SURVEY ACTION CENTER
6930 Carroll Avenue, Suite 240
Takoma Park, MD 20912-4468, USA
Phone: (301) 891-9192
Fax: (301) 891-9193
Website: www.sac-na.org

PROJECT ABSTRACT

The Landmine Impact Survey report summarizes the results of a socio-economic survey of the effects of landmines and unexploded ordnances (UXO) on communities in t he Awdal, Galbeed, Sahil, and T ogdheer regions (Phase 1). This sur vey was conducted from May 2002 to March 2003. This document is only one in a series of reports which collectively constitute the Global Landmine Survey Initiative. This initiative aims to catalog the socio-economic impacts caused by landmines and UXO and to store this data in a manner that supports strategic national planning and resource allocation decisions. The report on Phase 1 is designed to be read in conjunction with a document entitled The Global Landmine Survey Initiative, which describes the global project as well as the general methodologies used to conduct impact surveys. The project was managed by UNDP as part of its Somalia Mine Action Programme, executed on its behalf by UNOPS.

Funding for this survey has been provided by the European Community, the government of Sweden through the Swedish International Development Cooperation Agency (SIDA), the government of Finland, and the government of Canada through the Canadian International Development Agency (CIDA). Funding from the governments of Sweden and Canada was provided directly to the Survey Action Center. Quality assurance monitoring has been independently provided by UNMAS-UNOPS on behalf of the United Nations Certification Committee.

United Nations Mine Action Service,
Department of Peacekeeping Operations
Somalia

*N.B. The geographic boundaries depicted on the maps in this report do not imply endorsement or acceptance by the Danish Demining Group, the United Nations, or the Survey Action Center.

Introduction

Growing out of the wide collaborative efforts of the International Treaty to Ban Landmines, Landmine Impact Surveys (LIS) are executed to meet the needs of the international humanitarian mine action community, including donors, national authorities and mine action implementers.

The overall LIS vision is to "facilitate the prioritizing of human, material and financial resources supporting humanitarian mine action at the national, regional and global levels." To fulfill this vision, Landmine Impact Surveys are executed across the globe to the same rigorous standard.

Landmine impact surveys provide the three major partners of mine action- national authorities, donors and implementing agencies-with a common dataset. This data, as collected during the impact survey, offers clear improvements over past efforts in that it:

- Defines the entire problem in terms of scale, type, location, hazard and socioeconomic impacts experienced by local communities
- Improves national planning efforts by allowing for clear prioritization of resources
- Fosters development of national plans with well-defined immediate intermediate and end-state objectives
- Establishes baseline data for measuring performance

In sum, this implies nothing short of a major revision of how mine action programs are managed and how resources for such programs are allocated. Impact surveys are the first and most vital step in the overall transformation of humanitarian mine action. Impact surveys dramatically improve the quality of information available to support management decision making at all levels. The findings and information presented in this report are stored in the Information Management System for Mine Action (IMSMA) database and are intended to be descriptive in nature, providing the most comprehensive picture of the nature of the mine and unexploded ordnance (UXO) threat experienced by communities in Somaliland. While essential for national planning, this report is not a substitute for a national plan. It does not relieve national authorities or mine action professionals from their collective responsibility to gain a full understanding of the results of the survey and to use these results to set priorities, mobilize funding and allocate mine action resources in the most effective and rational manner. The survey has transformed the unknown into information and knowledge. The challenge now is for others to use this knowledge to bring about positive, constructive action.

As a global initiative with a stated goal of standardizing information across countries, Landmine Impact Surveys make a concentrated effort to ensure conformity of methods, procedures and processes. These are based on best practice in the fields of social science research and mine action. To ensure confidence in the results, impact surveys are supported by both internal and external quality control mechanisms. All surveys executed with the involvement of the Survey Action Center measure and score impacts in impacted communities in a uniform manner. The true value and nature of the impacts, however, cannot be ascertained by a quick tallying of colored dots on a map; instead readers should make a concentrated effort to comprehend all aspects of the landmine problem and then develop plans to address the problem.

Executive Summary

SUMMARY OF CONCLUSIONS

Phase 1 (Awdal, Galbeed, Sahil, and Togdheer regions) of the Landmine Impact Survey conclusively identified 357 mine-impacted communities, and 772 suspected hazard areas (SHA). The land contaminated by mines and/or by unexploded ordnance (UXO) directly impact the safety and livelihoods of an estimated 1.34 million people, and has led to the death or injury of 276 people in the last two years. The survey covered all communities to a high degree of confidence in these four aforementioned regions (Awdal, Galbeed, Sahil, and Togdheer-except its Boohuudle district). All 33 known communities in the Boohuudle district in the southeast corner of Togdheer were not surveyed due to security concerns. An additional 17 communities within the four Phase 1 regions were not surveyed due to being either abandoned or inaccessible and are assumed to be non-impacted by landmines/UXO. Only 231 of the 588 communities in the surveyed area were found not to be impacted.

The data collected during this effort affords extensive opportunities for research, analysis and project planning, and leads to several central conclusions relevant to mine action:

- Mine/UXO impact is spread widely across all areas in the surveyed regions, impacting all districts in the surveyed regions. Highly impacted communities tend to be clustered together.
- The majority of recent mine/UXO victims are children engaged in herding. Other activities that lead to frequent mine/UXO incidents include traveling along mined roads, and tampering with individual mines or pieces of UXO.
- A majority of the population in the surveyed regions is nomadic or seminomadic, and nomads have been specifically targeted in past conflicts.
- Suspected hazard areas (SHAs) in the surveyed regions are usefully divided into road and non-road types (574 and 198 SHAs respectively). Many roads are mined, but communities have in most cases adapted to this by creating diversions to mined segments. Non-road SHAs include former military camps, UXO stockpiles, and minefields that deliberately impact the livelihood of certain groups such as nomads.
- Most communities reported single-item UXO in their communities, but these were not counted as SHAs. However, as a considerable number of mine incidents occur outside SHAs, proper mine risk education is needed in these areas.
- The most prevalent resource blockages are of roads and pastureland, but the most serious blockages in terms of safety and socioeconomic security are of drinking water sources and irrigated cropland.
- Most SHAs in the surveyed regions contain a combination of anti-personnel mines, anti-tank mines and UXO, and are usually found on flat land covered by bush and scattered trees.
- One hundred twenty-six communities engaged in locally initiated mine action, usually mine or UXO clearance.

PROJECT OVERVIEW

A Landmine Impact Survey was conducted in the regions of Awdal, Galbeed, Sahil, and Togdheer from May 2002 to March 2003. Data collection was conducted between October 2002 and January 2003. This survey is the first phase of the Somalia Landmine Impact Survey project, which was first conceived in March 2001. The Survey Action Center (SAC) executed and the Danish Demining Group (DDG) implemented the survey in accordance with the principles and operating protocols established by the Survey Working Group.

A memorandum of understanding concerning the survey was signed between Danish Demining Group and the Ministry of Resettlement Rehabilitation and Reconstruction (MRR&R). The survey collaborated closely with MRR&R, the National Demining Authority, the Somaliland Mine Action Center, and the Vice President's office.

The European Community and the governments of Sweden, Finland, and Canada funded the survey. Technical support and material assistance was provided by the United Nations Development Program and the Geneva International Centre for Humanitarian Demining. The United Nations Mine Action Service and the United Nations Office for Project Services provided a quality assurance monitor to assist with the certification process.

DDG conducted the survey with four international staff members and more than 60 national staff during data collection. Field staff members were organized into two survey groups of one field supervisor, two field editors and eight interviewers each, and with four interviewers operating out of the Hargeisa office. Data collected was entered into the Information Management System for Mine Action.

SCOPE OF THE PROBLEM

Phase 1 of the LIS found all districts in these surveyed regions, a total of 357 communities, to be mine/UXO-impacted. While clusters of seriously impacted communities exist, the mine/UXO problem in the surveyed regions is widely dispersed, following roads and sites of armed conflict. The survey achieved 95% or better coverage among the districts covered.

IMPACT ON COMMUNITIES

A scoring mechanism categorizes communities by their degree of mine/UXO impact. Indicators for this mechanism include the number of victims within the past 24 months, blocked access to resources and the type of contaminating munitions. Based on this mechanism, the survey scored 45 communities as `high- impact' (13%), 102 as `medium-impact' (29%), and 210 as `low-impact' (58%) in the surveyed regions.

IMPACT ON SECTORS

The survey collected information regarding the types of resources that are denied to individuals and communities because of mine/UXO contamination in the surveyed area. The most frequently reported resources blocked by contamination were roads (74% of communities reported this as a blockage) and pastureland (36%). While drinking water and irrigated cropland were less frequently blocked- 18% and 6% of communities reporting these blockages respectively-they represent a much greater threat to safety and security, as alternatives to these resources are not readily available.

MINE INCIDENTS

The survey recorded 276 victims of mine/UXO incidents in the surveyed area in the past two years, of which 92 were recorded as fatalities. Seventy-five percent of mine victims were male, and just over half of all victims were between the ages of five and 14. Sixty-four percent of victims were harmed while herding livestock. Other significant activities that led to a mine incident include travel (10%), playing (9%), and tampering (7%). Less than one percent of recent victims (incident within 24-month period prior to LIS start) were military personnel. In the most general terms, the typical profile of an average victim in Somaliland is a male child engaged in herding livestock.

CAUSALITY

Statistical analysis generated a number of conclusions regarding causal factors that determine mine impact and the probability of mine incidents in a given community. Several factors tend to increase impact and the probability of recent victims. They include: more institutional connections to government, fewer commercial opportunities, proximity to other communities with recent victims, proximity of the community center to SHAs, involvement in Somaliland's civil war (1994- 1996), a larger population and significant UXO contamination.

CONCLUSION

The results of the survey clearly indicate that the Awdal, Galbeed, Sahil, and Togdheer regions suffer many adverse consequences from landmine and UXO contamination. The collected information will allow for the creation of a well-planned and targeted set of mine action initiatives. Safety and economic opportunity can be quickly restored to those communities that are suffering the most, with longer-term efforts aimed at concentrating resources where they will have the greatest benefit. The Landmine Impact Survey provides local authorities with the information it needs to put in place a program that will one day free it from the fear of landmines in the survey regions.

SPECIAL NOTE

This document has been produced with the financial assistance of the European Community and the governments of Sweden, Finland, and Canada. The views expressed herein are those of the Survey Action Center and Danish Demining Group and do not in any way reflect the official opinion of the European Community or the governments of Sweden, Finland, and Canada.

ABBREVIATIONS USED IN THIS REPORT

CIDA Canadian International Development Agency
CMA Cranfield Mine Action
DDG Danish Demining Group
DIMU Data and Information Management Unit
EC European Community
EOC Expert Opinion Collection
GICHD Geneva International Center for Humanitarian Demining
GISGeogr aphic mation Iynsftoerm S
GPSGlobal Positioning ystem S
HQ Headquarters
IMSMA Information Management System for Mine Action
LISLandmine Impact urvey S
MOU Memorandum of Understanding
MRR&R Ministry of Resettlement, Rehabilitation and Reconstruction
NDA National Demining Authority
NGO Non-governmental Organization
QAM Quality Assurance Monitor
QC Quality Control
SAC Survey Action Center
SHA Suspected Hazard Area
SIDA Swedish International Development Cooperation Agency
SMAC Somaliland Mine Action Center
SWG Survey Working Group
UNDP United Nations Development Programme
UNICEF United Nations Children's Fund
UNMAS United Nations Mine ervice Action S
UNOPS United Nations office Of operations Services
UXO Unexploded Ordnance

Phase 1 (Awdal, Galbeed, Sahil, and Togdheer regions) of the LIS identified 357 communities impacted by landmines and/or UXO in the four regions surveyed (see Map 2) and are included in this report. There is considerable variation among regions in terms of impacted communities: Hargeisa is the most contaminated region and Awdal and Sahil are the least contaminated. Within surveyed regions, all districts were impacted.

The survey also visited 231 non-impacted communities and recorded the population sizes in most cases. Approximately two out of every three people in the areas reached by the survey live in communities that are mine-impacted.

SETTLEMENT TYPE AND POPULATION SIZE

Mines and UXO in the surveyed regions primarily impact small rural communities. The nine impacted urban communities, however, accounted for almost 12% of the population suffering from mine impact. Community population size ranges over several orders of magnitude: the smallest impacted community contained 18 inhabitants and the largest had a population of 42,600. The median population size for settlements (2,400) is unusually large for such a dry country, though this high number might be caused by the fact that nomads may be counted as members of several communities simultaneously. The median core residential population for communities is 210.

MAP 2: LANDMINE IMPACTED COMMUNITIES (Sahil, Awdal, Togdheer, Galbeed)

Community status
-  Impacted (357 communities)
Non-impacted (231 communities)
-  Non-surveyed (abandoned or inaccessible)

 -  Non-surveyed region

 -  Non-surveyed district (Booduudle district)
- Region boundary
- District boundary

MAP 3: POPULATION SIZE IN IMPACTED COMMUNITIES

Population size 0-5,000 (282 communities)
5,001-15,000 (62 communities)
15,001-45,000 (13 communities)

Non-impacted (231 communities)
-  Non-surveyed Primary road
- Hargeisa

MAP 4: NOMADIC COMPONENT OF IMPACTED POPULATIONS
Nomadic component
-  0-25% nomadic
-  26-50% nomadic
-  51-75% nomadic
-  76-100% nomadic
Non-impacted
-  Non-surveyed
- Hargeisa

THE IMPACT OF MINES AND UXO ON NOMADIC PASTORALISTS

Pastoralism is the dominant means of livelihood in the survey regions: the majority of people directly or indirectly obtain their livelihoods from animal husbandry (see Map 4). Camels, goats and sheep are suited to dry land and semi-desert conditions. Pastoral life is structured around seasonal migratory patterns determined by rainfall and pasture. In the wet seasons (gu, April to July and dayr, October to December), pastoralists move with their animals into the Haud, arid zones where grass is plentiful and water holes and seasonal rivers are full. In the dry seasons (Hagaa, July to October and Jilaal, December to April) watering animals requires daily access to wells and pastoralists move their herds to the Ogo, less arid regions. During the dry season most families split into two groups. Herders, mostly unmarried youngsters, stay behind on the pastureland and come to the home wells about once a month to satisfy the less frequent, but more substantial needs of the camels. Women, small children and the elderly, move with the burden camels and the smaller livestock close to home wells. Because of their migratory lifestyle, pastoral families live in mobile huts, a number of which make up a settlement. Women and children take care of the goats and sheep that graze close to the hamlets.

