Cartography Final

The participation rates of high school students for each state aligned with the state’s SAT average, for 2009.  The data show participation rate in the SAT is low in the mid-west and heaviest on the East Coast.  High participation appears correlated with lower average scores

Our final cartography project involved presenting two sets of data on a single map.  The biggest challenge was to choose the best means to combine these two types of data in a way that meaningfully communicated the result. I chose a choropleth map of state participation rates as the base.  Then I added graduated symbols to represent the mean scores.  The biggest chore was dealing with all the labeling for the various combinations of large/small states and large/small symbols.  In the end, I'm pretty happy with the result, the colors, and the overall message that can be derived by studying the map.

Thanks for the class everyone!

Google Earth

I'm a huge Google Earth fan, so this week's assignment was pretty fun.  We learned how to convert ArcGIS maps and layers to KML files that Google Earth can use. Above is an image of my population density map from last week's assignment layered onto Google Earth's map.  We have the ability to turn on and off individual layers and adjust transparency as well.  I found the way Google Earth handles the legend from ArcGIS a bit strange (only have 9 positions that we can place it in), but otherwise the conversion was pretty painless. Google Earth seems to manage the projection just fine.

The second part of the lesson was creating a Google Earth tour across southern Florida.  Overall, this was pretty straightforward, though Google Earth did seem to be a bit "erratic" with knowing when layers were turned on and off during the recording and playback of the tour.  I think I eventually got it working correctly, starting with the dot map on, turning off to zoom to the Miami area, back on while zooming to Tampa Bay, off when looking at St Pete and Tampa, and then back on for the finale.  Overall, fun and pretty easy, though it can be hard to smoothly move the camera around when looking at city details.  They need to add some sort of "steady cam" feature for us amateurs.

Dot Maps

This lab focused on creating dot maps.  This map shows the population densities of southern Florida according to data from the 2000 US Census.

The biggest challenge of this map was creating a color scheme that allowed for the varied elements while still allowing the population dots to remain prominent. I wanted to give the frame background color and spent a long time with both light and dark backgrounds. I ended up choosing a gradient light blue that seems visually pleasing.  I think the gradient helps avoid the implication that southern Florida is an island while still helping the main terrestrial area stand out.  I always struggle with what color to use for “land”.  In the end the combo of sandy color with blue for water features and green for wetlands works for me. I chose to diminish the color of the boundaries between counties but not to eliminate them entirely.  If I set them to No Color, the Keys are lost and we don’t want that.   Red might seem a bit extreme as a color for the population densities, but also seems to work pretty well with this color scheme.  Less bright colors tended to get lost in the sand/blue/green combo.

As a computer guy, I found it interesting how processor-intensive it was to generate the masks.  I have to wonder if there is an inefficiency going on there.  One would think it should not be much more than a couple clips joined together to move points from county boundaries into the urban areas that intersect the country boundaries.  

Flow Maps

This map  was a bit of a departure for me as I like to keep my maps quite simple, tending to adhere to the Tufte-ism “more data ink, less non-data ink”. This began by creating a dark background layer that was not too “ocean blue” but also worked well with the continent colors.  I think the dark blue I chose provides good start on figure-ground organization.  It was quite a task finding colors that worked well with each continent AND the background color.

I really found the flow arrows (Asia/Oceania) that wrapped around the Cape of Good Hope in the lab example distracting and somewhat deceptive.  These folks are not coming to America on one of Magellan’s ships!  It made more sense to me visually (and spatially and realistically) to have the flow arrows breach the side borders and wraparound.  Unconventional, perhaps, but I think a better representation that doesn’t take over the whole map.

I lightened the country border colors as this map is about regional movement, not individual countries. This also helped the labels stand out more. I created a halo effect around the label text by duplicating the label level, increasing the stroke size (to 1) of the labels in the level copy, changing the copies label stroke color to white, and positioning this new layer behind the label text.  This further helps the labels stand out against the continents.

The most significant effect applied was a glow effect and drop shadows applied to the flow arrows.  Using Inner Glow, I created a minor glow effect that helped give the flow arrows more of a border.  I also applied Drop Shadows to the arrows that helped lift them off the surface.  I think this helps accentuate the movement of people from the regions to the USA. At one point I had the continents also raised by use of a drop shadow.  I later removed it and was so relieved by the reduction of clutter.  The important message is the movement of people from continental regions to the USA, not the application of as many visual effects as can be crammed in a map. On that note, I also did not use the bevel effect.  I found these reminiscent of the worst abuses of the PowerPoint epoch.

