People may not be as big as elephants, whales, or gigantic trees but they are still the most “space dominating” organisms on the face of the planet. What do I mean by “space dominating”? I hypothesize that humans tend to conquer over spaces so drastically that they greatly affect the quality of living for other organisms. Sometimes the affect is so great that species can go through drastic loss of numbers and possible extinction. I decided to investigate my idea further and use Finder! and Maker! to help me.

I investigated to see if I could find endangered species numbers by country and came across the website of the International Union for Conservation of Nature (IUCN). The IUCN regularly puts out their Red List which is an intense listing of the conservation status of plants and animals around the world. The list sets guidelines to evaluate the extinction risk of thousands of different plant and animal species. Once their list has been created they showcase their results to promote conservation across the planet. I created a dataset in Finder! of this list to show the number of threatened (critically endangered, endangered, and vulnerable) species by country in the world. The list is also broken up by number of mammals, reptiles, birds, fish, mollusks, other inverts, plants, and total threatened species. The map below shows the total number of threatened species by country. (click on images for larger views)

Finder! link - IUCN Red List of Threatened Species by Scientific Classification, World by Country, 2008

Now I wanted to find more data to test my original hypothesis (Are humans the most ‘space dominating’). I found human population data by country and also area of each country in square kilometers. With this data I then calculated population density’s for every country in the world. I then took the total number of threatened species by country and divided it by the area of the country. This gave me a value that I call my, ‘Threatened Species Density Rate’. I then decided to correlate this new Threatened Species Density with the Human Population Density. For my hypothesis to be correct I would need to see that countries with high population densities will have high Threatened Species Density Rates because humans have not allowed the animals living there to have the resources to properly survive. Below is the correlation analyzed in Maker!

The correlation above shows a slope of .64 which is somewhat high and shows that a somewhat high correlation is found in countries that have high threatened species density rates and high population densities. So my hypothesis can be seen as being somewhat accurate. Let’s take a look at this point shown in a map on Maker!

To really see this point illustrated lets zoom in on NW Africa on a map of Population Density and Threatened Species Density mapped out together (The purple polygons represent Threatened Species Density and the Green Dots represent Population Density)

Here we see countries with low population density (Algeria, Mali, Niger) have lower threatened species density. Compare these to the countries with high population densities (Liberia, Nigeria, Ghana)that have high threatened species densities.

It seems that if you keep animals and plants free from humans, they will likely be able to survive properly. Not all countries obey this rule, but the rate around the world is pretty high. It is also interesting that many of the countries that have low population densities have areas that are very desolate and unlivable by human standards. These places include deserts or ice fields where extreme temperatures and conditions don’t allow humans to live but allow specialized species to survive that can handle all the extreme conditions. Maybe one day if humans find ways to survive in such places causing population densities there to grow, we might then see an increase in the threatened species densities there as well. We will have to wait and see.

Popularity: 10% [?]

Over the course of Where 2.0 and WhereCamp TFL (Tobler’s First Law) came up several times and I though it might be a good time to revisit the concept. TFL is the idea that “Everything is related to everything else, but near things are more related than distant things”. It is a simple and powerful concept. After watching the host of location aware mobile application demos at Where 2.0, especially Jeff Holden’s talk on Footstreams and Greg Skibiski’s talk on Sense Network’s analysis tool for mobile data streams, I began to wonder how this might change TFL*.

With all the streams of mobile data coming online I’ve been wondering how this might affect the basic precepts of TFL. Will a population augmented with location aware devices change some of our basic perceptions of geography? Will time become as important as space and place for geography? The vast majority of geography and GIS deals with a static world. Few of our quantitative or qualitative analysis tools deal explicitly with time, yet all around us vast data sets are being built providing detailed location and behavioral information with very fine grained time stamps. Do we have the right tools in our analytical arsenal to deal with hairy fourth dimension of time?

Amazingly this is exactly what Tobler had in mind when he developed the ideas behind TFL. The concept originated in paper by Tobler entitled “A computer movie simulating urban growth in the Detroit area” presented at the 1969 Commission on Quantitative Methods of the International Geographical Union. That’s correct a temporal simulation of urban growth presented as a movie in 1969 - pretty amazing - at least in my opinion. When there was a forum at the 2003 meeting of the AAG to discuss TFL, Tobler stated that he was “just having fun doing an animation in order to bring time into geography more explicitly”. Apparently hacking geography was producing cool stuff even back in 69′.

Tobler was influenced by physics and specifically Feynman in his thoughts about the interaction between space and time. This is interesting because physics, probably more than any other discipline, has been focused on the multi dimensionality of phenomenon. As such it is a very fertile field for thinking about how time adds a fourth dimension to geography. While much of what motivated TFL involved time it is not explicitly part of the law. Only distance is identified as a factor that causes a decay of interaction/relation.

The natural corollary would seem to be that - things that are near for a long period of time are more related than things that are near for a short period of time. I think this corollary really strikes home when we think about mobile device that are location aware. Whether we are talking about friend that are near or points of interest (POI) that are near the temporal factor is possibly more important than the spatial factor.