Density of UXO contamination

Although UXO contamination is common in the surveyed regions, not all findings of UXO in communities were significant enough to be included in the survey and were not scored. While most communities surveyed had at least one or two pieces of UXO, these were often thrown under bushes or collected in small caches out of reach of children. The survey did not record these small UXO discoveries as suspected hazard areas, and so communities with only these hazards are not referred to as `impacted'. Instead, interviewers filled out a spot clearance request with enough information for a clearance team to locate and remove or destroy the UXO involved.

UXO contamination generally occurs at any of four density levels:

1. Single items, found in almost every village.
2. Small caches of ammunition or gathered UXOs, often right in the middle of the village.
3. Battle area contamination, with widespread UXO littering.
4. Larger stores of abandoned or partly blown up UXO resulting in substantial UXO distributed over a relatively well-defined, sometimes large, area.

Only UXO contamination of type 3 and 4 would result in establishment of a suspected hazard area for the survey.

Young people from permanent settlements may go away for several months of the year with the animals, but do not identify themselves as nomadic. On the other hand, women in nomadic communities will often cultivate the land when they stay for longer periods in the same place. Both types of communities rely on animal husbandry, or a mix of cultivation and animal husbandry. The survey results show that 96% of the surveyed communities have animal husbandry as one of their economic activities, 68% rely on both agriculture and animal husbandry, and 29% rely on animal husbandry only. The nomads themselves define a nomadic person as someone who owns livestock and follows the rain.

By their very methodology, Landmine Impact Surveys are biased towards permanent settlements. Phase 1 of the LIS was not equipped to undertake a more comprehensive study of the nomadic population. However, the survey did manage to gather some data on the nomadic population and partially overcame the challenge of characterizing the impact of the presence of mines and UXO on nomads. The survey registered the number of nomads participating in the community interviews, as well as the size of the nomadic population in the surveyed communities.

As anticipated, the survey found that nomads camp close to permanent settlements and towns, and the data showed that they prefer settling close to bigger villages. Nomads participated in the interviews and were therefore represented in the survey. Twenty-seven percent of the informants defined themselves as nomads. Nomads were present in the group interviews in 217 (61%) of the surveyed communities. The survey recorded that 54% of the population in the areas surveyed were described by informants as nomadic. Exact nomad population statistics do not exist, although the government estimates that 55% of the total population is nomadic. The United Nations Children's Fund (UNICEF) assumes that 80% of the population in Somaliland and Puntland (northeastern Somalia) are fully or partly nomadic.

Nomads are exposed to hazards when they enter new areas. Most suspected hazard areas in the surveyed regions are unmarked and several respondents reported that new people coming into the village are in danger because they do not know where the mines are. Some specify that it is especially a problem during the rainy season when people come to stay, and in settlements through which many nomads pass.

Mines were specifically laid to block access to nomads' water sources. Many berkeds-ground cisterns-were mined to disrupt the life of the nomads. The statistics on recent victims show that children and young men are at high risk when they are herding. They may be tempted to play or tamper with mines and UXO either out of curiosity or for collecting scrap metal.

GEOGRAPHIC DISTRIBUTION OF IMPACTED COMMUNITIES

Mine-impacted communities are found throughout Somaliland. In the surveyed area, the largest concentration of impacted communities is found in the Galbeed region. Map 5 shows the distribution of impacted communities and their impact in the surveyed area. The main cluster is the belt running southwest-northeast in the Galbeed region, but smaller clusters are also found in other regions.

HISTORY OF CONTAMINATION BY MINES AND UXO

The surveyed regions have suffered from several conflicts that scattered munitions through out the country, beginning with Italian occupation in World War II. Most of the UXO/mine contamination, however, occurred during three distinct periods.

1977-78: The Ogaden war with Ethiopia

In 1977, Somalia attacked Ethiopia to capture the Ogaden region, which according to Somali perceptions was part of their country. Initially the offensive drove deep into Ethiopia, but the Ethiopian forces managed to stop the offensive and push the Somali forces back to the border where they protected a number of military camps with mines. All of the actual fighting occurred inside Ethiopia. MAP 5: DISTRIBUTION OF IMPACTED COMMUNITIES (Sahil,Awdal,Togdheer,Galbeed)

Community impact
-  High (45 communities)
-  Medium (102 communities)
-  Low (210 communities)
Non-impacted (231 communities)
-  Non-surveyed
- Hargeisa

1988-1991: The Somaliland Liberation war

This conflict has had the greatest impact on the mine and UXO contamination in Somaliland. The war was fought between the Somaliland National Movement (SNM) from bases in Ethiopia and the Somali government in Mogadishu. The government response was to mine roads leading from the Ethiopian border. In May 1988, the conflict escalated with SNM's surprise attack against Burao and Hargeisa. The fighting engulfed most of Somaliland before it ended with the fall of Siad Barre, the Somali president, in January 1991. During this phase, both sides used mines to protect important installations and to restrict the movement of the opposing side. This resulted in the mining of roads and tracks, and areas around military camps and other installations. The nomadic population was also targeted with widespread mining of their water sources, disrupting their normal migratory routes.

At the end of this conflict, when the Somali armed forces withdrew from Somaliland territory, vast amounts of ammunition were abandoned and have contributed significantly to the Somaliland UXO problem.

1992 and 1994-1996: Civil wars

In 1992 a six-month conflict broke out in the port town of Berbera and extended along the major road towards Hargeisa and Burao. In November 1994, war erupted in the city of Hargeisa and engulfed large parts of the region before ending in late 1996, also contributing to the mine and UXO contamination. MAP 6: ERA OF MOST RECENT (Sahil, Awdal, Togdheer, Galbeed) MINE LAYING Mines/UXO contamination - Laid in or since 1988 (267 communities) - Laid in or before 1987 (90 communities) Non-impacted (231 communities) - Non-surveyed - Hargeisa

Map 6 shows the era in which the most recent mine was laid in each impacted community. Table 1 shows the contamination caused by the different conflicts in the region's history.

SUSPECTED HAZARD AREAS (SHAS)

The Survey Working Group (SWG) protocols characterize all areas known or thought to have landmines and/or UXOs as Suspected Hazard Areas (SHAs). The survey identified 772 distinct suspected hazard areas. The suspected hazard areas (SHAs) can be divided into road and non-road types. Five hundred and seventy-four SHAs were identified as road-related. Thus, there are almost three times as many road as non-road SHAs. Two hundred and seventeen communities (61%) were only impacted by road SHAs, and 84 additional communities have at least one mine-impacted road segment in addition to other blockages. Only 56 communities are solely impacted by non-road SHAs (Table 2, see previous page). However, roads do not always create a socioeconomic impact-often the villagers have been able to establish a detour without great inconvenience. The average significant detour is 7.9 kilometers long. Map 7 shows the number of SHAs at each impacted community in the surveyed districts.

TABLE 1: SUSPECTED HAZARD AREAS, BY CONFLICT

Number Non-road suspected Road suspected of hazard areas 
hazard areas Impacted communities Count Percent Count Percent population
Ogaden war (O) only 2 2 1.0% 0 0.0% 1,518
Somaliland Liberation war (SLL) only 208 97 49.0% 341 59.4% 732,852
Somaliland National Movement ar w (SNM) 2only 1 0.5% 30.5% 4,602
Other war only 2 2 1.0% 1 0.2% 1,308
O + SNM wars 1 0 0.0% 2 0.3% 3,155
O + SLL wars 18 16 8.1% 25 4.4% 74,925
SLL + SNM wars 91 46 23.2% 163 28.4% 373,259
O + SLL + 5 SNM 8 4.0% 30.5% 9,527
Other wars and one or two of the named wars 19 19 9.6% 26 4.5% 104,140
All wars 2 1 0.5% 5 0.9% 9,960
None 7 6 3.0% 5 0.9% 25,360
TOTAL 357 198 100.0% 574 100.0% 1,340,606

1 89 3 4 -
2 85 17 22
329 4 34
4 10 1 17
5 4 - 3
6 - - 6
7+ - - 2
TOTAL 217 56 84
Sahil
Awdal
Togdheer
Galbeed

MAP 7: NUMBER OF SUSPECTED HAZARD AREAS IN IMPACTED COMMUNITIES The diameter of the symbol represents the number of Suspected Hazard Areas- ranging from one to six, at each of the 357 impacted communities.

Non-impacted (231 communities)
-  Non-surveyed
- Hargeisa

- The number of recent victims (incidents within 24 months prior to survey start)
- The different types of socioeconomic and institutional blockage present
- The nature of munitions present (landmines and/or UXO)

The score is then used to classify communities' impacts as low, medium or high. Thirty-eight communities with the lowest impact-an impact score of two- reported only the presence of landmines and no recent victims or socioeconomic blockages. In these cases, the only resources that were at all blocked were roads that were easily avoidable by choosing alternate routes. The community with the highest impact score (29) had an incident in which a bus hit an anti-tank landmine, killing 13 people. Figure 1 presents the frequency distribution of impact scores. Two hundred and ten (59%) of the 357 impacted communities are classified as low-impact, 102 (29%) are medium-impact and 45 (13%) are high-impact (Figure 2, next page).

Seventy-five percent of impacted communities received a score of seven or less. The most common score was three, and the median score was five.

FIGURE 1: DISTRIBUTION OF IMPACT SCORES

- Low impact communities (210)
- Medium impact communities (102)
- High impact communities (45) Number of communities Impact scores Communities are classified as high-, medium-, and low-impact based on a framework that is common to all Landmine Impact Surveys.

POPULATION BY IMPACT CATEGORY

Approximately 13% of communities were identified as high-impact, 29% as medium impact, and 59% were identified as low-impact communities (Table 3).

DEMOGRAPHY OF RECENT VICTIMS

The survey identified 276 recent victims, 77% of which were male. The majority of victims were between 5 and 14 years old (Table 4, below and Figure 3, next page).

Most recent victims were farmers or homemakers at the time of the incident.Significantly, military personnel accounted for less than one percent of recent victims (Table 5, next page).

The vast majority of recent victims were herding at the time of the incidents. The majority of all incidents involved children tending animals (Table 6, next page). A large number of people have incidents while traveling, which corresponds to the fact that a third of all incidents are related to mine-impacted roads. Thirty

FIGURE 2: IMPACT CLASSIFICATION

(in surveyed districts)
High
Medium
Low
0 50 100 150 200 250
Number of communities

TABLE 3: COMMUNITIES, BY IMPACT CLASSIFICATION

Score Range Classification Count Percent
11 + High   45  12.6%
6-10 Medium 102 28.6%
1-5  Low    210 58.8%
TOTALS 357  100.0%

Male Female All
Age Number Percent of all victims Number Percent 
of all victims Number Percent of all victims
1-4 3   1.1% 1 0. 4% 4    1.4%
5-14    102 37.0% 45 16.  3%  147 53.3%
15-29   69  25.0% 10 3.6% 79  28.6%
30-44   26  9.4%  5  1.8% 31  11.2%
45-59   7   2.5%  0  0.0% 7   2.5%
60 +    5   1.8%  2  0.7% 7   2.5%
Unknown 1   0.4%  -  -    1   0.4%
TOTAL   213 77.2% 63 22.8% 276 100.0%

TABLE 5: RECENT VICTIMS, BY GENDER, MILITARY/CIVILIAN STATUS, AND CIVILIAN OCCUPATION

Occupation Male Female All
before incident Number Percent of all victims Number 
Percent of all victims Number Percent
of all victims
Military    2 0.7% - - 2 0.7%
Civilian    211 76.5% 63 22.8% 274 99.3%
Crop cultivation 8 2.9% - - 8 2.9%
Herding    152 55.1% 53 19.2% 205 74.3%
Household work 1 0.4% 4 1.4% 5 1.8%
Other      22 8.0% 0 0.0% 22 8.0%
Trading    4 1.4% 0 0.0% 4 1.4%
Unemployed 24 8.7% 6 2.2% 30 10.9%
TOTAL      213 77.2% 63 22.8% 276 100.0%

Male Female All
Number Percent of all victims Number Percent of all
 victims Number Percent of all victims
Military 2 0.7% - - 2 0.7%
Civilian 211 76.5% 63 22.8% 274 99.3%
Collecting food/water 4 1.4% - - 4 1.4%
Herding 127 46.0% 51 18.5% 178 64.5%
Household work 1 0.4% 5 1.8% 6 2.2%
Other 8 2.9% - - 8 2.9%
Playing 24 8.7% 2 0.7% 26 9.4%
Tampering 16 5.8% 3 1.1% 19 6.9%
Travel 27 9.8% 1 0.4% 28 10.1%
Unknown 4 1.4% 1 0.4% 5 1.8%
TOTAL 213 77.2% 63 22.8% 276 100.0%

percent of all incidents occurred outside the recorded SHAs. These incidents are probably caused by people removing items, particularly UXO, from SHA (Table 7). Child herders, constituting the largest group of recent victims, are often injured or killed when they play or otherwise tamper with single-item UXO.

Ninety-two of the 276 recent victims received fatal injuries. The fatality rate was higher for males (35%) than for females (27%). A large proportion of the survivors received emergency medical care, but very few received rehabilitative or vocational therapy (Table 8).

