Results 1 to 20 of 23

Thread: Aghanistan geospatial analysis

Hybrid View

Previous Post Previous Post   Next Post Next Post
  1. #1
    Council Member Ken White's Avatar
    Join Date
    May 2007
    Location
    Florida
    Posts
    8,060

    Default Of course you can post comments, the more the better

    Quote Originally Posted by Mark Eichenlaub View Post
    I assumed that you had the signature option on the board just like any other board. I can't post comments?
    and your sig option is okay -- it's just that occasionally, people drop by and post a one-liner or a cryptic sentence and drop a link -- and do little else.

    If you're going to comment, do remember it's not the best venue for communicating and it's better to be overly verbose than so concise that others aren't sure what was meant. Comment merrily away...

  2. #2

    Default

    Quote Originally Posted by Ken White View Post
    and your sig option is okay -- it's just that occasionally, people drop by and post a one-liner or a cryptic sentence and drop a link -- and do little else.

    If you're going to comment, do remember it's not the best venue for communicating and it's better to be overly verbose than so concise that others aren't sure what was meant. Comment merrily away...
    Ok, thanks for the tips Ken.
    Saddam Hussein and terrorism
    http://www.regimeofterror.com

  3. #3
    Council Member
    Join Date
    Mar 2008
    Posts
    1,457

    Default Project finished

    I finished my project and am posting some of the results below.

    To begin with, I would like to thank John McCreary of AFCEA for providing data for this project.

    Before getting into the maps themselves, I need to discuss the data and, specifically, the limitations of that data:

    1. The data is exclusively from open sources, primarily press reporting which is subject to collection bias, inaccuracy, probable undercounting and a variety of other errors. The data is for 2007 and 2008 only. I did not have access to data outside this 24 month period.

    2. The location of incidents was not always specific or readily apparent, most often because the original reports had errors or did not give specific locations. It was not possible to determine the district in which an incident occurred for approximately 23% of the data – a statistically significant number. The district maps (included below) are therefore undercounting events by about a quarter. (Note: Total # of reported clashes at the provincial level for 2007-2008 is 4,633. Of those, 3,576 are included in the district-level data which was used to generate the maps below.)

    3. Afghanistan has undergone one extensive administrative re-division of its province and district boundaries and is in the process of undergoing another. Some province and districts split, others merged or ceded territory as a result of these reforms. For incidents that occurred in or near those boundaries, it is often difficult to determine which model the original report was based on. For example, Uruzgan province was divided in two – the southern third remained Uruzgan and the remainder became Daykundi province. A report, even one today, of an attack in “Uruzgan” province could have taken place in Daykundi because not everyone is aware of the change, particularly the Western media, and others, for a variety of reasons (tradition, politics, etc.) may not acknowledge the boundary change in reports. To simplify this complex problem these maps use the older 32 province/329 district model. Standardization on this model caused some problems during transcription as some of the reporting utilized the newer model. In most cases, I was able to manually place the incident in the correct district or province, but in about two dozen cases at the district level this was not possible. Those incidents were removed from the district data.

    4. The population and demographic data (ethnicity) are estimates from the Afghan Central Statistics Office (CSO) and vary in currency from last year back to 2002. Ethnolinguistic group data is primarily based on studies done prior to the beginning of the Soviet intervention in 1979 as well as estimates by the CSO based on interviews with local leaders, NGO studies, and other reporting. These estimates are subject to substantial error.

    Because of these errors, I think one shouldn't look at the actual number of attacks per district, but the relationship between districts and where violence is concentrated.


    Discussion:

    Although not shown on the maps attached below, the data show a clear increase in violence in 2008 from 2007 in all areas, particularly IED attacks. There were a total of 1,704 reported attacks of all kinds in 2007. In 2008 there were 2,929 attacks of all kinds, an increase of almost 75%. IED attacks increased by 150%, from 239 to 596. Suicide bombings increase at a relatively smaller rate of 30%, from 124 to 162.

    Map 1: This maps shows relative violence overlayed onto district ethnicity. It's important to note this shows which ethnic group holds a plurality, not a majority, since not every district has a majority ethnic group (this is actually one goal for the various district reorganizations - to provide each district with an ethnic majority). It shouldn't be a surprise that violence correlates strongly to the Pashtun ethnic group. All the districts with significant violence that do not have a Pashtun plurality have large Pashtun minorities. The exception is the Nuristani areas.

    Map 2: This is a comparison of violence in Afghanistan for 2007 & 2008 by district. Again, no huge surprises, but it's interesting to note that levels of violence can vary widely over a single province.

