CAMBODIA NATIONAL
LEVEL 1 SURVEY

Executive Summary

Introduction

The National Level One Survey in Cambodia, funded by the CANADIAN INTERNATIONAL DEVELOMENT AGENCY (CIDA), is part of a global approach to defining and reducing the socio-economic impact of landmines. This global approach is coordinated by the UNITED NATIONS Mine Action Service (UNMAS) and is designed to meet the needs of the mine action community including donors, national authorities and mine action implementers.

Landmine Impact Surveys are intended to: To ensure the quality and comparability of survey results, UNMAS, funded by CIDA, contracted a UN Certification and Quality Assurance Monitor. The monitor's role is to provide technical advice to the survey and to report to UNMAS and to a UN Certification Committee. Providing results are satisfactory this survey will be UN-certified thereby providing confidence that the survey meets a high standard of quality.

CIDA contracted GeoSpatial International Inc as the Canadian Executing Agency (CEA) with a mandate to design, plan and implement all aspects of the survey, including field data collection, quality assurance and database creation.

Results

*One village was unable to be surveyed as it is under Vietnamese administration and surveyors were asked to leave. The other village was not surveyed as it was deemed too distant and difficult to access safely.

Cooperation with Cambodian Authorities

A Memorandum of Understanding (MOU) between the Government of Canada and the Royal Government of Cambodia established CIDA and CAMBODIA MINE ACTION CENTRE (CMAC) as project partners in April 2000. It also established detailed arrangements for cooperation between the two organizations. During the two year survey the Royal government created a new agency to coordinate humanitarian mine action known as the Cambodia Mine Action and Victim Assistance Authority (CMAA). The CMAA was assigned responsibility to create and maintain a national database of all mine related information and activities. The final survey database was therefore handed to CMAA at the conclusion of the survey in May 2002.

Logistical support and cooperation focused on CMAC as it has the operational mandate for humanitarian demining. The MOU stipulated that surveyors were to be recruited from amongst existing CMAC staff. The use of CMAC deminers as surveyors was of exceptional benefit to the LIS project. CMAC also provided GeoSpatial with some of the equipment necessary for the survey.

Survey Team Structure and Organization

The Project employed a total of 106 people - 4 expatriate managers and 102 local staff. As the educational standards of the CMAC staff were extremely variable, and some proposed hiring practices did not follow acceptable Canadian standards, a rigorous process was designed for surveyor selection. This included both Canadian and Cambodian representatives and proved to be extremely beneficial.

A basic training course was provided for surveyors involving an intense mix of classroom and practical work including role-playing to develop interview techniques and familiarity with the survey questionnaire. At the conclusion of the training period all candidate surveyors were assessed from which seventy-six (76) were selected and the remaining twenty-five (25) were placed on a reserve list. During the two-year survey only four (4) replacements were required from the reserve list

Refresher training was held at the end of each deployment in a particular location (between .5 and 2 days per session). The topics covered were based on problems identified by surveyors and survey management through the results of quality assurance activities.

The rigorous recruitment process, clear and comprehensive operating procedures and a strong management team meant there were few disciplinary issues. Incidents that did occur included one instance of unauthorized use of equipment and failure to comply with safety procedures; two instances of staff absent without leave; and one instance of failing to comply with survey implementation procedures.

Liaison with the Mine Action Community in Cambodia

The Cambodian mine action community had been engaged in collecting information and otherwise responding to the problem of mines and UXO for about eight (8) years at the time this survey was initiated. In light of this established capacity, a determined effort was made to assist the community to understand how the survey would improve long-term capacities to systematically respond to the mine/UXO problem.

The inception mission team met with all major mine action organizations. Initial meetings were followed by a workshop held early in the project which brought these groups together towards further eliciting information on past experience and existing requirements with respect to mine action survey work. Particular attention was paid to the information needs (database/query requirements) of these stakeholders.

Early survey results were shared with organizations as a means of eliciting feedback that more clearly and specifically defined the needs of an ever-expanding stakeholder group. Documentation included a monthly 'Progress Update' distributed to the Mine Action Coordinating Committee that provided a statistical overview of the progress of the survey and preliminary findings in terms of the contamination status of villages. A final briefing was held for all interested mine action and other agencies at the end of the survey project.

Developing the Questionnaire

The questionnaire developed for use in the survey was based on two previously developed questionnaires; one from the SAC in Washington DC and the other from CMAC. This latter instrument was seen as being of a reasonable length; relevant to the information needs of Cambodia; and written such that surveyors and villagers would understand the questions. A draft version was circulated throughout the mine action community for comment and underwent a number of revisions before being 'finalized' for use during surveyor basic training. Effort was made to further enhance clarity by soliciting surveyor suggestions for improvement during training. As a result, a revised version was drafted in Khmer and later translated into English. This version was tested in villages during training and again in a more formal 'pre-test' prior to the start of actual survey operations.