TABLE 7: TYPE OF SUSPECTED HAZARD AREA, BY INCIDENT

Outside suspected Road suspected Non-road
hazard area hazard area suspected hazard area
Number Percent of all victims Number Percent of all victims Number Percent of all victims
Not fatal 58 21.0% 45 16.3% 81 29.3%
Fatal 25 9.1% 35 12.7% 32 11.6%
TOTAL 83 30.1% 80 29.0% 113 40.9%

Occupation of Male Female All
survivors Number Percent of all victims Number 
Percent of all victims Number Percent of all victims
Military    2 1.1% 0 0.0% 2 1.1%
Civilian    136 73.9% 46 25.0% 182 98.9%
Cultivation 7 3.8% 0 0.0% 7 3.8%
Herding    90 65.2% 37 80.4% 127 69.0%
Household work 1 0.5% 3 1.6% 4 2.2%
Other      17 9.2% 0 0.0% 17 9.2%
Unemployed 20 10.9% 6 3.3% 26 14.1%
Unknown    1 0.5% 0 0.0% 1 0.5%
TOTAL     138 75.0% 46 25.0% 184 100.0%

TABLE 8: TYPE OF CARE RECEIVED BY SURVIVORS

Male Female All
Type of care Number Percent of all victims Number 
Percent of all victims Number Percent of all victims

Emergency      106 59.2% 35 19.6% 141 78.8%
Rehabilitative 2  1.1% 2 1.1% 4 2.2%
Vocational     0  0.0% 0 0.0% 0 0.0%
Other          6  3.4% 2 1.1% 8 4.5%
None           16 8.9% 8 4.5% 24 13.4%
Unknown        2  1.1% - - 2 1.1%
TOTAL          132 73.7% 47 26.3% 179 100.0%

The gender and current occupation of the survivors are given in Table 9 (previous page). A majority of the impacted communities and victims are in the district of Hargeisa as shown in Table 10, followed by the district of Togdheer. Map 8 (next page) shows the distribution of communities with recent victims.

TABLE 10: IMPACT AND VICTIM DATA, BY DISTRICT

Victims Communities Population Recently Recently Killed Injured All

District impacted impacted killed injured earlier earlier victims

Baki               5  26,119 0 0 4 3 7
Boorama            14 33,380 5 11 35 44 95
Dila               4  12,204 0 7 0 0 7
Lughaya            14 71,600 1 3 95 131 230
Sylaca             22 57,024 22 23 39 32 116
TOTAL              59 200,327 28 44 173 210 455
Allay Baday        4  29,110 1 2 15 11 29
Balli Gubadle      18 36,204 8 19 82 108 217
Cadaadley          11 33,685 2 4 50 46 102
Daarasalaam        12 31,532 0 5 13 18 36
Fara Weyne         17 44,542 2 9 24 28 63
Gabiiley           31 97,284 4 5 40 106 155
Hargeisa           26 107,868 14 15 79 79 187
Sabawanaag         17 45,784 1 3 39 69 112
Sallaxley          19 72,892 7 18 175 188 388
TOTAL              155 498,901 39 80 517 653 1289
Berbera            16 52,137  0 2 84 127 213
Bullo Xaar         4  12,120  0 1 0 1 2
Madheera           12 22,218  1 5 9 35 50
Sheekh             26 82,263  8 15 53 56 132
TOTAL              58 168,738 9 23 146 219 397
Burco              24 158,904 5 15 52 68 140
Durruqsi           5  20,520  0 3 38 96 137
Oodweyne           27 125,390 3 17 94 112 226
Qoryaale           19 84,918  8 2 51 92 153
Sheekh Xasan Geele 10 82,908  0 0 43 87 130
TOTAL              85 472,640 16 37 278 455 786
GRAND TOTAL        357 1,340,606 92 184 1114 1537 2927

MAP 8: RECENT VICTIMS IN IMPACTED COMMUNITIES

Community status
-  Recent victims (111 communities)
-  No recent victims (246 communities)
Non-impacted (231 communities)
-  Non-surveyed Primary road
- Hargeisa

Analysis of Blockage Impact

TYPES OF BLOCKAGE

Key informants in impacted communities most frequently identified roads, water, rainfed cropland and pasture as socioeconomic blockage (Table 11).

Roads were the most frequently reported resource blockage, and were often ranked by communities as the most significant. Many roads are mined but in most cases, detours have been established that minimize or eliminate impact. The distribution of road blockage types by community is shown in Map 9.

TABLE 11: PERCENTAGES OF COMMUNITIES REPORTING BLOCKED ACCESS

Communities

Areas of blocked access impacted
Minor roads 74%
Pasture 36%
Roads to administrative centers 23%
Rain-fed cropland 21%
Water, other uses, including irrigation 18%
Drinking water 18%
Non-agricultural land 17%
Other infrastructure 7%
Irrigated cropland 6%
Housing 4%

MAP 9: ROAD BLOCKAGE TYPE,BY COMMUNITY

Road blockage type
-  Blockage to Administrative Center AND other road blockage (79 communities)
-  Blockage only to other places (186 communities)
-  Blockage only to Administrative Center (2 communities)
-  No road blockage (90 communities)
Non-impacted (231 communities)
-  Non-surveyed Primary road
- Hargeisa

Table 12 presents the numbers of communities and SHAs associated with different blockage types.

FIGURE 4: BLOCKED AREAS: PREVALENCE AND VALUE 0 25 50 75 Percentage of communities Low-value land Rainfed cropland Irrigated cropland Housing areas Roads Water sources Food-producing areas: Listed in order of decreasing prevalence of blockage but increasing land value Non-food producing areas: Valuation difficult; listed in order of increasing importance, as expressed in community interviews

TABLE 12: RESOURCE BLOCKAGE STATISTICS

Area of blocked access Communities impacted Road 
SHA Non-road SHA Total SHA
Minor roads      265 452 22 474
Pasture          129 87 111 198
Roads to administrative centers 81 86 2 88
Rain-fed cropland 76 48 62 110
Water, all uses   63 32 45 77
Non-agricultural land 59 31 47 78
Other infrastructure 24 5 25 30
Irrigated cropland 23 13 13 26
Housing            15 2 15 17

TYPICAL COMBINATIONS OF IMPACTS

The Landmine Impact Survey seeks to understand the socioeconomic impact of blocked resources on communities. The survey has revealed certain common groups of blocked resources. These are called impact clusters. Clusters exhibit different recent victim counts averaged across the communities within a cluster, ranging from 0.26 to 1.45 per cluster community (see Table 13).

Type `A' exhibits the presence of landmines in both local roads and those leading to administrative centers; this characterizes 54 communities. Type `B' includes communities without any road blockage. Some of its members reported blocked access to housing, others of non-agricultural land. This type characterizes 63 communities.

Type `C' communities reported local road blockages as the only impact type. This type has higher recent victim counts than in A and B due to a few incidents involving a large number of recent victims. There are 86 such communities. Type `D' includes a variety of blockages in addition to roads, and characterizes 49 communities.

Types `E' and `G' involve water. Communities of type `E' reported water blockage in more than 80% of cluster members. Type `G' communities, in addition to reporting problems with blocked water sources, frequently reported serious local road, pasture and non-agricultural land blockages. Type `E' characterizes 43 communities; type `G' characterizes 20 communities.

Type `F' involves the blockage of rain-fed cropland and represents 42 communities.

TABLE 13: IMPACT CLUSTERS (Omitted)

Note: Cells in black designate blockage types that are present in more than 80% of cluster members. Gray stands for blockage types present with at least 67% of overall frequency. A number of results follow:

- The major factors differentiating impact are roads, pasture, rain-fed cropland, water, and non-agricultural land.
- `Other' roads (not leading to an administrative center) are a distinct impact category, reported, without other blockages, in almost a quarter of all impacted communities. The effect of these blockages on the communities is difficult to evaluate but could be minor: alternatives should be readily available, and many of the reported recent victims are from outside the community.
- Communities of types `E', `F' and `G'-characterized by water and rain-fed cropland blockages-have the highest recent victim averages, suggesting that these resources have few alternatives.
- Blockages of residential areas, irrigated cropland or other infrastructure occur among communities of all types except `C'. However, none of these blockage types discriminates strongly enough to define a separate impact cluster. The importance of access to water will be highlighted further in the analysis of mine incidents.

Jaleelo community

Jaleelo is a medium-impact community 40 kilometers northeast of Hargeisa. Similar to many villages visited by the survey, it is a dispersed rural settlement dependent on cultivation and animal husbandry, and has a seasonal nomadic population.

The community has three mined roads and an area with dispersed UXO. There have been several mine incidents, but none in the past two years. One involved the death of a man called Abdi, whose car slid off the road and hit an anti-tank mine. Abdi was a former soldier and had initiated local mine clearance efforts through community mobilization. Since his incident in 2000, no further mine clearance has been attempted.

UXO cache in Jaleelo Jaleelo Hargeisa

Summary of Past Mine Action

The survey gathered information about the nature of mine action activities- mine risk education, marking, surveys, victim assistance and mine clearance- that have already taken place in impacted communities. Their presence, whether implemented by outside agencies or locally initiated, is the foundation upon which future mine action programs can be built. These activities also reveal the magnitude of potential needs not yet fulfilled.

Table 14 represents the total number of communities engaged in some type of mine action activity during the past two years, and the incidence of mine action by external agencies.

CLEARANCE

Mine clearance initiatives by local people constitute the most common mine action activity reported. These initiatives included the collection or removal, at risk of injury or death, of dangerous items that represent an imminent threat or serious socioeconomic hardship. Communities will often place these devices in a safe place to prevent incidents, such as abandoned buildings, on roofs, or they may be thrown into bushes or buried.

The practice of mine/UXO destruction through incineration is even more dangerous. More than one in three communities has been involved in mine clearance. It is important to note that the local demining has removed some of the mines/UXO, but not in a systematic manner and other mines and UXO may remain. Clearance of SHAs by the external mine clearance agencies working in these regions-Danish Demining Group, HALO Trust and St. Barbara Foundation- has been reported by one in ten impacted communities in the last two years. These organizations have all been operational in these regions since 1999. In addition, Rimfire implemented mine clearance in these regions from 1991-1993 and Mine Tech and Greenfields under the United Nations Office for Project Services (UNOPS) in 1998.

TABLE 14: INCIDENCE OF MINE ACTION,BY SOURCE AND ACTIVITY

Communities Percent
External mine action agencies 56 15.7%
Marking and/or survey 41 11.5%
Clearance             34 9.5%
Mine risk education   5 1.4%
Victim assistance     2 0.6%
Community-initiated mine action 153 42.9%
Marking and/or survey 36 10.1%
Clearance             126 35.3%
Mine risk education   6 1.7%
Victim assistance     0 0.0%
Some mine action      171 47.9% (external and/or community level)

Note: Communities may have multiple types of mine action, and so may be included in more than one category.

MARKING AND/OR SURVEY

Survey and marking of SHAs have taken place in approximately 20% of impacted communities. Half were local initiatives and half were executed by external demining organizations.

MINE RISK EDUCATION

Only 11 communities reported mine risk education activity: half from external mine action agencies and half from local initiatives. These were executed almost exclusively through school presentations and posters.

VICTIM ASSISTANCE

Two communities reported some assistance to victims, and in one case an artificial leg was provided.

Factors Influencing Mine Clearance

This section describes the various types of terrain, vegetation and soil types found in the 772 suspected hazard areas (SHAs). It also illustrates the variation in types of ordnance found.

There is only limited range in variation of terrain and vegetation among the SHAs identified. The section begins with a general topographic description and then focuses more narrowly on four factors that influence mine clearance operations. The total land area in the surveyed regions is about 71,400 square kilometers.

TOPOGRAPHY

There are four distinct topographical zones between the Ethiopian border in the south and the Gulf of Aden in the north. These include a) elevated desert plateau with cross-cutting ravines to the south, b) a long string of arid hills lying east-west across the middle of the country, c) low-lying desert plateau in the north and d) sandy desert in the extreme north, by the sea.

Most of the communities are found in the high desert plateau, where water found in ravines and gullies allows for cultivation. Fewer communities are found in the hills or the low-lying plateau, where water is scarce. The sandy desert in the north is populated by fishing villages on the coast, and most of these are relatively inaccessible.

The survey found that permanent settlements tend to be built on flatter ground, especially in the higher desert plateau, and SHAs are concentrated around these communities and the road network that connects them.

VEGETATION AND GROUND PROFILE TYPES

Ground vegetation and elevation are two very important factors of mine clearance. About two-thirds of all SHAs are found on flat land in the elevated desert plateau to the south, where most communities are located. A majority of SHAs identified by the survey had vegetation consistent with the semi-arid desert: lowlying bushes with a scattering of trees.

The `bushes and trees' vegetation type is the most prevalent for both roadrelated and non-road SHAs, suggesting that contaminated roads have been mined for a relatively long period of time and hence overgrown.

CONTAMINATED SURFACE AREA, BY SOIL TYPE

Soil type is a significant factor for mine clearance, especially if mine identification is done by manual soil excavation and prodding.

Though there is arable soil in the gullies, washes and stream beds that crosscut Somaliland, the vast majority of suspected hazard areas lie on flatter ground and are made up of either sandy soil, rocky soil, or both.

CONTAMINATED LAND, BY ORDNANCE TYPE

Operational planning for mine clearance depends on the type of ordnance involved as well as the operating environment. The absence of intensive aerial bombardment during military conflict in the surveyed regions has meant that the majority of ordnance types are ones used in ground military operations: anti-tank and anti-personnel mines and munitions such as mortars and hand-grenades.

Geed Balaadh community

Geed Balaadh is a medium-impact community 40 kilometers south of Hargeisa, near the Ethiopian border. Ibrahim Mohammed Haji, a member of this community, left to work in Norway in 1987 and returned to Geed Balaadh in 2002. He owns some fertile land there. Haji bought a tractor with the money he earned in Norway and was hoping to use this for farming.

During the Somaliland Liberation War, Geed Balaadh was a government stronghold and was regularly under attack. When he returned to the village, Haji found that his land was mined and that there had been a mine incident there. He was disappointed to discover that he could not use the tractor he had purchased with all his savings. After three months of struggling financially, he went back to Norway.

