GIS 6005 Lab 9: Bivariate Choropleth Mapping

Bivariate choropleth maps provide an efficient way to display two variables and how they are related in a manner that is easy to interpret by the viewer. Like univariate choropleth maps, a bivariate map uses color progression to symbolize classes of data, the difference being its use of two color ramps for two related variables and the combination of those ramps to display the correlation of those variables.

Creating a bivariate choropleth map requires preparation of the data. We first must ensure the data for the two variables are normalized. In the case of the map above, the data for obesity and physical inactivity were already normalized by population. Next, we will create a new field in the attribute table for each variable and populate them with codes for each variable class. To keep the map understandable for the viewer, each variable is generally limited to three classes; this is because the number of classes displayed in the final map is exponential based on the number of classes for the two base variables. Thus, three classes for the two variables equals nine classes in the final map. Four classes for the variables would create a confusing map of 16 classes.

A third field is added to the attribute table and is populated with the concatenation of the two fields created above. This field will be used for the symbology of the final map and represents the combination of the two chosen variables (i.e., obesity and physical inactivity rates). The creation of the color symbology in the above map required some experimentation. The foundation of color choice for bivariate choropleth maps lies in choosing complementary colors for the two variables. Hue, saturation, and value options can then be adjusted for the overlapping classes. Once the symbology is finalized, the legend must be manually adjusted to create a suitable bivariate choropleth legend. This is accomplished by converting the legend to graphics, ungrouping the elements twice, and manually placing the color squares in their final placement. Text elements may then be added to label the legend as is appropriate.

GIS 6005 Lab 8: Analytical Data

Our task this week involved three distinct steps: Downloading and processing raw data, choosing variables that are suitable for comparison and visualization, and creating a map layout to display multiple maps, charts, and text. The data came from www.countyhealthrankings.org and included information on a host of variables broken down by state and county. We first chose two variables we thoughts may be correlated (negatively or positively); I chose obesity and unemployement with the hypothesis that there would be a positive correlation. In order to incorporate this raw data into ArcMap, we created new Excel worksheets containing only the data needed. These worksheets were then used to create charts visualizing our chosen variables. They were also joined to two separate United States maps in ArcMap and symbolized as choropleth maps.

The most difficult part of this assignment, however, was the final task of creating a layout to contain the various maps and charts. I chose to keep the layout as simple as possible with a white background and no neatlines or borders. This allows the viewer's gaze to flow easily from one element to the next. I settled on a tabloid-sized page in landscape orientation in order to stack the two maps on the left and to make room for the charts on the right.

Based on my end product, there does not appear to be a correlation between obesity and unemployment. If a correlation is present, it is relatively minor.

GIS 6005 Lab 7: Terrain Visualization

The various tasks for module 7 have taught us ways to aid map viewers in visualizing the terrain that is being displayed. One of the more precise methods is to display and label contour lines; such maps are familiar to hikers and anyone familiar with USGS map products. Another method is to apply hillshade to a DEM (digital elevation model). Hillshade tools, such as provided by ArcMap, take the elevation data of a DEM and simulate the visual effect of the sun at a specified angle and direction. This visual effect greatly aids the viewer in reaching an intuitive understanding of the topography of the displayed area. Hillshade can be combined with other map elements, such as the above symbology representing tree and landcover types for Yellowstone National Park, to provide the viewer with more information. In this case, the landcover layer is set to a transparency of 30% to allow the hillshade layer to be seen. This gives the viewer an understanding of the relationship between landcover, topography, and elevation that would not be possible if one used hillshade or landcover alone.

GIS 6005 Lab 6: Choropleth Mapping

One of our tasks this week was to use a diverging color scheme to represent positive and negative population change in the counties of our chosen state. We first needed to apply a suitable projection for our state. In my case, I chose UTM Zone 13N for Colorado. This UTM zone covers the majority of Colorado, leaving only a strip along the westernmost border outside the zone and thus mildly distorted. If we were required to work with detailed spatial statistics this week or were focused on that section of Colorado, I would have chosen a different projection or created a custom one.

The next step was to normalize the data. Rather than use the normalization options in the layer properties, we created a new attribute in which we used the field calculator to calculate the percentage of population change by county from 2010 to 2014. This was the field we used as the basis for our choropleth maps.

We then faced three main tasks: Decide on the classification details (method and number of classes), how to symbolize the classes, and how to design the legend. I began by using the Natural Breaks classification method with seven classes. Seven classes would allow for one class representing minimal population change and three classes each for population increase and decrease. Natural Breaks was a good start, but, for this general reference map without any other specified purpose, I wanted to create a more symmetrical and directly comparable classification scheme to more easily see how counties compared. I manually adjusted the population increase and decrease classes to mirror each other.

For symbolizing these classes, I based the diverging color scheme on one included in ArcMap. I did adjust the HSV values, primarily saturation, to even out the contrast somewhat. The orange to red color ramp seems natural to represent a decline and is easily distinguished from the green color ramp. The yellow representing the middle class is distinguishable from both ramps while fitting in to both color schemes.

The final piece of the layout was the legend. My original intention was to use the legend property options to create the legend, but these were too limited. The data naturally fall into three categories (as I have classified them), so I converted the legend to graphics in order to split the legend into three categories. This created a legend that was more legible and intuitive to interpret.

GIS 6005 Lab 5: Symbol Mapping

Our lab exercises this week introduced us to some of the difficulties cartographers face when displaying data using proportional symbols. It can be an intuitive way to present data, yet it can also be a struggle to create a map that does not create unnecessary confusion. In the above example, we were required to use proportional symbols to communicate job gains (a positive number) and job losses (a negative number) by state. The first hurdle was dealing with the negative job numbers; directly symbolizing both positive and negative numbers from a single layer in ArcMap does not produce an acceptable result. We needed to export selections of states with positive and negative job numbers into two new layers. A new field was then added to the 'states with job losses' layer in which the field calculator was used to convert the negative numbers to positive. The resulting two layers were then used as the basis for the proportional map above.

The primary variable in creating this style of map is choosing how large the symbols should be and if Flannery's compensation should be applied. In this case, compensation was not applied. However, there was a discrepancy in size between the two layers. Experimentation with the minimum symbol sizes created proportional symbol progressions that were equivalent. Once I finalized the legend layout (converting it to graphics and editing it within ArcMap), I wanted to apply a type of transparency that is not available within ArcMap. I exported the layout to Adobe Illustrator and applied a 'multiply' transparency to create the transparency effect seen above. This minimizes the interpretive problems created by symbol overlap.

GIS 6005 Lab 4: Working with Color

Color ramp	Notes	Stepwise Intervals
Linear progression		R20	G35	B27
	-Lightest color selected using approximately the same ratio of RGB as darkest -The RGB steps divided by 5 leads to R steps of 20, G steps of 35, and B steps of 27	20 20 20 20 20	35 35 35 35 35	27 27 27 27 27
Adjusted progression
	-Using the same dark and light colors of purple, the adjusted progression steps are based on the general rule (provided in lab text) that the darker color steps are a third higher than average and the lighter color steps are a third lower than the average.	27 24 20 16 13	47 40 35 29 24	36 32 27 23 17
ColorBrew
	-The steps provided by colorbrew created a more defined color progression with greater variations in the steps between each color.	69 2 -22 11 29	28 73 47 37 53	52 57 23 19 28

              

The differences between my linear and adjusted progression color ramps are subtle, but the increased step size between the darkest colors in the adjusted progression ramp does aid in differentiating them. The Color Brewer ramp, however, created a progression of individual colors that are more defined and easier to distinguish. The RGB values show that Color Brewer varied the steps between each color selection much more than the linear or progression ramps. For example, the step from the darkest purple to the next color is large for R and B but relatively small for G, but the step to the next color is tiny for R (2) and very large for G (73). The third step actually reduces the R value instead of increasing it. This reflects our text in showing how complicated the relationship is between RGB values when creating color ramps and how many options we have in creating our maps.

GIS 6005 Lab 3: Typography

I went through several rounds of experimentation in order to get the right combination of legibility and visual hierarchy for each of the map features. Starting with the water features of San Francisco Bay and Golden Gate, I wanted to use fairly standard font/color combinations for water features. I chose the built-in 'coastal features' symbol option that converted the labels to blue italicized Arial font with extra spacing. I made San Francisco Bay larger due to its larger size, and I used the Draw toolbar to rotate the labels to align with their orientations and to fit within the map. My label for Lake Merced, however, was created manually by changing the font color to blue and italicizing. I increased character spacing only slightly in order to fit within map. I also applied a white halo of 0.5 to allow the text to stand out from the background.

The San Francisco label needed to be prominent without dominating the layout. I maintained the Arial font, increased the size to 14, changed the style to Bold, and applied a halo of 0.5. I placed the label in the center of the map; this will naturally be one of the first labels a viewer is drawn to, and it also happens to be in an area of the map without competing labels.

I treated Marin Peninsula as a landform. After experimentation, I settled on a dark grey (70%) Arial text with increased character spacing (40) and a light grey (20%) halo. The combination of grey text and grey halo allows the label to stand out from the green park background without contrasting as intensely as a black/white combination would. I used the Spline Text tool from the Draw toolbar to create the moderately curved text that roughly follows the curve of the peninsula.

I kept the labels for Treasure Island and Angel Island as simple Arial text with no halo. The text stands out from the blue background without a halo or other alteration. Also, given the placement of the islands away from other features and labels, a placement to the upper right clearly labels each island without the need for pointers or having the labels directly on the features. I also kept Sausalito relatively simple, but its location required a halo as well as placement directly on it. I experimented with rotating the text to better fit the feature, but this never looked right to my eye.

The parks were a challenge due to their relatively small size and location near other features and labels. I began by fitting the text within the features for Golden Gate Park and the Presidio of San Francisco, but this was a tight fit . This also obscured the features themselves in the process. Instead, I moved the text away from each park and used the Draw toolbar to create a simple line pointing from the label to the feature. This eliminates any confusion of what the text is labelling, and the consistency of style works well. I used the same style for the Golden Gate Bridge; the bridge feature is far too small for a directly placed label, and it fits well with the park style.

Keeping the general labelling style simple, I directly placed the text for Russian Hill, Nob Hill, and the San Miguel Hills on the feature areas with haloed Arial text. The halo was necessary to allow the labels to stand out from the streets symbology underneath. I used the same text style for Twin Peaks, only adding a small pointer line to more clearly indicate the feature area.