Approximately one month after the start of field operations, work was suspended for two (2) weeks in response to UNMAS concerns that the survey was not using the SAC/IMSMA questionnaire and was not following standard survey procedures as reflected in draft guidelines prepared by UNMAS. Discussions amongst representatives of GeoSpatial, CIDA and UNMAS indicated that the problem was less the questionnaire and more the fact that there was no UN QAM assigned to the project at the time. Operations resumed with minor procedural changes.

Cambodia Mine Victims Information System (CMVIS)

Initially it was planned not to collect victim information but rather to use the victim information contained in the CMVIS collected by Handicap International and the Cambodian Red Cross. UNMAS agreed but directed the survey to ensure the quality of all information imported to the L1S database from any external source.

It was found that due to the methodologies used by CMVIS to collect and record victim information it would be difficult to confirm the exact contaminated area in which mine/UXO accidents had taken place. Exact location of accidents is particularly important for the survey as contaminated areas may be prioritized for clearance based on the number of accidents that have occurred within them. The collection of victim information became part of the survey for this reason.

A large amount of time was spent cross-checking L1S casualty information with that held in the CMVIS. While the overall number of casualties reported by each project is quite similar, there are significant differences in specific casualty information, particularly in terms of the location of accidents and personal information about casualties. An agreement was reached in early 2002, to merge the two data sets in an integrated database. It is expected that this integration will occur sometime in 2002. The integrated database will make available to users the advantages of the data available from each project.

Aerial Bombing and Cluster Bomb Contaminated Areas

Although UXO found in Cambodia date from World War II and from the French Indochina wars, most air-delivered UXO contamination in Cambodia is the result of US air attacks along the Ho Chi Minh Trail. Part way through the survey, the US Embassy in Cambodia provided the L1S with bombing records for the period 1965 - 1975. These records indicate that between 1965 and 1975, 10% of the 150,000 air attacks on Cambodia employed various types of cluster bombs. The target coordinates of all recorded attacks were entered in the L1S database to produce a map of potential air-dropped UXO contamination within the country.

Cluster bombs produce a class of UXO that while not technically mines may impact on communities in similar ways. Although not identified as a particular problem during the pre-test or in provinces that surveyed early in the project, in heavily-bombed provinces surveyed later in the project, surveyors quickly became overwhelmed reporting on high numbers of cluster bomb contaminated areas. The survey questionnaire was modified in response to this new problem.

False Negative Sampling and Adoption of the Census Approach

In areas where expert opinion suggests that all or many villages are probably not mine/UXO contaminated, a sampling technique may be used to confirm that this expert opinion is correct. The advantage is that only a small number of villages need to be surveyed, and if they are confirmed to be negative, the remaining villages do not need to be surveyed. This sampling program is knows as False Negative Sampling (FNS).

FNS was tested in several provinces in Cambodia. To compare results of the FNS test with actual village conditions, a full survey was conducted of all villages in the test area. This test indicated that FNS was not appropriate for use in provinces with moderate to high levels of contamination. After further tests in one of the few areas of Cambodia with suspected low levels of contamination the L1S team concluded that a census approach was required and that this would extend the survey from 15 to 20 months. CIDA accepted the recommendation and funded the additional level of effort.

Survey Operations and Logistics

The Field Manager was supported by an Assistant and by five (5) team leaders. Team Leaders in turn assigned the specific villages to be surveyed by each 2-surveyor detachment. Each team contained between five and seven detachments. Contingency plans were developed to avoid situations (ie. severe flooding) or issues (ie. commune election related security problems) that might impact survey activities.

In densely populated areas, each team surveyed approximately 300 villages per month-long deployment. In rural areas, where distances between villages were greater, each team surveyed approximately 200-250 villages during each deployment. In highly contaminated areas, team production was approximately 120-150 villages per deployment. Logistical procedures adopted enabled all areas to be completed on or before schedule.

Provincial governors were always approached for permission to operate in the province and to obtain available information prior to the survey. Team Leaders would consult with local district and commune authorities to inform them of the survey work in their area of jurisdiction. These visits also provided information on the location of new villages, road and access conditions, criminal activity or security issues that might impact on survey activities that was incorporated into operational plans as relevant.

Village-Level Procedures

Each village was visited by one two-man survey detachment. One member of the detachment conducted the interview while the other recorded information in a diary and produced sketches when required.