Community Background and Mine Effects

COMMUNITY ADAPTATION

This section deals with the responses of impacted communities to their mine/ UXO problems. Community-based mine clearance has occurred in 126 out of 357 communities, suggesting that there has been a great deal of locally initiated response to mine contamination. This compared to 34 interventions by international non-governmental organizations (NGOs), although it could be assumed that each NGO intervention was more comprehensive.

Where mine hazards exist, communities continue adapting to the threat. This adaptation includes discussion among community members on the threat and responses, the restrained use of resources in suspected hazard areas and the continued search for alternatives.

The response depends not only upon the nature of the hazard, but also upon the social context of the community. Historical factors as well as current organizational capacity and resources define group and individual adaptation.

Unfortunately, given that the mine/UXO problem impacts a diverse set of communities; it is difficult to find explanations that are universally applicable and that can characterize the degree of adaptation.

A potential indicator is the ability to avoid mine incidents. The risk of incidents should decrease when knowledge of mine location, alternatives to blocked resources, and mobilization of clearance resources increase. This assumes that recent victim data is reasonably accurate and also that communities possess these capacities at different levels.

Statistical methods can be used to find associations between recent mine incidents and the social characteristics of communities. Establishing these correlations is extremely important for directing mine action. This exercise attempts to quantify community adaptation in terms of certain variables, with the understanding that these correlations can never replace the experiences of individual communities.

FACTORS CONSIDERED FOR THE ADAPTABILITY OF THE COMMUNITIES

Many factors affect a community's ability to deal with mines and UXO, and the survey could not gather information on all of them. The survey did, however, collect data on a number of variables that may be relevant to community adaptation.

1) Size of the current population

Other things being equal, a larger population would mean more chances of interacting with the hazard. In addition, poor communities with large populations face greater pressure on available resources. Thus, a larger population should mean more mine incidents. However, it is possible that mines impact only a few people in a community, in which case the relationship between population size and incidents is not that direct.

2) Institutional endowment of the community

If a community has more local institutions, it probably has a greater ability to respond to the mine problem. The presence of institutions may be related to the amount of mining and UXO littering because these institutions are the resources that armies fight over. In Somaliland, communities with more institutional density tend to have larger suspected hazard areas.

3) Extent to which mines block critical resources

Communities that are denied access to critical resources can be expected to suffer more incidents. In the dry climate, and with the high community emphasis on water and irrigated cropland, blockage of these resources increases the number of recent victims considerably.

4) Legacy of the conflict

This factor is expressed in three dimensions: the estimated surface area of total SHAs, the intensity of regional landmine use as measured by the distance to the nearest other community with some recent victims, and the communities' exposure to hostilities and mining. In some models, the location of contaminated areas in relation to social space, as measured by the distance from the center of the community to the nearest SHA, was also considered.

THE INSTITUTIONAL ENDOWMENT OF THE COMMUNITY

Social science assumes that more complex institutional formations exchange information more effectively and can thus help communities solve problems.

However, this needs to be tested in the field against the problems of bureaucracy and social barriers and the possibility that communities with more institutions attracted more intense conflict. To measure the institutional complexity of the surveyed communities, the survey collected information on certain institutional indicators. Statistical analysis of 19 indicators revealed highly significant correlations.

These 19 indicators cluster into four groups. The groups reflect:

- the relative strength of the connection with government institutions;
- the strength of civil society;
- the degree of commercial opportunities; and
- the modernization of local water supplies and the presence of veterinary services.

The characterization of these four factors is somewhat arbitrary. If a community has access to health care facilities, fuel, radio and a secondary school, then it will have better communication with local and national authorities. These categories are useful because the variables of several indicators are closely linked to the strength of community-government linkages. The second cluster, broadly consisting of associational resources, reflects a community's ability to develop solutions to problems and take initiatives. A third factor is the presence of organized markets and of permanent transport facilities.

The fourth consists of water resources and veterinary services. Table 15 summarizes the frequency of these indicators.

THE LEGACY OF CONFLICT

Another of the mine impact indicators is exposure to the various armed conflicts, or similarly, the number of years since contamination. The assumption is that if mines/UXO have been around a long time, the community has had more time to adapt. Therefore self-reported exposure to the Ogaden War (1977), the Somaliland Liberation War (1988-91), and the civil wars (1992 and 1994-96) is used to characterize the legacy of conflict.

Intensity of armed conflict is spatially concentrated and extends to the density of mining or UXO contamination: the more intense the conflict, the greater concentration of mines or UXO. Therefore, incidents in one community may predict incidents in neighboring communities. The distance between a community and the nearest other community with some recent incident victims approximates the above measure. In Somaliland, this median distance is 11.9 kilometers.

TABLE 15: COMMUNITY MODERNIZATION INDICATORS

Percent impacted communities
Modernization factor Indicator with this feature

Strength of Ordinary village 94%
government connection
Medical facility 24%
High frequency radio 18%
Police station 17%
Intermediate school 15%
Electricity 11%
Auto fuel 11%
Telephone 5%
Civil society Primary school 36%
Women's association 36%
Youth association 31%
Mobilizing association 29%
Commercial opportunities Regular transport 75%
Permanent market 74%
Livestock market 47%
Improved water supplies Berkeds 68%
Irrigation pumps 22%
Water kiosks or hand pumps 20%
Veterinary service 12%

Communities with suspected hazard areas close to the village center should have higher levels of resource blockages. The typical distance between SHA and village center is 500 meters. While all but two of the impacted communities have some mines, only about a third reported significant UXO contamination.

FACTORS INFLUENCING THE PROBABILITY OF MINE INCIDENTS

As noted previously, 111 of the 357 impacted communities suffered one or more recent incidents. There were a total of 276 victims. The number of victims in a particular incident depends on the situation and is probably not related to the social structure of the community. The key informants may not be accurate in stating the number of mine incidents, so this analysis is limited to investigating the association of risk factors with the number of recent victims.

The number of victims is a good indicator of whether a community has adapted successfully to contamination. Their ability to avoid, or greatly reduce, the number of mine incidents reflects the degree to which a community has developed strategies for avoiding interaction with the threat.

In this analysis, two types of influences are distinguished. As described in Figure 5, the green bars represent the strength of factors that improve the chances to avoid incidents completely. Blue bars signify that the associated factor is helpful in reducing the number of victims, without completely avoiding incidents.

Among the factors with blue bars, a linkage to community- government is the strongest factor. Town and villages with stronger government connections tend to have more recent victims, even controlling for population size. The distance from the village center to the nearest SHA is significant: communities with larger distances between the center and the nearest SHA have fewer victims.

FIGURE 5: INSTITUTIONAL FACTORS THAT CONTRIBUTE TO AVOIDING MINE INCIDENTS

No UXO littered Smaller population Nearest other village with recent victims is far away Not in internal Somaliland war 1994-1996 Has more commercial opportunities Nearest danger area is far away Has fewer government-related institutions Relative strength of influence
- Helps avoid incidents altogether
- Helps reduce incidents

The commercial factor is also relevant in determining incident probabilities: communities with permanent markets or with established transport facilities tend to have fewer incidents.

Among the war exposure indicators, only the latest of the three wars influences the number of recent victims significantly. Communities spared from the internal conflict that raged between 1994 and 1996 tend to have fewer victims. A greater distance from the nearest other community that had recent victims decreases the incident probability. In the surveyed regions, this factor has a built-in threshold effect, symbolized by the green bar beside the blue: beyond a certain distance, chances improve for communities to be completely incident-free.

The last two factors also influence the level of mine impact for a community. Smaller communities have a greater chance of avoiding mine incidents, though this depends a great deal on how many people are in contact with mines. Nevertheless, in the surveyed regions, the population size effect is very strong. Finally, communities without UXO are more likely to be incident- free.

For statistical reasons, the influences of road and water blockages on incident probabilities were investigated separately. Road blockages do not increase the probability of incidents. Blocked access to water points, however, significantly increases that probability. Figure 6 represents the probability of a mine incident occurring in the next two years for communities with and without blocked roads, and Figure 7 represents that probability for communities with or without water blockages. In Figure 6, the two groups' probabilities are virtually identical. Figure 7, however, shows that communities with water blockages are nearly 20% more likely to incur an incident in the next two years.

FIGURE 6: MINED ROADS AND INCIDENTS Probability that some incident will occur in the next two years No mined roads Mined roads 142 communities 215 communities The thick horizontal line is the median probability of a new incident in the next two years for an impacted community. The upper and lower bars are extreme instances and additional outliers are shown as round symbols.

FIGURE 7: WATER BLOCKAGES AND MINE INCIDENTS

Probability that some incident will occur in the next two years
No water source Water source
blockage blockage
315 communities 42 communities

The thick horizontal line is the median probability of a new incident in the next two years for an impacted community. The upper and lower bars are extreme instances and additional outliers are shown as round symbols.

COMMUNITY PROFILES

Four community profiles are presented in Table 16. Two of the profiles represent communities with highest probability of having mine incidents and two represent communities with the lowest.

One of the two high-probability communities has had some incidents in the two-year period prior to the survey; the other did not. Neither of the two low-probability communities has had an incident. From lower to higher incident probabilities, the number of blockage types reported increases, although in terms of calculating the model, incident probabilities and blockage types were independent. The types of blockage that each of the community's rated as the most serious differ. There are also very significant differences between high- and low-probability communities in terms of population, SHAs, and institutional features.

TABLE 16: MINE INCIDENTS AND COMMUNITY BACKGROUND VARIABLES-EXAMPLES Communities with two lowest Communities with two highest incident probabilities incident probabilities

Gargaara Galbeed Heemaal Dheere Oodweyne (Shacabka) Burco (Faarah Omar)
Region Awdal Awdal Togdheer Togdheer
Probability of some mine incident in 2 years 0.017 0.022 0.973 0.977
Impact score 2 3 6 14
Number of recent victims in the past 2 years 0 0 0 3
Population 60 600 8,400 36,000
Settlement type Seasonal, Dispersed, Compact, Compact,
100% nomads 100% nomads 20% nomads no nomads
Resource types blocked - 1 3 5
Blockage ranked highest - Road Cropland Water
Year mines last laid                                 1994 1991 1986 1995
Distance to nearest other community with recent victims (km) 33.2 19 0.4 1.7
Water kiosk or hand pumps No No No Yes
Electricity No No Yes Yes
Telephone No No Yes Yes
Medical facility No No Basic Hospital
Primary and intermediate schools No No Both Both
Women's and youth associations No No Both Both

Dhalaaddu community

Dhalaaddu is a dispersed community of 1,100, located in a flat valley surrounded by mountains, 55 kilometers west of Hargeisa. Most permanent residents operate farms irrigated by water pumps, and as many as 500 nomadic families settle there during the rainy season.

Dhalaaddu was the site of intense fighting in 1988. Most inhabitants fled to Ethiopia and only returned at the end of the war in 1991. As a legacy of that war, two roads around the community are mined, but these have been clearly marked and diversions have been established. There have been no recent mine incidents, though several people were killed when inhabitants first moved back from Ethiopia.

The LIS has categorized Dhalaaddu as a lowimpact community, and can be considered a success story of community adaptation to a mine Dhalaaddu teashop threat. Dhalaaddu Hargeisa

Consequences for Mine Action

GENERAL PLANNING CONSIDERATIONS

There are a number of consequences for practical mine action that follow from the LIS data.

Clusters of high-impact communities contain communities with both low- and medium-impact. These communities may have had serious mine incidents, but none within the past two years. Thus, to be fair to all communities, when technical survey teams visit high-impact communities, they should also visit low- and medium-impact communities within the same cluster.

1. Smaller communities must not be discriminated against in the allocation of mine action resources. Although the LIS impact score purposely does not use population size as an indicator of impact, some people still assume that clearing larger communities eliminates more risk and impact than clearing smaller ones. While smaller communities have a greater chance of being incidentfree, victim rates in smaller communities are much higher. In fact, the average victim rate in smaller communities-with a mean population size of 606-is 49.5 victims per 100,000 person-years, in while larger communities-with a mean size of 9,480-the rate is only 6.1.

2. The survey reveals two distinct types of suspected hazard areas: road-related and non-road. Communities can usually establish alternative routes when roads are mined, and thus road-related SHAs tend to have little or no real impact. However, some communities do suffer serious impact from blocked roads, either in terms of economic hardship or risk to community members. In many cases, road incidents involve people not familiar with the contaminated roads. In cases where alternatives exist, fencing and marking contaminated roads should prevent these incidents from occurring without needing to clear the roads immediately. For those roads that do need to be cleared, road rehabilitation is also important. Once cleared, these roads will remain unusable without clearing accumulated brush, mending potholes or completing additional repairs.

3. The effects of blocked access to water are serious. Communities with blocked access to water almost always have more recent victims, and so providing access through blocked water resources should be a major priority. The LIS data indicates that 154 SHAs block access to water in 126 communities. While these numbers are relatively small, they do account for a significant amount of the impact from mines and UXOs in Somaliland.

4. The UXO contamination problem in the surveyed regions has not been fully captured by the LIS, which focuses on suspected hazard areas within a community. When the survey teams encountered individual UXO items, these items were recorded on a spot clearance request, and forwarded to other mine action agencies for clearance. Fortytwo percent of impacted communities were contaminated by significant UXO contamination.

Two hundred and twenty-two spot clearance requests were recorded in 26% of all visited communities. However, because a number of incidents occurred outside suspected hazard areas individual item UXO clearance requests should be targeted for follow-on mine action activities. Table 17 characterizes the number of communities that require spot-clearance requests for individual UXO and the number of communities with full UXO SHAs.

TECHNICAL PLANNING CONSIDERATION

Landmine impact surveys are not designed to investigate technical mine action issues in detail but rather to define the nature of the landmine/UXO problem at the national level. The LIS does provide information to subsequent technical surveys and gives an indication of some of the challenges that will be faced by future mine action planners.