    Map 3: This map takes the violence from Map 2 and adjusts it on a per capita basis (based on the 2005 CSO population estimates), so the most attacks per capita are shown in red. The effect is that rural, less populated districts with significant violence are highlighted. Notice Kamdesh district in Nuristan province where last weekend's attack on FOB Keating took place. It has one of the highest violence per-capita ratios in Afghanistan.

    I have some more maps from the project, but these three are the most interesting. I have all the data in ArcGIS geodatabase format that I'm willing to share for anyone who wants to PM me. I also have much of the data in excel format for anyone that's interested. I can also generate more maps if there's a specific request - at least for the next few weeks before my ArcGIS license expires.
    Last edited by Entropy; 10-08-2009 at 05:35 PM. Reason: added stuff

  4. #4
    Council Member Ken White's Avatar
    Join Date
    May 2007
    Location
    Florida
    Posts
    8,060

    Default Lot of work

    and a great job, thanks.

  5. #5
    Council Member marct's Avatar
    Join Date
    Aug 2006
    Location
    Ottawa, Canada
    Posts
    3,682

    Default

    Entropy,

    Really nice job there ! Just out of interest, is your dataset available?

    Cheers,

    Marc
    Sic Bisquitus Disintegrat...
    Marc W.D. Tyrrell, Ph.D.
    Institute of Interdisciplinary Studies,
    Senior Research Fellow,
    The Canadian Centre for Intelligence and Security Studies, NPSIA
    Carleton University
    http://marctyrrell.com/

  6. #6
    Council Member
    Join Date
    Mar 2008
    Posts
    1,457

    Default

    Thanks Ken and Marc!

    Marc, I have the data in excel or ArcGIS geodatabase. Send me a pm with your email and I'll send what I have to you.

  7. #7
    Council Member
    Join Date
    Apr 2009
    Location
    Maryland
    Posts
    827

    Default

    Kivlonic:

    Kudos.

    FYI: One thing I learned from Iraq, as you saw in Afghanistan, is that sub-district and district boundaries, unlike in the US, are dynamic factors that must be tracked, updated in real time to have relevance. Actually, as with Kurdish areas in Iraq, the boundary changes themselves are typically leading or trailing factors of conflict---either creating them as one party jockeys for control over the area of another, or trailing in that a newly resettled majority has now gained control, and wants to affiliate with a different political structure (leaving the original remaining residents in a lurch).

    One of the perennial problems (which you are not able to solve withoin your assignment) is that each change in districts actually represents a substantial change in demographics. In most cases, these census figures are based (and reported) on an underlying census block basis, so moving these blocks from one district to another moves the pop figures around.

    The "disputed" district of Mandeli (Diyala) in Iraq was a circumstance where there were layers of depopulation/destruction as Kurds were pounded by successive attacks by Sadaam, then Iran, then Sadaam again. By 2009, you have a town which was once a large and important provincial district seat where, now, virtually everything has been destroyed, and only a few Kurdish stragglers exist or have returned. Questions: What data do you use to determine who belongs to this place? Who should be there? Who should control it (Erbil or Baghdad?)? How accurate and valid is the data?

    Bear in mind, too, that once a change is made, all the place and town names can change, too. So dynamic administrative/political boundaries are reflected by the multiple alternative names given to places. Before the federal highway system (and Euro Counterparts) standardized names in the West, we did the same as them---used descriptive placenaming: Shiek Abdullah's place is only Shiek Abdullah's place if he lives there; once he moves somewhere else, the name follows him, and his old place is called Peshmerga HQ. Dynamic placenaming, and name tracking for all the alternative names is another big universe beyond your project.

    One of the perennial, but unrecognized problem is conflicting administrative boundaries. Just because a province or district changes it's boundaries, what boundaries and "service areas" are used by national ministries, or other regional service providers? In Iraq, US recon folks couldn't understand a lot of what the national ministries were doing because, in part, they were looking at them from the provincial level. National roads (as opposed to provincial, district or local roads), and many other national-level services were managed based on a regional "service area" basis, or, in most cases, based on older provincial boundaries (they never changed them). Very confusing stuff if you don't realize these things as dynamic factors.

    The US has a system for event reporting. In many cases, I found that there were substantial discrepancies between the event and the coded location. So your stuff is as good as it gets sometimes.

    Great work.

    Steve

Similar Threads

  1. Center of Gravity Construct
    By Strickland in forum Doctrine & TTPs
    Replies: 152
    Last Post: 02-25-2013, 08:06 PM
  2. Class Analysis and COIN
    By AmericanPride in forum RFIs & Members' Projects
    Replies: 21
    Last Post: 02-26-2009, 02:51 AM
  3. Overhauling Intelligence
    By SWJED in forum Intelligence
    Replies: 14
    Last Post: 05-05-2008, 06:26 PM
  4. Replies: 52
    Last Post: 12-23-2007, 05:55 AM

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •