A group of UCLA geographers published a paper yesterday in the MIT International Review entitled “Finding Osama bin Laden: An Application of Biogeographic Theories and Satellite Imagery”. The UCLA team used purely open source data, including “Landsat ETM+, Shuttle Radar Topography Mission, Defense Meteorological Satellite, QuickBird”. Then used a variety of commons geographic analysis techniques, “distance-decay theory, island biogeography theory, and life history characteristics” to predict high probability locations for Osama Bin Laden. The story has already been picked up by 90 media outlets and has been popping up on the front page of several major news outlets.

It would never make it out of the labyrinth of classification schemas in the US government, but it would be fascinating to see what a crowdsourced search for Bin Laden would turn up if better data was made available from the intel/defense community. Since the government data will never be released we thought we could at least help make the open source data easily accesible. So, we took the available data in the MIT article plus relevant data on Afghanistan and pushed it into GeoCommons. We’ve embedded a map with our own take below.

To view this map in GeoCommons Maker! click here.

In addition to the UCLA data we’ve added gridded population data for the area. A big part of the UCLA thesis was Osama would be, “in a larger town rather than a smaller and more isolated town where extinction rate would be higher”. So, the gridded data gives a rough view of population densities in the remote Tora Bora region.

Source data for the maps is here:

Structure Locations of Possible Hiding Spots of Osama Bin Laden, Parachinar, Pakistan, 2009
Tora Bora 10 KM Buffer Rings
Gridded Population Data, Afghanistan and Pakistan border near Tora Bora

Would be great to see what other folks can do with the data to promote other perspectives. Also a nice opportunity to show the power of opening data up for better analysis, QA and alternative perspectives. Kudos to the UCLA team - great to see geographers in the news for doing what they do best.

Popularity: 18% [?]

We’re wrapping up a great time at GEOINT this week, and wanted to share just a few short posts that caught our attention about the show. Sean Gillies gives his hilarious opinion of the GEOINT blog, while All Points Joe Francica shared a biting commentary for one vendor who denied him a picture. In all seriousness, it’s great that GEOINT has taken steps to be plugged in to social media through their blog, Twitter account and Facebook group. Although, it would have been nice to hear about the blog before the announcement on Tuesday so that people could connect online before the show.

Google Earth for the iPhone came out this week, and it’s pretty slick. A really cool feature is that Google Earth is available in offline mode through the iPhone and desktop by simply choosing to “continue without network”. It also remembers your cache, so any searches or locations you have viewed in the past will transfer between desktop and phone with or without internet connection.

Just in time for Halloween, Virender at Mibizaar posted a mashup with the creepiest places on earth. Bhangarh, India topped the list. Very Spatial listed some data of all the corn mazes in the country. Something fun for everyone!

Popularity: 10% [?]

This is the final blog post in our eight part series on violence in Iraq and Afghanistan before and after the military surge. We have discussed our seven key findings in great detail and provided our in depth analysis and the raw data for our analysis on Both Finder! and Maker!. We welcome anyone else to use this data or any other data to continue to look into various aspects of the wars in Iraq and Afghanistan. GIS Professionals, Neo-Geographers, and Spreadsheet Jockeys can all work together using publicly available data to do high-end analysis using GeoCommons, and we hope you do.

Here are abstracts of our key findings with links to the more detailed blog posts:

1. There has been a shift in violence from Iraq to Afghanistan. When all events of violence were tabulated there was a 13% increase in violence (attacks) in Afghanistan and 23% decrease in violence (attacks) in Iraq between the pre and post surge time periods. Blog Post Here.


2. When the total violence is examined, Afghanistan’s percentage share of violence has increased from 12% to 17% between the pre and post surge time periods. Blog Post Here

3. While there has been a shift in percent of change from Iraq to Afghanistan, the total violence in Afghanistan is dwarfed by Iraq. Since the surge, total violence in Iraq is equal to 5,465 attacks while Afghanistan’s is considerably less with 1,104 attacks. Blog Post Here

4. From the beginning of 2005 through to the end of the 1st quarter of 2008, the percentage of total attacks that were improvised explosive device (IED) related has increased by 13.19 % in Afghanistan, and 14.75% in Iraq. The pre-surge average for percentage of total attacks that were IED related in Afghanistan was 18.81% and 35.66% in Iraq. Post-surge averages show that 23.76% of attacks in Afghanistan were IED attacks and 41.59% of attacks in Iraq were IED attacks. Both countries are favoring IED attacks over other types, such as suicide bombing and assassinations. However, there has been a significant decrease in suicide bombings in Iraq since the surge but an increase in Afghanistan. This could indicate a move of more radical elements from Iraq to Afghanistan. Blog Post Here

5. From January of 2005 to March of 2008, there was a decrease in reported attacks on energy infrastructure by %70, although during the same time period the total number of attacks increased by %40. Blog Post Here

6. The overall trend of violence in Iraq has been moving gradually east towards the Iranian border. In Afghanistan there has been no consistent pattern or shift in violent attacks. Blog Post Here

7. Violent attacks in Iraq have been geographically concentrated and consistent over time, but attacks in Afghanistan are far more dispersed and locationally erratic. This could mean that the same strategy of hold and clear hostile territory by systematic progression, might not work in Afghanistan where attack patterns are not stable over time but spatially volatile. Blog Post Here

For Access to our public datasets in Finder! On Iraq, Click Here
For Access to our pulic datasets in Finder! on Afghanistan, Click Here
To view the maps that have been created on Iraq, Click Here
To view the maps that have been created on Afghanistan, Click Here

To get a sense of how attacks played out week by week, we used a time-stamped kml file in Google Earth. The change in the color of the points roughly indicates pre and post “surge” (early may 2007, in this case).