DISTRIBUTION CONSIDERATIONS

The LIS found that the Awdal, Galbeed, Sahil, and Togdheer regions are mine impacted and that all districts within these regions are mine contaminated. Contamination is concentrated in clusters, the largest of which surrounds Hargeisa. Many impacted communities are located on the border with Ethiopia. Contamination in the surveyed regions exists throughout the area due to the nature of the conflicts.

CLASS OF MUNITIONS

Ordnance type is an important factor in determining clearance methods and equipment requirements. For instance, mechanical anti-personnel mine clearance devices cannot safely be used for clearing anti-tank mines.

Of the 772 SHAs, only 22 are reported to contain only anti-personnel mines. Two hundred twenty-five SHAs have a combination of anti-personnel mines, antitank mines and UXO. Of the 421 SHAs reported to contain only anti-tank mines, 388 or 92% are mined roads. This means heavier, more expensive equipment for clearance will be needed in those cases where fencing mined roads and improving alternate routes is not the preferred solution.

TABLE 17: COMMUNITIES REQUESTING SPOT CLEARANCE OF UXO

Number of Number of communities communities requesting surveyed spot clearance Percent

Impacted 357 138 39%
Non-impacted 214 15 7%
Total communities 571 153 27%

VEGETATION AND SURFACE

More than 90% of SHAs in the surveyed regions are covered by bushes and scattered trees. Sixty-nine percent of all SHAs are on flat land. This vegetation/terrain combination allows for clearance using mine detection dogs and manual mine clearance teams. The soil conditions in the contaminated areas of these surveyed regions are sandy or rocky, but even sandy soil can be packed into a hard crust, making prodding or excavation operations more difficult.

Contamination is largely rural, impacting village communities. Investments in urban mine clearance, both in terms of equipment and training, are therefore largely not necessary.

MINE RISK EDUCATION

The survey results can assist in planning and establishing targeted mine risk education programs. LIS data includes the age and gender of each recent victim and the activity they were involved in at the time of the incident. The results show that an overwhelming proportion of incidents involve children between 4 and 15 years of age, and incidents most often occur when these children are herding animals. This type of information should provide a basis for determining the appropriate direction and audience for mine risk education approaches.

Significantly, one-third of all incidents happen outside survey-defined SHAs, suggesting a prevalence of single-item UXOs or mines in communities. Many incidents involve individuals, particularly children, picking up or tampering with UXO items.

Age and gender of the victims: Of the 276 recent victims, 63 (23%) were female and 213 (87%) were male. Children under 15 years of age account for 55% of all victims, while female victims account for 26% and male victims for 74%.

Location of victims: Reports of mine incidents existed in all the surveyed regions. The Galbeed region has the highest number of recent victims, but the risk probabilities are highest in Awdal. Incidents are least likely to occur in Togdheer (Table 18). Activities of landmine victims: For both males and females, most incidents occurred while herding animals. Sixty-four percent of all the recent mine incidents involve herding, while traveling, the next highest category, accounts for only 10%. Other activities that involve mine incidents include playing (9%), tampering (7%), household work (2%), collecting food/water (1%), and military activities (1%).

The survey database contains a wealth of information on victims and a more detailed analysis of victim data at the local level will lead to better mine risk education planning. The survey results can be used to create preliminary profiles of recent victims and identify the location of incidents for further investigation.

TABLE 18: RECENT VICTIMS, BY REGION

Victims in impacted Victims in total
Recent Population communities per Population population per
Province victims impacted 100,000 person years total 
100,000 person years

Awdal    72  200,327   18.0 400,212   9.0
Galbeed  119 498,901   11.9 888,995   6.7
Sahil    32  168,738   9.5  205,198   7.8
Togdheer 53  472,640   5.6  563,512   4.7
TOTAL    276 1,340,606 10.3 2,057,917 6.7

Types of Injury:
Of the 276 persons involved in mine incidents in the last two years, 92 lost their lives. Of the remaining survivors, 47 had a limb amputated, 18 were partially or fully blinded and 126 reported other injuries (Table 20).

TABLE 19: CARE RECEIVED BY MINE VICTIMS

Region Emergency Rehabilitation Other None Fatal Total
Awdal    28  0 3 13 28 72
Galbeed  66  3 1 6  39 115
Sahil    18  1 4 0  9  32
Togdheer 29  0 0 3 16  48
TOTAL    141 4 8 22 92 267

TABLE 20: TYPE OF INJURY SUFFERED BY MINE VICTIMS, BY REGION

Region Fatal Amputation Loss of sight Other Total

Awdal    28 7  3  35  73
Galbeed  39 20 12 54  125
Sahil    9  5  1  16  31
Togdheer 16 15 2  21  54
TOTAL    92 47 18 126 283

The table shows a total of 283 injuries; 7 victims had multiple wounds.

Sayla Bari community

The dispersed rural community of Sayla Bari has a permanent population of 1,400 among 200 farms, and nomads settle there in the rainy season. Most individuals engage in subsistence rain-fed agriculture and animal husbandry. Water is collected in rainfed cisterns called berkeds.

Somali government forces established two camps near Sayla Bari in 1977, which were abandoned when the SNM occupied the community and established their camp in 1988. During this period, there was heavy fighting and most community residents fled to Ethiopia. Community members began returning at the end of the war in 1991.

The former government camps to the north and northwest of the community are mined and block access to pasture and cropland. The owner of land near the northern camp has indicated that he would cultivate crops there if the mines were cleared. All three camps have significant UXO stockpiles. Sayla Bari was designated as a medium-impact community by the survey. In addition to blocked access to scarce cropland, the mines and UXO represent a continued threat to herders who wander into these areas.
Female informants in Sayla Bari
Sayla Bari
Hargeisa

Background & Methodology

Project Timeline

The initial planning for the Somalia Landmine Impact survey began in the spring of 2001. It was decided to divide the project into several phases as only the Awdal, Galbeed, Sahil, and Togdheer regions (Phase 1) had sufficient political stability to facilitate a survey. Phase 1 of the LIS was completed in the spring of 2003.

- January 2002-SAC seminar-Bo Bischoff (Director of DDG), Paul Mackintosh (team leader) and Helene Ruud (socioeconomic expert) participated in a 4-day LIS training seminar conducted by SAC in Oslo, Norway.

Detailed project planning-The team leader and DDG management worked on detailed project planning and budgeting. The results were finalized with SAC at a meeting in Copenhagen in late January.
- February 2002-Team leader returned to Hargeisa-The team leader was the program manager for Danish Demining Group's demining program, until the LIS commenced.
- March 2002-Start delayed-The arrangements between the United Nations (UN) and SAC were not yet finalized, which led to a delay of the start date. Coordination-The team leader refined project plans and initiated contacts with the government. It was established that the survey would operate under the Ministry of Resettlement, Rehabilitation and Reconstruction (MRR&R). Coordination with UNDP Mine Action began.
- April 2002-Local negotiations-Final contracts were signed and the project start date was set to 1 May. Throughout April, the team leader continued negotiations with the national authorities regarding the memorandum of understanding (MOU) and other practical matters, including the principle of employment based on a free and open selection.

Startup information meeting-At the end of April, a meeting with national stakeholders was held to inform them of the upcoming LIS. MRR&R pledged full support for the LIS.

Logistic preparations-The team leader found a suitable compound in Hargeisa for the survey and supplied it with equipment provided by UNDP and established an office.
- May 2002-Official project start-Phase 1 of the Landmine Impact Survey officially started on 1 May. Survey staff (Socioeconomic expert, Administration/ Finance officer, and geographical information system (GIS) and mapping officer) arrived.

MOU-On 12 May, the MOU with MRR&R was signed. The MOU was a onepage document stating that the survey would be conducted under the general DDG MOU for mine clearance activities with minor modifications.

Office set up-By 19 May, the move to the new office compound was completed. The LIS compound was fitted with a very reliable generator, a computer network and a database center.

IMSMA center staffed-The database center was staffed with four data entry operators seconded from the Somaliland Mine Action Center (SMAC), the National Demining Agency (NDA) and UNDP. Under the guidance of the GIS and mapping officer, they started training on IMSMA.

IMSMA support from the Geneva International Center for Humanitarian Demining (GICHD)-From 23 May to 30 May, a GICHD consultant visited the IMSMA/GIS Mapping Office to help with training, setup of a network environment and to discuss customizations in the IMSMA database.

- June 2002-Adaptation of survey questionnaire-The questionnaire was adapted for local conditions in cooperation with the national social scientist. Several questions were added or changed, database consistency was checked and discrepancies were resolved. The national social scientist translated the questionnaire into the Somali language.

Recruitment of senior staff-After advertising in the national newspapers for field supervisors and field editors, the survey received more than 170 applications. Approximately 50 applicants were given a written test of basic skills and English proficiency. Of the 30 who were invited for interviews, three supervisors and six field editors were hired for the training.

Training-On 22 June, SAC training staff arrived and began the training on 25 June.

Expert Opinion Collection (EOC) international organizations-Information on contaminated areas were obtained from the three demining organizations operating in the surveyed regions: DDG, HALO Trust and St. Barbara Foundation. A questionnaire, with an attached map, was sent to all international NGOs requesting their knowledge of where contaminated areas are located.

Obtaining maps and gazetteer data-The GIS and mapping officer visited the UN Data and Information Management Unit (DIMU) in Nairobi to get all available digital maps and gazetteer information. Community lists from various ministries were also used to build the survey gazetteer.

Support to the IMSMA/GIS center-In June, Franco Serena, funded by the Swiss Government through UNDP, joined the IMSMA/GIS department for a three-month consultancy,
- July 2002-Pre-test 1-At the beginning of July, the senior staff trainees, with observers from SAC and DDG, conducted pre-tests in nine suspected communities. Each trainee participated in three community interviews. It was decided that after the changes to the questionnaire and some additional training was conducted, a second pre-test would be needed. This added one month to the project duration. On 15 July, the SAC training team departed.

EOC district level-Approximately half of the trainees formed three teams to collect expert information at the district level. Preparation of interviewer training-The remaining trainees assisted in the translation of the training material from English to Somali in preparation for the interviewer training.

Refinement of survey questionnaire-The questionnaire was updated and procedures to counter false reporting were developed.
- August 2002-Retraining and pre-test 2-After completion of the EOC, retraining commenced with the new procedures and an updated questionnaire. This culminated in the second pre-test, conducted in six suspected communities. SOP development-Standard operating procedures (SOPs) for field operations were developed during August and September.

Supervisor and editor operations training- Supervisors and field editors were trained in operations and introduced to the SOPs.

Recruitment of interviewers-The practice of hiring equal numbers of men and women was actively pursued. The survey hired 11 women and nine men as interviewers from all parts of Somaliland.
- September 2002-Interviewer training-Supervisors and field editors conducted two weeks of interviewer training.

Pilot test-A one-week pilot test was executed immediately following interviewer training. The test was conducted in Gabiley, about 80 kilometers west of Hargeisa, and involved community visits, continued training, and performance evaluation. Interviewers then participated in several days of evaluation and retraining in Hargeisa.

UN QA monitor-The UN quality assurance monitor joined the survey on the first day of the pilot test.
- October 2002-Public information campaign-A public information campaign was launched to coincide with the start of data collection. A letter of endorsement was obtained from a local official and published in major newspapers along with an article about the survey. Campaign information was also broadcast over the radio and the team leader conducted a television interview at the point of final deployment.

Deployment-On 5 October, the two survey groups were deployed to their operational areas of Togdheer and Sahil. Each survey group was made up of

Evaluation during the pilot test (top); Pilot test base camp four interviewer pairs, with the remaining two pairs based in Hargeisa. Field work began on 6 October.

Operations-The administration/finance officer took on the role of monitoring operations, and tracking progress based on new operational guidelines. Daily reports from the field were received and progress monitored closely in Hargeisa.

New QC unit established-A Quality Control (QC) unit was established to reduce the number of mistakes in the initial surveys. This unit checked all surveys for completeness and consistency, and returned surveys to the field for correction. The QC unit also visited the survey groups to retrain the interviewers and field editors.
- November 2002-Data entry accelerates-Due to problems with the quality of completed questionnaires, data entry did not begin until November. The team of four data entry operators entered survey information without serious backlogs. Prolonged data collection-The survey groups found more villages than were present in the gazetteer and many of the non-suspected villages turned out to be mine impacted. Because of this and the additional quality control procedures, the survey fell behind schedule. SAC and the LIS team agreed to extend the data collection period by one and a half months.
- December 2002-Instability in Sool region-Because of instability in the Sool region, the eastern survey group relocated further west. After the situation settled, the survey group returned to the east and started surveying communities in Sanaag.
- January 2003-Fighting in Sanaag region-On 1 January, the eastern survey group encountered militia forces from Puntland and was forced out by renewed fighting in the east. The eastern survey group was then redeployed to the west. In consultation with UNDP and SAC, the LIS decided to omit the Sool and Sanaag regions because of security concerns.

End of data collection-By the end of January, the Phase 1 regions had been surveyed.

Data entry quality control-In January, a second QC unit was established to review all data entry to ensure quality.
- February 2003-Cleanup operations-Some field staff members participated in surveys for missed communities. The QC units finished the last surveys and the data entry operators completed the database, including verification and corrections.

Training exercise

Equipment stored-Field equipment was stored with UNDP to wait for the next phase of the Somalia LIS.

SAC visit-The SAC Director for Survey visited the surveyed regions to review the survey operations with the LIS staff.

Analysis and report writing-A SAC team was sent to the surveyed regions to assist with data analysis, mapping and report writing.
- March 2003-Close down-The compound was emptied and handed back to the owner and the remaining staff members were released from service.

Key Participants

The following organizations were essential for the execution of Phase 1 of the Landmine Impact Survey.

