepMapUSA_Sample.

5)     I then calculated positional accuracy using the horizonal accuracy statistic worksheet.

6)     After calculating statistics, the GIS analyst would then prepare a formal accuracy statement. My statistics were off by a large margin, so I am in the process of getting assistance to ensure they are correct before I published them here.  With that said, once I figure out what I did wrong, I will add the correct accuracy statement.  The statement would be written as follows:

**Tested at XXXXX feet horizontal accuracy at 95% confidence level

**Compiled to meet XXXXX feet horizontal accuracy at 95% confidence level

It should be noted that 95% percent confidence level for horizontal accuracy is calculated by taking the RMSE and multiplying it by 1.7308.

Source: Positional Accuracy Handbook. 1999. Minnesota Planning, Land Management Information Center, St. Paul, MN

Module 1- Calculating Metrics - Spatial Data Quality

During Module 1, we learned about the difference between accuracy and precision. Accuracy is calculated using "assessment points", in which the true value and estimated value is known (Bolstad). Whereas, precision is the value pertaining to how accurate the estimated value is to the true value. 

The horizontal precision between the reference point and the average waypoint measured at approximately 3.26 meters. I calculated the horizontal precision (68%) to be 4.4 meters.  There is over a meter difference noted here.  This discrepancy could be due to the use of a recreational GPS device versus the use of a professional unit.

We also measured the horizontal distance between a reference point and the average waypoint location, and determined there was a 1.14-meter discrepancy.  The vertical accuracy, which was calculated by subtracting the elevation of the average waypoint (28.54m) from the reference point (22.8 meters), showed a difference of 5.74 meters. 

=