Kosovo: Applying GIS in an International Humanitarian Crisis

by David G. Smith, Department of State

ArcUser July-September 2001

One positive outgrowth of the recent tragic events in Kosovo was a cooperative effort to apply GIS tools for managing the international response to a complex humanitarian emergency. A team from the Department of State responding to this situation faced the challenge of creating a common geospatial and data-organizing framework for a variety of national government agencies, international organizations, and nongovernmental organizations (NGOs) that were addressing a multiplicity of problems ranging from destruction of housing and infrastructure to residual hazards posed by land mines and unexploded ordnance.

Mapping damaged buildings.

Damaged buildings were inventoried using imagery analysis and mapped
in ArcView GIS.

This project, which began in fall 1998, proved most successful in providing a rapid assessment of damage to housing and other infrastructure. Information from remote sensing was combined with ground evaluations gathered using an assessment form designed to populate a georeferenced database. The geospatial data framework for the international response in Kosovo had two components--a foundation map data created prior to field operations and operational data generated in the field.

Most of the geospatial foundation data was developed by the U.S. National Imagery and Mapping Agency (NIMA). The agency produced a "humanitarian planning map" that was designed to be widely released and used a standard commercial GIS data format. The humanitarian planning map included layers such as elevation, transportation, populated places, hydrography, and administrative boundaries. Equally important, NIMA produced a gazetteer of standardized place names. The gazetteer, available in both digital and hard-copy form, helped clarify the confusing mélange of Serbian and Albanian variants of place names.

This data was refined and released in successive versions, each more inclusive, accurate, and detailed than the previous release. For example, road layers were initially released at 1:250,000 scale and later at 1:100,000 scale. Incrementally releasing data gave relief workers reasonably good data to work with early on rather than waiting months for the release of "perfect" data.

The long gestation of the Kosovo crisis made response planning somewhat easier. The humanitarian planning map and data, the product of many months of work by NIMA, provided a common reference for geospatial data that was essential for agencies--whether U.S. government, international, or NGO--dealing with difficult crisis response issues.

The Kosovo crisis can be divided into three phases--the build-up phase from mid-1998 to March 24, 1999; the NATO bombing phase from March to May 1999; and the resettlement and rebuilding phase from June 1999 to present. The following are some examples of ways in which GIS was applied in each of these phases.

An escalating series of killings, bombings, house burnings, land mine explosions, and other acts of violence prompted various international efforts to reduce violence, relieve tensions, and alleviate suffering. On the humanitarian side, the United Nations High Commission for Refugees (UNHCR) worked with a constellation of NGOs and other humanitarian organizations. On the security side, the Kosovo Verification Mission (KVM), sponsored by the Organization for Security and Co-operation in Europe (OSCE), took over in late 1998 from an earlier international coalition attempting to reduce tensions and violence.

In November 1998, a small Department of State team was sent to Pristina to promote the use of GIS as an efficient means of organizing and sharing information among the various organizations operating in Kosovo. The humanitarian and security agencies working in the area were collecting data for their own operations. Each agency also had strong and legitimate needs to access some of the data collected by other agencies. Why not put certain types of security and relief data in a standard georeferenced database that could be shared in a GIS?

For example, the UNHCR was collecting information on housing conditions and food and medicine needs in villages throughout Kosovo. KVM was collecting information on land mine reports and security incidents. UNHCR teams needed to know where hazards had been reported, and KVM needed information on conditions in villages and where relief teams were operating.

Working with these organizations, the State team designed prototype databases using standard commercial relational database software that could be easily integrated into a GIS. The team worked with KVM and UNHCR officials in assessing the data that was already being collected and creating a database structure to accommodate that data with as few changes as possible. Teams were using low-tech methods of data collection (e.g., clipboards and ballpoint pens). If the data collection system was too complicated, no one would use it. Although conditions deteriorated and UNHCR and KVM removed operations from Kosovo, GIS use and other concepts continued to be implemented. For example, the land mine monitoring team adopted the proposed database, continued populating it with data, and used that data in a GIS.

Damaged buildings

After KVM and UNHCR curtailed operations in Kosovo, information demands changed. For example, policy makers were very interested in estimating the damage to housing in towns throughout Kosovo. Imagery analysis provided a way to make such an assessment. Village by village, NIMA analysts counted damaged buildings and saved this information in a spreadsheet that included coordinates in latitude/longitude in a process that was labor intensive.

The completed file contained approximately 2,000 locations. Putting this data into a GIS provided a very striking picture of the devastation that occurred as ethnic Albanians were driven out and their homes destroyed. Graphic displays of the data--prepared using a GIS--were very helpful in presenting the issue to the public, as well as in helping different agencies plan future relief efforts.