Let’s look back at Jeff Holden’s talk where he collected 642 days of footstream traffic with his GPS enabled device. He found that he visited Starbucks 2-4 times per day and his mean time between Starbucks was 29 hours and peaked at 12 hours in one week. Obviously there is a strong relation between Jeff and Starbucks, but I’d argue the strength comes from the duration in time he spent at Starbucks as much as the proximity of Jeff to Starbucks. I believe as the Web fully morphs into the GeoWeb - location in browsers (HTML 5), location in desktops (Windows 7), location in mobile hardware (everyone), location in mobile software (Google latitude), location hybrids (Skyhook Wireless) our understanding of the temporal dimension to all this data will become critical. Is the corollary to Where 2.0 then When 2.0?

*More generally you can read Paul Ramsey’s take here on the two talks above.

Popularity: 11% [?]

Chrysler has recently announced that by June 9th, 2009 they will be closing 789 of their 3200 dealerships across the USA. Sales for many dealers have been very low and a little over 50 percent of all the dealers account for close to 90 percent of all the company’s U.S. sales. This shows a definite need to eliminate dealers that are not performing at high enough levels. Below is a map of all the dealer closings across the country. (click on the map and all images, including the charts, below for a larger, clearer view)

Finder! link to Chrysler Dealer Closing List, USA, 6.9.2009
Maker! link to Chrysler Dealership Closings, USA, 6.9.2009

From the news it is certain that most of these closing have been based exclusively on sales performance of the dealership. But could there have been other factors that lead to poor performance in the area? I decided to use Finder! and Maker! to look into finding an answer to this question.

I started at the state level and made a map of number of dealerships closed by state. (see below)

Also in this dataset I added many different economic and non-economic factors that I thought might correlate with the number of dealerships that were closed by state. These factors included: Unemployment Rate by State, State Population, Number of Foreclosures in the State, Foreclosure Rate in the State, Number of Registered Automobiles in the State, Registered Automobiles per Person in the State, and Population Density. The correlation results are below:

I also ran correlations with the number of dealerships closed per state normalized with the state population. The correlation results are below:

From all these correlations I realized that nothing really correlated. The highest was the correlation between Number of Dealerships Closed by State with Total Number of Registered Automobiles in the State at .69. This shows that states with large amounts of registered vehicles tended to lose more dealers than states with less registered vehicles. These states with high amounts of registered vehicles also tended to have large populations where the demand for vehicles was greater. This data is not too remarkable and still tells us very little.

I then tried to focus my geography down a little bit more to see if I could find any different results. I looked for counties in the country that had three or more dealers closed and looked at Unemployment Levels in those Counties for the month of April 2009. Below are my findings:

With the national unemployment rate being about 6.5% across the nation we see that unemployment at the local level did not have that high of an effect. Two of the five counties are below the national unemployment level, two are above the level, and one is right around the level.

To get a better understanding of these closings I have concluded that I need more data. The two types of data that I would love to collect are: Percentage of dealerships in a state that are being closed and Sales Figures for all Chrysler dealerships across the country. With percentage of dealerships in the state closed I could see what states lost the majority of their dealerships and correlations could have been more telling. With the sales data I would have been able to see if the closures were solely based on sales and have a better picture of where sales were doing well and where they were doing poorly. If anyone out there has such data or has other ideas about data to try and correlate please respond. With these new ideas, I will then reexamine this topic and see if I can find any interesting trends.

Popularity: 14% [?]

We’ve been collaborating with our co-founders back at George Mason for the last few months on a paper modeling oil dependency/vulnerability from a geographic perspective. We wrapped up the paper yesterday and it got me thinking about what a fully interactive version of the paper would look like. What if all the maps and charts were embeds? What if you could download all the data sets used for the analysis right from the paper?

While many journal have come online and some even in openly accessible venues - I don’t think we’ve really tapped the power of the Web for interactivity, data sharing, innovation, or peer review. Having more interactivity in charts and maps could make research more accessible and engaging. Further, having the data for a paper downloadable could provide better peer review, and create the opportunity to innovate and extend the research. A fellow resercher could have an idea to extend or optimize an equartion test it on the same data set and see if it yielded better results.

Currently the academic peer review system is quite limited with a only a few colleagues reviewing papers. This is often a bit of a buddy system, especially in fields with only a few experts. Opening up the commenting and feedback process could foster even better critique of work. By also making data available, an incentive is created for fellow researchers to interact with the research, provide feedback, and collaborate with authors. Potentially you could create a journal in such a format leveraging interactive tools across the web like Swivel, InfoChimps, ManyEyes, YouTube, Google Charts, DataMash, or Flickr, To give this idea a go I’ve created an example of what such an article could look like with our oil paper as the guinea pig:

The Repercussions of Being Addicted to Oil: Geospatial Modeling of Supply Shocks

Laurie Schintler – George Mason University School of Public Policy
Rajendra Kulkarni – George Mason University School of Public Policy
Tom Buckley – FortiusOne Inc.
Emily Sciarillo – FortiusOne Inc.
Sean Gorman – FortiusOne Inc.

Abstract

In a world addicted to oil and the prodigious infrastructure that produces it there are distinct spatial variations in oil dependence and vulnerability. Depending on a countries location they have dependencies of different sources of oil. Disruptions in any one source of oil will have differing impact in both magnitude and breadth of countries affected. To begin to understand such a volatile landscape this paper will review pertinent literature surrounding oil shocks and propose a model of how they can be geospatially modeled. Specifically the modeling will calculate an oil import vulnerability index, oil dependency index and the percent reduction in import diversity for 63 countries.

Read the rest of this entry »

Popularity: 25% [?]