The village chief or deputy village chief was the first person interviewed in each village. If both were absent the detachment moved to the next village and returned to conduct the survey when one or the other was present.

If the information provided by the village chief indicated that the village was contaminated with mine or UXO fields, a comprehensive survey was undertaken in the village. Once information provided by the village chief was recorded, surveyors would identify and interview at least three residents of the village who resided close to or were in some way directly affected by each distinct mine or UXO area reported. Where possible, surveyors aimed to interview respondents of different gender, age and socio-economic class.

On average, the complete survey of a contaminated village took a full day to complete. Depending on the degree of contamination affecting a village, the survey could be completed in as short a time as a half day, or in instances as long as three (3) days per village.

Team Leaders checked each survey report for omissions and conducted field checks to ensure data accuracy before submitting the reports to the Field Managers for further field checking and a review by the Field Editors. Once checked by the Field Editors, the reports were given to the QA/QC team who conducted additional field checks before finally submitting the questionnaires to the database unit. Missing, incomplete or inaccurate information identified at any stage throughout this process was corrected by a review of information contained in the surveyors' diaries relative to the village in question or a revisit to the village by the detachment originally undertaking the survey.

Safety

There were no mine/UXO related injuries. SOPs were developed at the beginning of the project and were vigorously enforced and revised throughout the project.

Over the course of the project, 90 field staff used 50 motorcycles and eight 4-wheel drive Toyota pickups to conduct activities in twenty-four provinces, traveling over an estimated 1,000,000 kilometres. By project end only one surveyor had been involved in a traffic accident that resulted in serious injury.

Detailed planning including days of preparation and reconnaissance ensured that staff faced the minimum risk possible. This activity included security assessments, and evaluation of the operational area, including lines of communication, local infrastructure, and physical threats, if any. Risks faced by the staff included on occasion, difficult river crossings and travel over water.

Effective communications was an essential element in maintaining safe operations, in addition to the use of HF and VHF radios, extensive use was made of other communications networks including local administration, military, and police communications networks. Where possible use was made of mobile phone networks.

The effectiveness of these systems was demonstrated when a motorcycle accident occurred in a remote location resulting in a surveyor being seriously injured. The time from injury to hospital was less than eight hours.

Two full time mechanics were employed to maintain, repair and ensure the safety of the vehicles. Strict standards relating to the serviceability of equipment, especially vehicles and communications equipment were enforced. Medical supplies, and first aid equipment were periodically inspected and replenished or replaced as a first priority of the field managers.

Quality Control and Quality Assurance Procedures

The survey was designed and implemented such that systematic quality assurance and quality control (QA/QC) systems and procedures were adhered to. Identified QA/QC issues became the focus of surveyor refresher training until adequately resolved. As a result, errors were systematically reduced and a high quality of survey information was ensured throughout the project.

Team Leaders continuously monitored the surveyors under their control, carrying out a 100% review of all reports submitted and field checks of at least 20% of completed survey questionnaires.

The original project design did not include Field Editors. However, as it became evident that Team Leaders were not capturing all errors due to the volume of their work, two surveyors were placed in the field office to assist in capturing errors and omissions.

The Field Manager and the Assistant Field Manager monitored survey rates and results; recommended necessary changes to survey procedures; carried out field checks of at least 1% of all completed survey questionnaires; and designed refresher training in consultation with the Database and QA/QC Managers.

Survey reports arriving at the Database faced comprehensive examination. Firstly, the database was designed such that incomplete or illogical entries were flagged. Simple administrative errors were corrected at the point of input; more serious errors produced a 'rejected report' for return to the field for correction. Secondly, every village and area sketch were checked by the GIS Operator and the Database Manager.

Independent of other field operations, the QA/QC Team conducted field assessments on 3% of the village survey reports and followed up on error reports and rejected survey questionnaires produced by the database. A total of 486 village survey reports were assessed by the QA/QC team. Every province and municipality in Cambodia was sampled. On average the work of each survey detachment was assessed 13.5 times. In no case was it ever shown that a detachment had not visited the village in question.

Not all errors were caused by the surveyors - in fact omissions by respondents produced the most serious errors. On occasion, a QA visit to a village elicited information on mine/UXO contamination that had not been made available to the surveyors. In some cases additional respondents provided the new information; in other cases previous respondents had simply forgotten certain facts or been made aware of additional information after the initial survey. In one instance a village leader identified an additional mined area to the QA team - this after being previously visited by the surveyors, the team leaders, and a field manager. It is estimated that there are 850,000 discrete data entries in the survey database. Based on QA-checked reports a probable error rate for information contained in the database was estimated to be most probably less than 1% of the data contained in the database.