GOVERNMENTAL AUTHORITIES

The Ministry of Resettlement, Rehabilitation and Reconstruction (MRR&R), under the jurisdiction of the Ministry of Planning and Coordination, is the national authority responsible for mine action in the surveyed regions. DDG signed an MOU with the MRR&R on the survey, and coordinated closely with the Ministry's liaison officer. The survey was also provided with valuable support from the Vice President of Somaliland.
- The Somaliland Mine Action Center (SMAC), one of two national mine action agencies, is responsible for maintaining the IMSMA, supported by UNDP. The LIS cooperated with SMAC and the National Demining Authority (NDA) in developing its IMSMA unit. SMAC will be responsible for maintaining the LIS data in the IMSMA database.

INTERNATIONAL ORGANIZATIONS

- Danish Demining Group is a non-profit mine clearance organization based in Denmark. DDG, under a sub-contract from SAC, served as the main implementing organization for Phase 1 of the LIS.
- The Survey Action Center is the executing agency of the Survey Working Group and an international NGO responsible for most Landmine Impact Surveys.
- The Geneva International Center for Humanitarian Demining is an international center dedicated to research and development of humanitarian mine action. GICHD sent a consultant to assist the LIS team with establishing an IMSMA database unit.
- United Nations Mine Action Service was formed in October 1997 to serve as the UN focal point for mine action. At the global level, it is responsible for coordinating all aspects of mine action within the UN system. At the field level, it is responsible for providing mine action assistance in the context of humanitarian emergencies and peacekeeping operations. The United Nations Mine Action Service (UNMAS) provided support to Phase 1 of the LIS through the provision of a Quality Assurance Monitor.
- The United Nations Office for Project Services has been designated as a principal provider of mine action services within the United Nations system. In March 1998, UNOPS established a dedicated Mine Action Unit (MAU) with specific project management, technical and legal expertise for mine action. UNOPS provided support to the LIS through providing a Quality Assurance Monitor.
- United Nations Development Program (UNDP) is the main coordinating agency for UN country operations, and is involved with supporting mine action programs. UNDP provided the LIS with most of the computer and global positioning system (GPS) equipment needed for the survey. The European Community (EC) provided funding through UNDP.
- The European Community, the government of Sweden through the Swedish International Development Cooperation Agency (SIDA), the government of Finland, and the government of Canada through the Canadian International Development Agency (CIDA) provided the funding for Phase 1 of the Landmine Impact Survey.

Administrative Structures

PROJECT ORGANIZATION

- Headquarters: Phase 1 of the Landmine Impact Survey operated out of a compound in Hargeisa containing offices, a classroom, storage buildings and parking spaces. A generator was installed. The survey fitted the offices with a network that connected all computers to a central file server and a number of shared printers.

- Staff and organization: The survey had four full time international staff members: a team leader, a socioeconomic expert, a GIS and mapping officer and an administration/finance officer, who also assumed operations monitoring and execution roles during the data collection period. The national staff numbered more than 60 during the data collection period.

- Two field survey groups (SGs): Survey groups consisted of a supervisor, field editors, interviewers and support staff. They worked from mobile field camps as independent operating units to cover the eastern and western parts of the surveyed regions. They moved around in four-wheel-drive vehicles and communicated with satellite phones. Each group had tents, camping equipment, a generator and a field kitchen. A paramedic was assigned to each SG. Two interviewer teams operated from headquarters (HQ) and handled the communities near Hargeisa.

- SAC technical assistance: SAC deployed three staff members to the survey for 20 days to conduct the senior staff training. The training material (lessons, exercises, etc.) was then translated and used for the interviewer training. During this period, SAC also rendered assistance in adapting the questionnaire to local conditions and sharing experience from earlier surveys. SAC advised on a number of critical decisions for project implementation. At the end of the project, two SAC staff members went to Hargeisa to assist in the final data analysis and report writing. SAC was responsible for the final drafting and production of the final report.

- Reporting: Weekly and monthly progress reports covering project activities and outstanding issues were submitted to SAC A monthly financial report was also submitted. SAC was responsible for reporting to donors.

- UN Quality Assurance: UNMAS/UNOPS provided the survey with a quality assurance monitor (QAM). The QAM continually monitored and documented the progress of the project based on certification guidelines. The QAM also made recommendations to the survey staff based on his observations and analysis and thereby assisting the survey and ensuring compliance with UN certification guidelines.

Strategic Planning

The establishment of a strategic planning framework that makes use of the LIS data to assist authorities in establishing medium- and long-term priorities for mine action is a central part of the LIS mission. The strategic planning process usually takes place concurrently with survey operations. For this survey, however, strategic planning was postponed because of the absence of a permanent operational mine action center and preoccupation with the April 2003 presidential elections. The stakeholders-including the United Nations, the European Community and the Survey Action Center-agreed to begin the strategic planning process as soon as the requisite conditions are met after survey completion. In May 2003, the Mine Action Unit of Cranfield University (CMA) conducted an operational review in Hargeisa in preparation for strategic planning exercises later in the year.

Finances

BUDGET AND EXPENDITURES

The original budget for the survey was developed following the SAC advance survey mission in March 2000. In January 2002, the budget was increased from US$874,000 to US$966,800 to reflect local wages and price adjustments and an extension of the data collection period. In August 2003, the budget was increased to US$976,900 to account for publication and printing expenses associated with the final report. Out of this budget, US$43,600 was spent on non-expendable equipment, which was transferred at the end of the project to UNDP for a future Landmine Impact Survey in Puntland. All survey vehicles were rented locally, which was a considerable cost saving in comparison to purchasing vehicles. UNDP and SMAC provided DDG office equipment during the survey. Funds expended by the United Nations to cover the costs of their support, equipment, the quality assurance monitoring and certification are not reflected in these figures.

Survey budget

DDG (survey)            $702,400
SAC (technical support) $262,000
CMA (strategic planning)$12,500
Total                   $976,900

Funding for Phase 1 of the LIS was provided directly to SAC by the governments of Canada and Sweden. Funding from the European Community and Finland has been contracted through the UNDP office for Somalia and UNOPS in New York.

European Community (EuropeAid) $504,970
Sweden (SIDA)                  $285,970
Finland                        $88,496
Canada (CIDA)                  $45,500
UNDP/SMAC in-kind(office equipment)
Total                          $924,936

SURVEY METHODOLOGY-OVERVIEW

The poor quality of the different sources for the gazetteer at the start of operations, as well as the lack of mine-related information at district and regional headquarters, forced some methodological adaptations. In Phase 1 of the LIS, the survey conducted a full census of all village communities for mine impact. Those communities that were found impacted were then surveyed: a community interview and associated activities were conducted.

Communities that stated they were not impacted were not surveyed, as well as those with only slight UXO contamination. If UXO were found as individual items and not in SHAs, interviewers filled out a spot-clearance request but the survey would not include the community (see below). As in other impact surveys, the data was entered into IMSMA, which formed the basis of subsequent data analysis.

Field staff was recruited from each of the four surveyed regions to ensure that the survey had sufficient geographic and cultural knowledge. The society is a traditional Muslim one, and thus is largely gender-segregated. Having women and men working closely together on the same survey teams and in the DDG LIS office caused no problems, however. As much of the population is nomadic or partially nomadic, nomads were included in community interviews.

TRAINING

SAC, in coordination with the DDG LIS staff, trained the field editors and supervisors. During the three-week course, the first of the two pre-tests was carried out. Later, the LIS team conducted additional training. Subsequently, the trainee group was split into two groups. One group with the supervisors and some of the field editors collected expert opinion at regional and district levels and the other group with the three field editors stayed in Hargeisa to prepare the interviewer training and to translate training materials.

The interviewer training course included 11 women and nine men and was two weeks in duration. Supervisors, field editors and interviewers went directly to the field after interviewer training in order to conduct the pilot test of the survey. The pilot test covered 20 mine-impacted communities, during and after which more training was conducted and logistics, communications, and field organization were reviewed.

The training of supervisors, field editors and interviewers addressed the setting, organization, detailed components of the survey, the questionnaire, and the scoring system. The field staff and the four database operators were trained in community mapping, key informant selection, interview technique, mine and UXO recognition, first aid, safety procedures and visual inspection of SHAs. Exercises in the use of global positioning systems (GPS), compasses, cameras, satellite phones and maps supplemented formal instruction.

QUESTIONNAIRE AND TESTING

Phase 1 of the LIS adapted the questionnaire template provided by SAC to local conditions with the help of the national social scientist. The survey underwent two pre-tests: one during the training of senior survey staff in which SAC staff participated and another after the re-training of the senior staff. The questionnaire was modified after each of the pre-tests and adjusted after the pilot test. The trainees made valuable contributions to the questionnaire evaluation and to translation of the document.

EXPERT OPINION COLLECTION

Digital maps and community names and locations in the surveyed regions were only partially available. Digital maps were obtained from the UNDP Data and Information Management Unit (DIMU) in Nairobi. A complete gazetteer did not exist for Somaliland, and so the survey received partial data from DIMU and from national authorities to create an operational version, compiled from various sources. Demining agencies, as well as international NGOs and UN agencies, were approached for information on contaminated areas and suspected communities. Data from the Somaliland Mine Action Center (SMAC) was incorporated into the LIS database.

The supervisors and field editors also gathered expert opinion and district levels and 170 individuals were interviewed on the gazetteer and the location of suspected communities. In these meetings, field staff also requested information to update the gazetteer.

Subsequent quality control efforts revealed that gazetteer sources and expert opinion were both so defective that a full-census approach became necessary, with survey staff visiting every permanent settlement. Eight hundred forty-one communities were thought to exist in the surveyed regions. Eastern districts became inaccessible for survey teams, for security reasons, during the survey. (A few eastern communities were surveyed before their regions had to be abandoned. These, however, are not referred to elsewhere in this report.) In the districts covered by the survey, teams visited all 588 communities except for 17, which were completely abandoned or inaccessible and are assumed to be not impacted. Some of the 588 communities had not been included in the initial gazetteer and others in the original gazetteer could not be found.

The number of false negatives is relatively high; this too justified the census approach to determine the initial community contamination status. Because of the full census taken, the question of survey coverage hardly arises. In the most extreme scenario, all of the 17 communities not visited could be imagined as being inhabited and impacted. In this very unlikely event, the survey would have covered 357/(357 + 17) = 95% of all impacted communities in the surveyed regions.

FIGURE 8: RESULTS OF EXPERT OPINION COLLECTION

Not suspected and visited: 341
True positives:            180
Not visited, abandoned:    2
Suspected and visited:     247
False positives:           65
False negatives:           177
True negatives:            149
Not visited, abandoned:    15
Total impacted communities:357
Surveyed area communities: 588
All communities in Somaliland: 841

COMMUNITY INTERVIEWS

LIS staff made initial visits to communities to determine whether a community was mine-impacted. If three community members certified that a community was not impacted, then no community visit and interview was conducted but instead a nonimpacted sheet was filled out. When a community was identified as impacted, appointments were made for interviewers to conduct a community interview with key informants from that community. For communities that had only UXO contamination, field staff would decide whether the level of contamination required a full community visit. When UXO formed large stockpiles or was widely littered, the community was surveyed. If UXO were present only as individual items, field staff would fill out a spot clearance request and not schedule a full visit. Spot clearance requests were also filled out in more widely impacted communities.

The interviewer team visited one community each day, and spent most of their time in interviews with key informants. Group interviews included community mapping and questions concerning socioeconomic blockage and recent victims of mine/UXO incidents. Interviewers, in addition to conducting the formal question-

FIGURE 9: CHARACTERISTICS OF IMPACTED COMMUNITIES (compared to non-impacted communities)

Larger community (current population, log10)
More active civil society (civic factor scores)
Distance from community center to nearest mined area (meters, log10)
Distance from Hargeisa (meters, log10)
Distance from Hargeisa, dampened (meters, log10 squared)
Total estimated mined area surface (square meters, log10)

Has UXO, in addition to landmines less likely more likely
- Not statistically significant
- Statistically significant

Tea shop naire, collected background information, visually inspected SHAs and resolved inconsistencies. Visual inspection of SHAs from a safe viewing point is a means of verifying information gathered in the community interview and obtain geographic coordinates and a photo that may be of some use to subsequent technical surveys.

THE DEMOGRAPHICS OF COMMUNITY INTERVIEWS

In the 357 communities surveyed, a total of 4,060 key informants participated in interviews. The majority of the informants were men (76%) and between 30 and 44 years old. Nine hundred and thirty-six informants were women (23%). In 97 out of 357 surveyed communities (27%), no women participated at all. In four communities all the informants were women. The average number of female informants in a community was 2.7. Most participants were settled farmers (28%), followed by nomads (27%).

FIELD EDITING, QUALITY CONTROL, DATA ENTRY AND ANALYSIS

Interviewers were required to complete the questionnaire and attachments before submission to the field editor. Field editors checked the material for consistency, completeness and legibility, and translated the questionnaires.

Supervisors held weekly meetings with their survey staff to discuss operational plans, problems, results and impact scoring. Completed questionnaires were sent to Hargeisa for data entry.

QUALITY CONTROL PROCESS

The first completed surveys were found to be of poor quality, and survey management established a quality control unit in Hargeisa and retrained field staff on the quality aspects of the questionnaire. When the QC unit found omissions or mistakes in submitted questionnaires, these were returned to the field for correction. These procedures led to substantial improvement in the quality of the data.

Community mapping (top); Women in a community interview

LOCAL ADJUSTMENT FACTOR

The protocols for the LIS allow for survey field staff to adjust community scores if those scores did not adequately reflect the community's impact. This procedure was available to interviewers but was not used in Phase 1 of this LIS. Some survey- wide modifications to LIS methodology-excluding contaminated roads for which easy detours have been established and individual item UXOs-were adopted.

RESURVEY OF ONE PERCENT OF SURVEYED COMMUNITIES

According to the guidelines for UN certification, one percent of the surveyed communities must be resurveyed. Five randomly selected communities were resurveyed: two from the Galbeed region and one from each of the other regions. A survey team other than the one that conducted the first survey visited the communities. The differences between the original and the resurveyed data were minimal and, most significantly, no impact score was changed.

DATABASE MANAGEMENT

The IMSMA database was used to receive, store and manage the information gathered in the field. The default database design was adjusted during the pilottest to accommodate modifications in the questionnaire.

A quality control function was established to reduce errors introduced data entry.

PARTICIPATION OF WOMEN

No female candidates for the supervisor and editor positions had the required English language proficiency. The survey, however, successfully recruited 11 women for interviewer positions, as English were not required for conducting the community interviews.

These interviewers were essential for gathering information from female informants who would not communicate with male interviewers. Men would often be more available for interview than women, who were preoccupied with household work: childcare, gathering water and firewood, taking care of animals and being engaged in cottage industry. Group interviews were most effective when held at a time and a place that was convenient and appropriate for female informants.

Meeting a community leader

NOMADS

The majority of the population in the surveyed regions is nomadic or seminomadic. As land used by nomads nominally belongs to a certain community, a community interview would include nomads staying on this land. The survey made special efforts to include nomads as key informants in the community interviews. Nomads interact frequently with permanent community members, and this sustained contact allows for the exchange of information about the mine/UXO problem.

The survey documented the participation of the nomadic population. One thousand seventy-seven out of 4,060 informants (27%) defined themselves as nomads. Nomads were present in the group interviews in 217 communities (61%). The nomadic population in the surveyed areas was estimated at 735,000 or 54% of the total population.

Community Scoring,Weighting, and Classification

THE MINE IMPACT SCORE

The central element of the LIS is to score and classify mine-impacted communities according to the severity of impacts. The purpose of the impact score is to allow mine action agencies to prioritize communities for a variety of activities ranging from victim assistance to mine risk education to clearance.

Landmine impact scores are calculated only for communities, rather than for suspected hazard areas or recent victims. The number and size of suspected hazard areas or by the size of the community do not affect the score. It is a composite of three aspects of the mine problem: 1) the nature of munitions present, 2) the types of institutional/socioeconomic blockages present, and 3) the number of recent victims.

The nature of munitions is scored as two points for mines, one point for UXO or three points for both. Two points are given for each victim in the last 24 months, and specific surveys define the scoring for socioeconomic blockages, within a framework set by the Survey Working Group (SWG). (See Map 10.)

Sahil Awdal Togdheer Galbeed MAP 10: CONTRIBUTIONS OF MUNITIONS, BLOCKAGES AND VICTIMS TO IMPACT SCORES
Community impact
- Victims
- Munitions
- Blockages

Non-impacted (231 communities)
- Non-surveyed

The radius of the pie chart represents the actual impact score, ranging from 2 to 29 points. The pie segments indicate the contribution of each type of impact as indicated.

ELEMENTS OF THE SCORING MECHANISM

In the IMSMA default configuration, the following indicators are included:
- The presence of mines
- The presence of unexploded ordnance
- Access to crop land
- Access to community's pasture
- Access to water points
- Access to non-cultivated area
- Access to housing area
- Roads that are blocked
- Access to other infrastructure
- Mine/UXO victims in the last 24 months

Irrigated land is generally distinguished from rain-fed land, and blockage of drinking water is usually considered separately from water used for other purposes. Fifteen different variables enter the default configuration. However, the database also allows surveys to add five user-defined variables to account for countryspecific conditions. Scoring for presence of mines and UXO and for recent victims is set across all surveys and cannot be modified. The SWG has mandated standardization of the scoring to allow for broad comparability among surveys.

MEASUREMENT

The nature of munitions and specific socioeconomic indicators are calculated based on the presence or the absence of the value in question. Only the recent victim element is related to quantity. This largely non-quantified approach was chosen for a number of reasons involving consensus, validity and reliability.

The recent victim part of the score is measured as a count. Each person who was killed or injured in the previous 24 months adds two points to the score. Data is collected for victims further in the past, but does not affect the score.

WEIGHTS

The scoring weights for munition types and recent victims are fixed. For the socioeconomic indicators, however, surveys choose, within limits, how much impact each indicator contributes to the score. The weights system for Phase 1 of the LIS impact score calculation is shown in Table 21.

TABLE 21: WEIGHTS USED IN COMPUTATION OF IMPACT SCORES

Variable Weight

There were mines:                                2
There was UXO:                                   1
Access to some pasture was blocked:              1
Access to some rain-fed crop was blocked:        1
Access to some irrigated crop land was blocked:  2
Access to some water points was blocked:         2
Access to some non-cultivated area was blocked:  1
Access to some housing area was blocked:         1
Some roads were blocked:                         1
Access to some other infrastructure was blocked: 1
Mine victims in the last 24 months: 2 x number of recent victims

The survey works with four impact categories, based on the score range:
0: No known impact problem
1-5: Low impact
6-10: Medium impact
11+: High impact

THE WEIGHTS BUDGET

The Danish Demining Group and its socioeconomic expert, in conformance to Survey Working Group protocols and in close coordination with the national authorities, chose the default weights for socioeconomic blockage used in this survey. The sum of weights for socioeconomic indicators must be equal to 10 and weights must be 0 (indicator excluded), 1, 2, or 3. This prevents impact inflation and ensures some inter-survey comparability.

SPECIFIC CHANGES MADE IN THE SCORING FRAMEWORK

Two main modifications were made in this survey. First, blocked roads were not counted as blockages in the score if alternatives had been found and thus the blockage had no serious impact. Second, individual items of UXO were not counted. UXO was only considered in the score if found in large stockpiles or scattered widely as battlefield contamination.

IMPACTS RANKED BY KEY INFORMANTS

The survey has been successful in getting community members to rank their own impacts. This ranking did not impact the score, but provides useful information about the community's own concerns and impact priorities.

In 159 out of 357 community interviews, communities ranked their own impacts. These weights, as described in Table 22 (next page), are separated from the frequencies with which the blockage types were reported. These alternate weights are computed by scaling the probability that a given blockage type is ranked highest, to a maximum of ten.

The weights implied by the key informants' rankings are varied, from a low of 0.49 for buildings and facilities to a high 3.07 for water.

Several of the implied weights make sense for this geographical region, with the largest weight given to water and cropland. The relatively high weight placed on road blockage was unexpected, since most communities found easy alternatives for the blocked road.

SENSITIVITY OF THE COMMUNITY CLASSIFICATION TO WEIGHT CHANGES

As an experiment, impact scores were recalculated based on the impact weights from the community ranking above (see Table 23). Throughout this experiment, the weights budget of 10 was maintained. The number of high-impact communities changed little when different weights were applied, which indicates that the scoring system is not very sensitive to blockage weight changes. In fact, the overall distribution over the three impact categories does not change much with variations in indicator weights. This indicates that the LIS scoring system is strong enough to express the real impact of landmines on communities no matter how blockage weights are adjusted.

TABLE 22: FIRST RANKS AND BLOCKAGE OCCURRENCES

Probability that this Type of blockage Occurrences 
Times ranked first is ranked 
first Implied weights
Roads (any type) 136 57 0.42 2.14
Pasture          121 31 0.26 1.31
Cropland (any type) 78 27 0.35 1.77
Water (all uses) 60 36 0.6 3.07
Non-agricultural land 55 4 0.07 0.37
Buildings and facilities 21 2 0.1 0.49
Houses           12 2 0.17 0.85
Total            483 159
Sum of conditional probabilities 1.96
Weight sum 10
All surveyed communities = 357

As implied in
As used the impact rankings by in this report the key informants
Mines 2 2
UXO 1 1
Pasture 1 1.31
Rain-fed cropland 1 -
Irrigated cropland 2 -
Cropland, any type - 1.71
Non-agricultural land 1 0.37
Residential areas 1 0.85
Other buildings and facilities 1 0.49
Roads 1 2.14
Water 2 3.07
Recent victims 2 2
High-impact communities 45 53
Medium-impact communities 102 113
Low-impact communities 210 191
All communities 357 357

This report mainly describes the rationale behind the operations of Phase 1 of the Landmine Impact Survey. A number of experiences, issues and lessons learned during the survey should be useful to other organizations implementing the LIS in the future.

The survey lasted from May 2002 to March 2003. The project started out with an eight-month timeline: with four months allocated for operational setup and training, two and half months for data collection, and six weeks for completion. In actual fact, it took five months for set-up, four months for data collection and two months for completion. Even this expanded timeline was only possible because of the survey staff's familiarity with local working conditions.

The survey faced two main operational problems:
- The EOC-generated list of suspected communities and gazetteer compiled by the survey at the start of operations did not reflect the situation on the ground.
- It was difficult to maintain consistent quality interviews and data entry. By successfully addressing these issues, DDG was able to conduct a survey that meets the requirements for UN certification and gives Somaliland a tool for mine action prioritization.

The LIS was able to accomplish this through the dedication and hard work of the survey international and national staff and the UN quality assurance monitor. I wish to thank them sincerely for their commitment to the LIS process and product.

ADMINISTRATION AND LOGISTICS

International staff

There were four international staff members employed by Phase 1 of the LIS: the team leader, a socioeconomic expert, a GIS and mapping officer and an administration officer. The team leader and the socioeconomic expert attended a four-day seminar in Oslo in January 2002 on the methodology and execution of the LIS. The GIS and mapping officer participated in a course on IMSMA at GICHD. The administration officer was given an introduction to the administrative procedures of the DDG before joining the team. These workshops and briefings prepared the staff for the start of the LIS project on 1 May 2002.

DDG had been involved in mine action in the surveyed regions since 1999 and thus knew the country and the operating environment well. DDG has a very good relationship with the Hargeisa government and is well respected by the population. By the time the survey got underway, the team leader had already spent nearly a year in the region, half of this as the programme manager for DDG mine clearance. As the team leader was present months before the start of the survey, he executed many of the set-up tasks before 1 May, when the survey officially started.

National staff

Education in the surveyed regions has been in disarray for decades and many educated people have fled the country. Unemployment is very high, and only a few job opportunities in the country can provide people the skills and experience necessary for proper LIS execution. Therefore, it was extremely difficult to recruit qualified personnel.

Furthermore, the lack of education and English proficiency prevented women from being considered for senior staff positions. Female interviewers, however, were recruited for the survey teams and they ensured that women's opinions and experience were captured. While there was much initial skepticism, the recruitment of female staff greatly enhanced the quality of the survey.

Society in this area is clan-based, and the clans and sub-clans compete for resources. As a result, DDG had to be very careful with the distribution of survey employment and purchasing benefits in the country. Therefore, recruitment and vehicle hire had to be geographically balanced among the surveyed regions. While this made hiring complex at times, the survey was accepted everywhere and had a pool of local knowledge from which to draw.

The first recruitment efforts were for the administration staff at HQ. The survey quickly hired six guards to protect the compound, logistic and financial administration assistants for the office, and a cleaner. Hiring a national social scientist represented more of a challenge, but the survey eventually found Ibrahim Musa Wais, who with his extensive knowledge of Somali society soon became a keystone component of the survey.

The survey then recruited three field supervisors and six field editors from more than 150 applications. These senior staff members were required to be proficient in English, which, as a result, had to exclude female candidates. The survey then hired 20 interviewers from Hargeisa and the different regions, 11 of which were women.

Four IMSMA data entry operators were seconded from the Somaliland Mine Action Center (SMAC), the National Demining Authority and UNDP. Three of these employees had already been trained in mine action-related database operations. The LIS recruitment process was praised by the national government for being open and transparent while also achieving gender and regional balance among employees.

Equipment

The survey had very few problems with equipment and logistics. Most of the computer and GPS equipment was available from UNDP. Other equipment and materials were bought locally or were shipped from Denmark. The survey had no problems with customs regarding the import of equipment.

Compound

In the surveyed regions, office space is only available in fenced compounds originally meant for residential use. The survey was able to rent a compound that perfectly matched survey needs. The compound had five offices and a classroom in the main house, and storage space and extra offices in smaller buildings. Parking was available for several vehicles. Survey staff installed a generator that provided reliable electrical power throughout the project.

Computer equipment

Thirteen computers were used in the project. Other database-related equipment was provided by SMAC. The survey purchased laptop computers for fieldwork. A network was established in the compound where all computers were connected to a central server and a number of shared printers. The database center, with four workstations, was also established and connected to the network.

Survey equipment

Every interviewer team was equipped with a compass, a Global Positioning System (GPS) receiver, and a digital camera. Supervisors also used GPS receivers.

Communication

The survey maintained constant contact with field staff throughout operations. As the survey changed vehicles several times, it was not practical to mount HF radio sets in the vehicles and VHF units did not have the range. Instead the survey used satellite phones, which were programmed to allow calls only to a defined list of other sat phones in the Survey Group and headquarters. At HQ, the survey used a Globe Connect Base station that allowed a satellite connection in the operations room. Using these satellite phones turned out to be a cheaper investment than what would have been required for HF radios, as the satellite phones cost a third as much to purchase and operate.

Transport

The data collection period was too short to require the survey to purchase its own vehicles for the project. Instead, the survey rented four-wheel-drive vehicles on the local market. Vehicles were available for rent along with drivers throughout the country, and so the survey contracted vehicles and drivers locally when the survey groups operated in a certain area. Renting vehicles, instead of purchasing them, resulted in significant cost-savings for the survey.

Survey vehicles

Field accommodation and food

During field operations, the field staff lived in the districts they were surveying. When the necessary accommodation and facilities were not available, the survey groups were issued with equipment and tools to allow them to establish a completely self-sufficient camp. This included tents, camp beds, blankets, mosquito nets, a generator, light fixtures and cabling, plastic chairs and tables, all necessary cooking equipment, water tanks, and tools for building toilets and washing facilities. As it turned out, field supervisors were always able to find unused buildings, often-unused schools or public installations, for use by the survey group. The camps were usually a combination of tents and permanent structures. The LIS provided both survey groups with a cook. Each staff member also received a daily subsistence allowance.

COORDINATION

Government authorities

The survey coordinated closely with the Ministry of Resettlement, Rehabilitation and Reconstruction (MRR&R), the national authority responsible for mine action. DDG has a long-standing relationship with MRR&R and had signed an MOU covering demining operations in the surveyed regions. An MOU specifically relating to the survey was negotiated with MRR&R, establishing a legal foundation for the survey.