Isometric Mapping

This week in Cartography, we created an isohyetal map that shows annual rainfall in the state of Georgia.  We were provided rainfall data values at weather station points across the state and then manually interpolated the isohyet lines. Manually interpolating the lines is not unlike Sudoku or similar brain teaser games.  You have to find the best-fit dividing line between diverse data points while keeping the "rules for drawing isolines" in mind.  Since this map was created completely in Adobe Illustrator, we also needed to manually label the isolines (including ticking for depressions) and the normal essential map elements (legend and North arrow in particular).

Proportional Symbol Mapping

This week in Cartography we learned about proportional symbol mapping.  This is basically the idea of presenting a symbol (as simple as a circle or something more representative, like a bottle) that is scaled in a manner that reflects scale to the user. The first map above was created entirely in ArcMap.  It is not much to look at, but it was quick and fairly simple to create.  The symbols are scaled to the cube root of the actual value, which allows us to see a value for every country.  I also applied a halo effect to the labels to make them a little easier to see.  I'm not sure why ArcMap insists on labeling islands within a country individually (see Greece Greece Greece for example).

I'm happier with this map created in Illustrator, though it would be a lie to say that I am happier with Illustrator.  Illustrator does allow for some powerful "post processing" of a map.  In this case, marrying the partially transparent circles (wine colored!) with a bottle gif and circular text for good measure.  But Illustrator comes at a cost. It seems unnaturally difficult to make simple color choices for example. The big challenge with proportional symbols seems to be finding the sweet spot where the max value looks appropriately "max" but the smallest values are not lost visually.  In this case, Luxembourg was the real challenge as tried to find a size that eliminated some of the crowding but not so much to lose Luxembourg entirely.  At the same time, the circular text often collided with the country borders in the background.  In the end, I also applied a halo-like effect to the "Belgium" text to prevent it from getting too entangled with the coastal border.

Choropleth Mapping

A choropleth map using color

A choropleth map in grayscale

This week in Cartography our assignment was to use a simple map of the USA state and a spreadsheet of US Census data from 1990 and 2000 to create two choropleth maps.  A choropleth map is a map that uses shading intensity proportional to the data value to visually communicate the results to the reader.

In the color map, I presented the percentage in population change for each state, grouping the states into five color groups using a classification method called Natural Breaks. In the grayscale map, states were grouped by regions and again presented in five groups using just variations in shading intensity.  Overall, I'm pretty happy with the results.  The maps look clean, convey the central message and are approachable.

Typology

This week's emphasis was on Typology as it relates to Cartography.  There are many conventions when presenting text (either as titles, legends or feature labels) that has been standardized in maps over the centuries.  Expert cartographers have gone before us and established the "rules" of where and how to best place text on the complex and challenging background that every map presents.  Over time, many of these "best practices" have been recognized and codified.

Our task with this map was to place 17 labels that follow these conventions and best communicate the goal of the map.  I've kept the overall title and legend quite simple.  All the key (island) names/labels share the same style and color and blend in to match the overall "sandy" tone of the keys.  As much as possible, I placed these "on land" if I could avoid overprinting. Some of the keys are quite small or busy so it was necessary to place their labels "on water".  Aquatic features are in blue and italicized. Most of these are slanted to fit their geographic positions. Symbols were placed for Marathon and Key Colony Beach "cities". Marathon Shores is technically a neighborhood, not a city/town, so it didn't get a symbol but followed the color/style used for cities. The airport and country club were two features that required call-outs to pinpoint their exact location.

It is interesting to note that Google Maps and Bing Maps have slight differences on where exactly some of these features are. It took a while to figure out Vaca Key is actually the whole of the center island.

Map Composition

Our assignment this week was to take a rather jumbled map and rearrange it in a manner that follows map composition principles and will best communicate the desired information to the user.  To do this, I primarily relied on the principles of screening and visual weight.  The most important element is the weighted coloring of the counties.  These fill most of the available space and utilized a sequential color scheme that intuitively communicates "higher" population percentages with darkening colors. The title also has a high visual weight and cements the contract with the user of what we are trying to get across.  The legend also takes a prominent position as it is needed to interpret the counties in more detail.