Check out the screen shots:

Attacks tend to concentrate in a few areas. In order to make sure all of them showed up, some points were pushed out just slightly from the point to which they were geocoded.

Popularity: 12% [?]

Key Finding #7
Violent attacks in Iraq have been geographically concentrated and stable over time but are far more dispersed and locationally erratic in Afghanistan. The shape file, which was downloaded from Finder!, was used to run spatial statistics on the trends, particularly the distribution of the attacks in each country. To display and share the results, we exported shape files from ArcMap and then uploaded them into Finder! to then be displayed using Maker!.
The figures below show the spatial means of fatalities for Iraq and Afghanistan, calculated monthly, going back a year from March, 2008. Orange indicates fatalities from all violence and blue are fatalities from IED attacks. Spatial means show the spatial central tendency of a set of points weighted by a specific feature, in this case that is the average geographic center of attacks weighted by fatalities. A comparison of the means across different periods provides information on how the distribution of attacks may have shifted over time. The map of Iraq shows that the centrality of fatalities has remained relatively constant over the last year (the average locations are tightly clustered) with a few exceptions.

In Afghanistan, the spatial means for both fatalities and IED fatalities show more movement. The attack patterns in Afghanistan are far more erratic and geographically dispersed as shown in the map below.

Figure 16: Calculated Spatial Means in Iraq and Afghanistan

The map below also shows the spatial means, however, the size of the circles represent points in time: the largest circles are post-surge and the smaller ones are pre-surge attacks.

Figure 17 and 18: Calculated Spatial Means: Fatalities from Violence in Afghanistan and Iraq.

The map illustrates an outward movement of attacks. In Afghanistan, the population is located around the periphery of the country since the interior is mountaneous. In Iraq the map shows a much tighter clustering of attacks with a slight shift north since the surge. Spatial standard deviations for the monthly fatalities and IED fatalities were also calculated. These statistics indicates the degree of spatial dispersion of the attacks weighted by some feature such as the number of fatalities or number of wounded. On the chart, the higher standard deviation score indicates that attacks are more spatially dispersed. With the exception of October, 2007 in Afghanistan, the spatial variation of fatalities has remained relatively constant in both countries and in terms of both fatalities and IED-related fatalities. Although, in comparison to Afghanistan, fatalities have been more geographically concentrated in Iraq. The statistics show that attacks are roughly twice as dispersed in Afghanistan as in Iraq—again reinforcing the different pattern of attacks being seen in Afghanistan.

Figure 19: Spatial Standard Deviations: Iraq vs. Afghanistan

Another method for examining the data is to map the spatial standard deviation ellipses for each month in both Iraq and Afghanistan. On both maps the blue ellipses reflect total fatalities and the yellow reflect IED fatalities. In both countries, IED attacks show a similar pattern to the overall pattern of violent attacks in the region. Although in both cases IED attacks are slightly more concentrated than violent attacks in general.

Figure 20: Spatial Dispersion of Fatalities in Afghanistan

Figure 21: Spatial Dispersion of Fatalities in Iraq

The more narrow ellipses in Iraq illustrate that the attacks are more concentrated in a southeast to northwest direction. The wider ellipses in Afghanistan indicate more dispersion and are located in a southwest to northeast direction.

In conclusion, the analysis shows a much different pattern of attack in Afghanistan than in Iraq. Pundits have pointed to the clear and rebuild strategy, used successfully in Iraq, for reducing violence in Afghanistan. In Iraq “First, US and Iraqi security forces project into an area and provide initial security, then locals are recruited into auxiliary security forces, and reconstruction and aid projects quickly follow,which encourage the population to engage with security forces for a new tier of security.” This strategy was systematically used to secure neighborhoods, sometimes on a block by block basis. Since the geographic pattern of attacks in Iraq was very consistent, it makes sense that this strategy would work well. The areas of violence were well understood and could be rooted out in a systematic way. The evidence produced by the analysis of Afghani attacks shows that the same strategy would be challenging since attacks are dispersed and there is not a stable pattern of violence. This means forces could spend a large amount of time securing neighborhoods where there were no enemy forces or potentially future attacks. This could further alienate the local population.

Datasets:
Spatial Mean, Fatalities in Iraq
Spatial Mean, Fatalities in Afghanistan
Spatial Mean, Fatalities from IED attacks in Iraq
Spatial Mean, Fatalities from IED attacks in Afghanistan

Popularity: 12% [?]