new jersey geographer

Over on MetaFilter today, there is a great post filled with links to interactive maps detailing various aspects of the recession and the eventual rebound in the economy. In looking through these maps, I’m irked by the cartographic conventions employed by some of them. Google Maps has started a dangerous trend: representing everything possible as a point on a map.

For example, take Richard Florida’s “The Shaping of America” interactive map in The Atlantic. The map relies on points of varying size to show the number of patents, the population and income levels for selected US cities. The size and color of the point is an indicator of the city’s performance relative to the surrounding “metro average.” There’s no definition of what these “metro areas” are. They are not delineated on the map. The boundaries of the city are also not reflected.

Why is this a problem? While not having the city outlined is concerning, the truly egregious flaw is that the theme of the map is dependent on a ratio without well-defined boundaries. Take, for instance, Trenton, New Jersey. Trenton is slightly north of the geographic center of New Jersey, however it is routinely grouped with “South Jersey” and is rarely grouped in “Central Jersey.” The parts of the State that identify themselves as “Trenton Metro” are limited to adjacent municipalities, if that. So what is the “metro area” of Trenton? Is Princeton included in Trenton? That would absolutely set Trenton above average for all three indicators mapped.

Map depicting just a few of the boundaries for Trenton, NJ

Now consider the New York Times Immigration Explorer. The Times has been cranking out some amazing maps lately and this one is no exception. This temporal, thematic map is rendered using Flash. It shows the 3,000 or so counties within the US with great detail and clarity. Ethic groups as a percentage of total population are reflected on a chloropleth map while the overall population is shown using dots of varying size. We’re back to the dot map, but it’s very different from the Google Dot Map above. The dots are sized in proportion to the total population, not an ill-defined sample. Also, the Flash interface allows the user to manipulate the base size of the dots, which allows the user to discern differences in population in the most sparsely inhabited regions.

Immigration Explorer would still convey its intended message if the cartographers behind it employed a dot map like the Shaping of America. Considering the data is explicitly by county, a point map could be used without introducing ambiguity. However, considering the geographies represented by the Shaping of America map are not well-defined, we are left guessing what we’re actually trying to represent with those dots.

Google Maps (and KML, the language for user-defined data in Google Earth) supports lines and polygons. Granted, there are more hoops to jump through to get vector data into Google Maps, but there are ways to do it. Maps that really require representation using polygons should not be constructed using points & Google Maps. Using Google Maps isn’t what’s important. Making sure your map delivers its intended message is essential.

Starting with Kenya, Google is allowing users to download the base data collected through the Map Maker service, as either KML or Shapefile format. Google’s licensing allows for only non-commercial use. Hopefully, this data will support some of the non-profit mapping efforts taking place on the continent. The license also restricts (or at least, severely limits) competition with services provided by Google. This puts OpenStreetMap in an odd position – OSM is restricted from incorporating the data into its own service. Considering OSM has been around since 2004 and Map Maker only 2008, and seeing how many people in over 100 countries have been offering Google data, OSM needs to better align (or contrast) itself with the work Google is doing. One of the reasons I believe Map Maker has gathered the amount of user-generated content in the past year is that Map Maker is incredibly easy to use and if you do run into issues, there is clear and concise help. This is compared to the OSM Beginner’s Guide. Now, I understand that OSM is geared towards a more technical audience than the Map Maker service, but OSM needs to spend more time fostering a community outside of the devoted submitters. What is going to prevent Google from offering up a GIS service akin to the Virtual Earth on ArcGIS platform currently being offered by ESRI and Microsoft? A service built on TA/Navteq data, enhanced with local knowledge? A service that is available now from OSM, but much easier to use? Providing base data for Kenya is just the beginning.

Mikel Maron has some more thoughts about OSM and Map Maker, as well as some comparison screenshots. (Via Mapperz)

Before Google Earth 5 had come out, I had commented to a few people that the dates for the aerial photos (which used to be displayed in the bottom center of the screen) were way off for New Jersey. The date was reported as 2004, but it was clearly 2007 imagery.

Well, this little problem is now a bigger issue. In browsing the area around Rowan University (where considerable development has occurred in stages between 2002 and 2008), I noticed several places where the “2004″ imagery shows a further progression of development than the (actual) 2006 photography. I’ve prepared a KMZ file with annotated points, highlighting places where the development that is visible in the “2004″ aerials had not yet occurred in 2004.

Considering I had noticed this discrepancy throughtout the State before Google allowed you to transition through the older imagery, I took a look at another location in the State where development is occuring, Newark. Google Earth has 11 different sets of photography available for Newark. The very big problem is that the default imagery is pre-Prudential Center. This is even reflected in Google Maps. The default images in both Maps and Earth is now the “December 30, 2002″ images. Many of the buildings that are in Street View for Newark are not to be found on the aerial photography. The 2002 data is the second-oldest data available, but it’s now the default.

This seems like a pretty big mixup for Google to make. We’ve gone from posting imagery within the same day to reverting back to a 4 to 6 year lag.

Find another place in the State (or elsewhere) where the dates are obviously wrong? Let me know in the comments.

A Google Street View car hit a deer while imaging a road in upstate New York. You can still see the deer in the driveway in this Street View. The Register has the pictures of the incident. The pictures have already been removed by Google – considering the driver stopped and notified police, why were they even posted to begin with?