Once the fighting stopped, hundreds of thousands of refugees began quickly returning to Kosovo. At that point, the agencies started implementing the United Nations' Rapid Village Assessment, a survey of infrastructure damage. UNHCR coordinated with other relief organizations in designing a survey questionnaire that would gather enough data to be operationally useful, but not require so much data input that field survey teams would find it onerous. To facilitate data entry and minimize errors, the paper form carried by field teams corresponded exactly with the structure of the relational database.

Making an accurate survey of housing reconstruction needs was the primary concern. The survey form used a five-level categorization of housing damage that ranged from little or no damage to destroyed or irreparably damaged. A set of simple illustrations showed the criteria for each category. In addition to counting and classifying housing in each village, teams gathered data on the condition of the water supply, electrical power, school and health care facilities, and presence of displaced persons and relief agencies.

GIS requires that locations be accurately geocoded. Despite the valuable NIMA gazetteer, geocoding locations was challenging because unique identifiers did not exist for each village.

UNHCR devised a five-digit identification system called p-code (p for place). The first two digits referenced the municipality (or opstina) in which the village was located and the last three digits assigned a unique number to each locality. The p-code allowed an accurate link between the database and the GIS regardless of the GIS software used. The p-code model pioneered in Kosovo is now being used in Mozambique and elsewhere. In practice, the system worked quite well. UNHCR was able to issue an initial report covering some 500 villages within 10 days of dispatching the first field survey teams.

Unexploded mines and ordnance

Locating land mines and unexploded ordnance was another concern. Data from several sources was combined in a GIS and made available to civilian humanitarian organizations.

Another concern in Kosovo was locating land mines and unexploded ordnance. In addition to the previously mentioned database of land mine reports, the NATO military coalition force (KFOR) received minefield data provided by the former Yugoslav army as well as information on target sites for NATO cluster bombs. This data was put into a GIS and made available to civilian humanitarian organizations.

GIS is also being used in forensic investigations of atrocity sites. While this application of GIS has been more slowly realized, awareness of its potential value is growing. Among other advantages, GIS has been particularly effective in displaying data visually, showing both spatial and temporal patterns of war crimes, and communicating what happened to the public.

The experience in Kosovo has provided some useful lessons for applying a GIS-based data management system in a complex humanitarian emergency. Here are some of those lessons:

  • Although not readily available for many parts of the world, a foundation of good geospatial data is essential. In particular, accurate data on subnational administrative divisions and populated places is necessary. Making partial data, even if imperfect, available in a timely manner is more important than waiting many weeks for a fully developed dataset.
  • A standardized reference gazetteer for place names with unique, unambiguous identification codes avoids confusion among users. This resource may need to be created on an ad hoc basis.
  • Prior planning is key for getting all the participants onboard and in agreement on data needs. Waiting until the crisis has advanced causes problems because no organization wants to retrofit collected data into another format.
  • The database structure must be adapted to the needs and practices of data-gathering organizations, not the reverse.
  • Data management does not need to be centralized in one location if all data collectors use a common geospatial reference. However, some entity should be in charge of data entry and maintenance for each type of data and make data available to other interested entities in a usable form. In Kosovo, the UN's Humanitarian Community Information Center performed this function. Many entities can collect the same type of data if a common format for recording it is used.
  • Initially, low tech may be better than high tech. Internet dissemination and exchange of information is great in theory, but it does not work in a country with no working telephones or Internet access. Planners should assume that data collectors will not be using GPS or other technically sophisticated means of data collection.
  • A GIS for a humanitarian emergency needs the support of senior-level personnel in the organizations involved. This support will overcome potential resistance at the working level. Relief workers face a myriad of stressful and conflicting demands of a chaotic environment. To enlist their support, they must realize that a GIS and data collection project is not an additional burden but a tool that will help them do their jobs.

For more information, please contact
David G. Smith, Chief, Geographic Information Unit
Office of the Geographer and Global Issues
Department of State
Washington, DC 20520
E-mail: ACdsmith@us-state.osis.gov

About the Author

David G. Smith has been a State Department officer since 1974. He holds a bachelor's degree in economics from Rice University and a doctorate in Latin American history from the University of Texas. He has done postdoctoral study in demography at Georgetown University and in geography at George Mason University.

Note: The views expressed in this paper are solely the author's and do not necessarily reflect those of the Department of State or the United States government.

Table of Contents for the July–September 2001 issue

Contact Us | Privacy | Legal | Site Map