All the other items on the map (the insets, scale bar, north arrow, author, date, data source) are supplemental and would detract from the main message.  To minimize their visual impact, these elements have been set to 40% gray and place in less prominent positions on the page. Also, any color was removed from the inset maps except for the rectangles that show the area of interest.

The double inset created a bit of a dilemma, as I'd not seen many maps that utilized two insets.  I ended up placing the USA inset first and furthest from the counties.  Then, I placed the Florida inset map between the USA and county maps.  This seems to provide a visual hierarchy of sorts and a linear progression.  Overall, I am pretty pleased with the outcome and look forward to seeing how other students tackled this problem.

Intro to Illustrator

The goal this week was to export a map from ArcGIS format (.mxd) into Adobe Illustrator (.ai) where it could be further edited/manipulated and then exported as an image file (.jpg - seen above).  This lab walked us through all the steps to accomplish this as well as detailed exercises in working with Illustrator objects, groups, and layers.  It is especially important to make sure that ArcGIS objects (such as the scale bar and the geographic data) are grouped together into a common layer when working in Illustrator.  This grouping will ensure that if any items in the group are resized, all the other objects in that group will stay within the relative scale relationship that existed in ArcGIS. Our lab also provided an overview of many of the most common tools in Illustrator for enhancing maps.

Data Classification

Combined Data Classification Methods

This week we focused on common methods of classifying and presenting data.  Our first project presents four views of the African American population of Escambia County.  Each of the methods used has different advantages and disadvantages, depending on the type of data and the spatial relationship that is being explored.  These differences are easily seen when we apply different methods (quantile, natural breaks, equal interval, and standard deviation) to the same data and map. Since we were comparing methods for the same data and map, I thought it important to use the same color ramp for each map.  This helps us compare the impact of the method better and not just be distracted by different color schemes.

 

Of the four methods used, I thought the quantile method seemed the best fit for the data. The quantile method breaks the total population up into classes (five in this example) of equal population size.  So for a map of population density, each class represents a 20% increase (with five classes) in the subject population. This is an easy to understand distribution and presents a well graded map.

Statistical Foundations

Histogram with Custom Bins

This week in GIS3015 we concentrated on basic statistics and graphing.  The histogram above (with trendline) demonstrates the use of custom "bins".  In this case, we are using a custom bin size of 20, which presents a neat appearance on the horizontal axis and a better distribution (in my opinion) of the stock data.  I've also edited the trendline axis on the right to remove a 120% increment (which doesn't make sense in this case) and cleaned up the labeling to only use whole numbers.

Good Map / Bad Map

Good Map!

http://videoonly.com/assets/wavo.pdf

This is perhaps an interesting choice for my first “well-designed” map. This is from a “weekly specials” flyer for an electronics store called Video Only (VO) which has stores around the Seattle area (Bellevue and South Center are near Seattle).  These maps are trying to do one thing and one thing only – help you get to the local store – and as such score high in Tufteian values of clarity and efficiency.  They “know” you know the local area and are simply communicating major cross streets in relation to the major highways around Seattle (520, 405, and I-5).  They also include just a couple clearly-labeled local landmarks, showing the stores’ relationships to these larger landmarks (Sears, Ross, Southcenter Mall) in ways that help orient the user.  The maps are not particularly concerned with scale, but emphasize the routes the user will need and minimize visual clutter.  As a local, I could easily find either of these stores. 

Bad Map!

http://mapbox.com/blog/election-mapping-usatoday/

Those of us in the USA are all painfully familiar with maps like this election day map from USA Today.  I think we could agree that it is a technically beautiful map (many cases of graphical excellence with regard to layout, muted colors for areas out of the main focus, white borders between states and counties, clear labeling unless you live in Sioux City, etc.). However, I’m presenting this as a poorly-designed map for failing to deliver the right impression. If one was to take a quick look at the map, who won Iowa?  Clearly red is dominant in the image, so Mitt Romney must have won.  There is no “scale” to indicate that the population of some counties is extremely low and others high, just a binary red/blue indication. The map relies on the bar scale below the map and a green check-box to confirm that the state went to President Obama.  Without this “addendum” I think this map doesn’t "tell the truth about the data" (Krygier & Wood, 2005).

Reference:
Krygier, J., & Wood, D. (2005). Making maps: A visual guide to map design for GIS. New York: Guilford Press.