Revisits / Resurveys

Village revisits and resurveys were undertaken in the normal course of the survey when any quality control procedure identified errors or omissions in a report. In addition:

The Database Unit

The Level 1 Survey Database Unit was comprised of six staff employed to undertake data entry, scan survey questionnaires; and produce maps. In addition a database analyst was seconded from the Cambodian Mine Action Authority (CMMA), to serve the rapidly expanding client base with requested maps and queries. All database staff reported to the Database Manager.

Eight (8) computers were used by the database unit on a daily basis and an additional 2 computers were available as backup. Once the demand for Survey results started to grow and larger format equipment was needed, CMAC lent the project a large format (A0) plotter and scanner in exchange for the LIS repairing and maintaining the equipment and allowing CMAC staff to use the machines as necessary. This arrangement worked extremely well and resulted in efficiencies being achieved by both the Survey project and by CMAC.

Maps were produced and updated using GIS software. Surveyor-produced sketches of contaminated areas were digitized onto base maps and verified using Landsat imagery or aerial photos. This was necessary as the only complete set of map sheets of Cambodia are 30 years old. Other data entry was carried out using an application developed in FOXPRO for Windows and tailored to the survey questionnaire. FOXPRO is available free from Microsoft and is accessible to all potential survey users.

This data entry application replaced the recommended IMSMA system, which was found unsuitable in light of conditions in Cambodia. A sophisticated consistency checking system was developed to identify and correct logical errors during data entry. We were unable to easily implement this system in IMSMA. An export routine was developed to transfer data to IMSMA to enable the Cambodia database to become part of the UNMAS global survey.

A variety of QA/QC controls are built into the Cambodia L1S data entry system:
  1. The application automatically checked for logical inconsistencies in the data.


  2. A smart user interface ensures operators enter all required information in the correct fields.


  3. Typing is kept to a minimum by using pop-up menus populated with data dictionaries.
In the case of an error, data can only be saved once the system administrator overrides the system with his password. Differences between field reports and database files are documented in an error file that is accessible from all screens.

Internal and external back-up systems were used to ensure that original data could not be lost.

National Gazetteer

The National Gazetteer of Cambodia lists all known provinces, districts, communes and villages and assigns an eight (8) digit code to each. The survey used the 1998 version that had been modified and updated by CMAC. Surveyors identified 474 new villages; modified information on an additional 577 villages; and deleted 25 villages from the list provided in the gazetteer. On occasion it was found that the village coordinates provided in the gazetteer were not accurate. While all changes are unofficial they are fully documented in the survey files.

Use of Topographic Maps, Satellite Images and GPS

The only complete map coverage of Cambodia is provided by 1:50,000 topographic maps published by the U.S. Army Topographic Command in 1960. These were the primary maps used by the survey. A partial new set of 1:100,000 maps covering central and northwestern Cambodia, recently produced from SPOT imagery was used. The survey also purchased Landsat 7™ scenes. Though some images were not useful due to heavy cloud cover, in general, these images greatly enhanced the capacities of the survey to accurately plot results. The new 1:100,000 maps and satellite images enabled field crews to navigate efficiently in areas where the US maps were particularly out-dated.

The survey corrected local administrative area files obtained from the Geography Department to ensure that the information is correctly shown on survey maps. These modified files are a survey tool only and are not official. All field personnel were equipped with GPS to assist in overcoming deficiencies in the available maps.

Access to Results

There has been widespread interest in the results of the Level One Survey from a large number of organisations in Cambodia with extremely varied technical and practical capabilities. In response, a variety of user applications were built using the same FoxPro software package used for data entry. Over the second half of 2001 this application took shape and was demonstrated to a wide range of users. Based on their comments, the application was steadily refined. This application and the survey information was provided to CMAC and to the CMAA. It was also provided to the Battambang Land Use Planning Unit (LUPU) to assist with their annual planning activities.

All other data requests were met through the provision of hard copy reports. These requests were the basis from which a standardized set of reports and maps, easily generated at the country, province, district, commune and village level, were designed.

A number of the requests received were extremely complex. The preferred option in responding to this type of request is to provide an overview of the application to the user and a digital copy such that the user can run queries and produce reports as desired. The application will run on any PC running Windows 95 or later versions, requires about 60MB of space on the hard drive and requires about an hour to understand its range of functionality and how to operate it. The database, application and maps are available on CD.

All maps produced by the Level One Survey include the following disclaimer in English, French, and Khmer:

This map indicates areas of known and suspected mine and UXO contamination. The information presented is not necessarily a complete indication of the actual extent and location of contamination. Other areas may contain mines and/or UXO as yet unknown.

Next Section - Acknowledgements
 Back to Table of Contents