A liaison officer was appointed for regular contact between the government and the LIS. The survey established a good working relationship with the liaison and was able to solve all outstanding issues with the government in an expedient manner.

Local officials were also kept informed of survey progress and provided assistance when needed. This contact was especially useful during the public information campaign, launched to coincide with the start of field operations.

The national mine action structure, consisting of a UNDP-supported Somaliland Mine Action Center and a MRR&R-controlled National Demining Agency, entered a restructuring that lasted almost the entire survey operations period. They therefore did not play the active role that the survey originally envisioned. On 18 February 2003 a Presidential decree established the government's Somaliland Mine Action Management Structure which will receive the survey IMSMA database and assume responsibility for its update and maintenance.

UN Quality Assurance Monitor

The UN Quality Assurance Monitor (QAM) did not join the LIS until almost five months into the project. In the meantime, however, the necessary documentation was collected and kept on file by the administration officer. The QAM arrived on the first day of the pilot test and immediately joined the field pilot test camp. To maintain an independent role, the QAM established an office at UNDP, but visited the survey frequently to discuss issues and collect survey samples to check.

The survey had a very good working relationship with the QAM. The survey management and the QAM constantly discussed LIS issues and then took corrective action. The QAM process was a supportive resource for the survey.

Survey Action Center

The survey had very good relations with SAC, and SAC staff assisted the survey on a number of occasions. DDG submitted a weekly status report to SAC giving the current status of the project compared to the timeline. A monthly progress report was also submitted. SAC staff executed the senior staff training, and provided training materials (lessons, exercises, etc.) that were used for interviewer training. The SAC training team visited for three weeks, during which the survey also had time to discuss various other issues relating to the survey. Dr. Aldo Benini's analysis of the questionnaire and proposed corrections based on the results of the pre-tests were enormously helpful. SAC staff members were also available to answer questions relating to the survey mechanisms and operational issues. During the reporting phase, SAC deployed two persons to assist with data analysis and report writing.

CAPACITY BUILDING

The survey incorporated the information department of SMAC into its own IMSMA/database unit, ensuring that the survey results can be maintained in the region. After the survey concluded, all equipment necessary for the functioning of the SMAC information department, both from UNDP and from the survey budget, was transferred to the national authorities.

Field staff members were given a thorough training in the community interview and the use of equipment-GPS receivers, digital cameras, compasses, and maps-necessary for survey execution. These skills, combined with the experience gained through four months of field operations, have given the staff useful qualifications for work with other international and national organizations, and several found employment shortly after the completion of the survey.

OPERATIONS

Phase 1 of the LIS set out to conduct a standard survey, as described in the general LIS methodology and protocols. However, after field operations began, the survey was forced to abandon the EOC process and instead conduct a full census of communities. This change from normal methodology was recommended by the survey staff and approved by SAC.

FIELD ORGANIZATION

One survey group each was assigned to the eastern and western areas of the surveyed regions. One field supervisor, two field editors, eight interviewers and support staff made up a survey group. An extra supervisor, two field editors and four interviewers operated out of Hargeisa.

Three community visits were to be conducted each day, leaving the last interviewer team to complete paperwork in the camp. Community visits were conducted five days a week and a weekly SG meeting would occur on the sixth day. This timetable resulted in the total production of between 30 and 40 surveys a week.

TRAINING

Two training sessions were conducted. The first was for supervisors and field editors and included the data entry personnel. SAC staff conducted this training in English, with participation and input from the survey staff. While there were problems relating to the level of language proficiency and training on the part of the trainees, the training succeeded and culminated in the first pre-test, which provided important feedback for the adaptation of the questionnaire. It also revealed the need for a second pre-test, which was conducted one month later after some retraining and corrections of the questionnaire.

The two-week interviewer training program was conducted in the Somali language and the senior staff participated as instructors and assistants. The survey conducted a one-week pilot test as a final check on the training and all logistics, communications and operational guidelines. Every day half of the interviewers conducted community visits while the others evaluated completed surveys and continued training. Twenty community visits were conducted that week and all systems functioned well.

SURVEY DOCUMENTS AND PROCEDURES

The questionnaire was adapted from the LIS template to fit the local conditions. The questionnaire was then tested during the two pre-tests and the pilot test. The final questionnaire contained almost 200 questions. It was divided into four modules, which were printed on different colored paper and translated into the Somali language by a national social scientist.

The scoring sheet and scoring weights were also modified to fit local conditions. Also a system for not scoring suspected hazard areas that had no impact on the community was introduced. The latter change took the place of the standard LIS local adjustment factor mechanism.

Coding sheet

Instead of having the field editors transfer all the information from the questionnaire to a coding sheet, the survey printed the data entry instructions directly in the questionnaire. This reduced the coding sheet to one page that was inserted directly in the suspected hazard area module.

Other paperwork

The non-impacted sheet was a micro survey of communities that declared themselves to be free of mines or larger UXO contaminations. It collected the GPS location of the community, the population size estimate and some other basic data. The surveyed regions are heavily contaminated with unexploded ammunition of various types. Some are large, concentrated contaminations that the survey recorded as suspected hazard areas. However, much UXO was in the form of single items that were thrown under bushes, or placed on roofs out of the reach of children. These single items were not recorded as suspected hazard areas, but interviewers filled out a spot clearance request sheet with the GPS position, a description and a digital image. These reports were then sent to clearance organizations in Somaliland to rid the communities of these items.

UXO recognition sheets were used to give a short introduction to the group interview participants.

For the pre-visits, the supervisors presented a letter from local officials and another from the team leader requesting that the village provide all possible support to the survey. These letters stated that the survey does not guarantee and is not connected with any kind of assistance. These papers were seen in the villages as a treasured certificate.

OTHER PROCEDURES

Large-scale maps showing only a few main roads and villages and a series of 1:100,000 topographic maps showing a large amount of detail, including roads and tracks, were the two types of maps generally available. The latter maps were produced in 1976 by the Soviet Union, and were not used for logistics or operations. The only way to navigate safely in the surveyed regions is to ask for local information and only drive on tracks that are frequently used by other vehicles. An A3 print of the 1:50:000 Soviet maps with the community were issued to plot suspected hazard areas.

Standard operational procedures were developed by the survey to guide all aspects of the field operations. The survey developed SOPs on disciplinary actions, tracking of interviewer movements, emergency procedures, etc. In all, 32 separate instructions were issued and translated into Somali.

Once the survey got under way, a great deal of paperwork was generated and sent back and forth from Hargeisa to the field. A pouch system was developed to keep track of them all. Every time a pouch was dispatched, it would contain a list of all the papers it contained and an inventory check would be performed when it reached its destination. Because of this, no questionnaires were lost.

THE COMMUNITY SURVEY

The group interviews were conducted according to the protocols established by the Survey Working Group. A day or two before the community visit was to take place, the supervisor would visit the community to present the survey's credentials, confirm that it was impacted and request participation in the survey process.

The field supervisor would stress that the interview group should include persons from all parts of society, including women and nomads. On the day of the interview, the interviewer team would arrive some time before the meeting to acquaint themselves with the community.

The formal meeting would start with drawing a community map on a poster size piece of paper, guided by key informants. After that, the actual interview, based on the questionnaire, would begin and normally last for one to three hours. During this phase, the interviewer team would often treat the participants in the meeting to tea as is customary in the region. After the meeting the Interviewer team would go to a safe viewing point in the vicinity of the identified SHAs to conduct visual inspection. A safe distance of 100 meters was mandated. During this inspection, a digital image, the GPS position and a sketch map of the SHA were taken. It was virtually impossible for the interviewer team to assess the validity of the claims during this inspection, but it gave a much better idea of the location of the SHA. In communities with many suspected hazard areas, these inspections would take several hours.

ACCURACY OF INFORMATION

During the pre-tests, the survey found that communities exaggerated claims of mine impact. To counter this, the survey trained interviewers to verify the validity of claims through a series of crosschecking questions. Interviewers also talked to people after the meeting to verify claims. These techniques seemed to bring the problem under control. It is important to recognize, however, that the survey captures only what the local population tells the interviewers. Thus, the LIS captures

GAZETTEER

The lack of a comprehensive gazetteer created a number of problems for the survey. All available data was used to compile an operational gazetteer for the survey, but this turned out to be incomplete and, in many cases, misleading. In the areas surveyed, 193 communities not identified by the gazetteer were found and 72 communities identified in the original gazetteer did not actually exist. Further discussion on this and other related issues, including the survey's response, are covered in the rationale section above.

QUALITY CONTROL

The survey faced difficulties with the quality of work submitted by survey groups. Returned questionnaires often contained a number of mistakes, omissions and internal inconsistencies, and field editors would only correct some of these problems. There was a similar lack of attention to detail by the data entry operators at headquarters in Hargeisa.

To respond to these problems, the survey created extra quality control units at various points in the process. When completed surveys were delivered to HQ, a QC unit checked them: mistakes would be listed and they would then return them to the field team for correction. In some cases, interviewers had to revisit the community. In the data entry unit, a report generator was created that printed out the data entered in a format similar to the original questionnaire. A different QC unit would then check the entered data with the original questionnaire. The QC units would often find between five and fifty mistakes in a single questionnaire. The LIS national staff repeatedly said that the attention to detail required of them was unprecedented in the history of Somaliland.

RECOMMENDATIONS

This survey has three main recommendations regarding LIS methodology and operations:

1. Surveys should think carefully about compressing LIS projects into very tight timelines. In particular, enough flexibility should be given to adjust to operational problems that arise during data collection.

2. In countries where the civilian administration is still weak, where no proper gazetteer exists and where the knowledge among civil servants about the mine problem is limited, surveys should carefully consider adapting the standard Expert Opinion Collection and False Negative Sampling procedures to local conditions.

3. The QAM should visit the survey shortly after project start-up to establish procedures required for certification. The QAM should also be present when the questionnaire is being adjusted and when field staff is in training.

The survey has four recommendations regarding the administration and use of the data from Phase 1 of the LIS:

1. The national mine action framework needs to stabilize and begin operating in cooperation with all involved parties.
2. The LIS data should be used to create a national priority list for technical surveys and clearance in conjunction with the national government and mine action agencies.
3. The IMSMA database should be kept fully updated through receiving and incorporating data from all mine action organizations in the survey regions.
4. It is important that the situation in Sool and Sanaaq be monitored to establish when a survey can be completed there. I would like to thank national and international staff working on this project for all the hard work that was put into the successful completion of Phase 1 of the Landmine Impact Survey. It is now up to the local authorities to make use of the wealth of data collected.

PAUL MACKINTOSH, Team Leader
Landmine Impact Survey
Danish Demining Group

The successful completion of t he Landmine Impact Survey in t he Awdal, Galbeed, Sahil, and Togdheer regions (Phase 1) and the publication of this report would not have been possible without the input and effort of a large number of people and institutions. With respect to this, we would also like to acknowledge the invaluable contributions of the following:

Project Support and Facilitation

- Ahmed Yusuf Yassin, Vice President of Somaliland
- Mohamed Said Gees, Minister of Foreign Affairs
- Ahmed Haji Dahir Elmi, Minister of Planning
- Abdillaahi Hussein Iman, the Minister of Resettlement, Rehabilitation and Reconstruction (MRR&R)
- Mohamed Osman Ahmed, Program Manager, Somaliland, Mine Action Center
- Engineer Jidhif Hassan Abdillahi,Former Mine Action Advisor,MRR&R
- Mohamed Ali Ismail, Assistant Program Manager, Danish

Demining Group, Somaliland

- John Dingley, UN Chief Technical Advisor
- Eddie Boyle, Deputy Programme, Manager, RoLS, UNDP
- John C. Brown, UN Quality Assurance Monitor
- Bodil Jacobsen, Administration and Finance officer/Operations officer
- Franco Serena, Consultant, UNDP/Swiss Government
- Kristoffer Waage Beck, DDG
- Wais Ibrahim H Mussa, national social scientist
- Bo Bischoff, DDG
- Robert Eaton, SAC
- Nelly Schneider, SAC
- Danish Demining Group
- The HALO Trust
- Saint Barbara Foundation

Drafting and Editing
- Paul Mackintosh, DDG
- Oystein Juul Nielsen, DDG
- Helene Ruud, DDG
- Dr. Saeed Ahmad, SAC
- Tom Haythornthwaite, SAC
- Mike Kendellen, SAC
- Dann Naseemullah, SAC
- Peter Harvey, SAC
- Dr. Aldo Benini, SAC Consultant
- Eddie Boyle, UNDP
- John Dingley, UNDP-UNOPS Graphic and Report Design
- Betsy Rubinstein, InForm
- Rashne Green, InForm

A special thanks to all the Somaliland nationals who made up the field teams and office staff, without whom none of this would be possible.

SOMALILAND MINE ACTION CENTER
Hargeisa, Northwest Somalia
Phone: (+252 8) 220 030
Fax: (+252 2) 523 174

SURVEY ACTION CENTER
6930 Carroll Avenue, Suite 240
Takoma Park, MD 20912-4468, USA
Phone: (301) 891-9192
Fax: (301) 891-9193
Website: www.sac-na.org

DANISH DEMINING GROUP
Borgerade 10, Post Box 53
1002 Copenhagen, Denmark
Phone: (+45) 3373 5329
Fax: (+45) 3393 8730
Website: www.danishdemininggroup.dk/

UNITED NATIONS DEVELOPMENT PROGRAMME SOMALIA
Centenary House, 2nd Floor
Ring Road, Westland Lane
Nairobi, Kenya
Phone: 254-20-4448433
Fax: 254-20-4448439
Website: www.so.undp.org

UNITED NATIONS MINE ACTION SERVICE
304 East 45th Street
Room FF-360, United Nations
New York, NY 10017, USA
Phone: (212) 963-1875
Fax: (212) 963-2498
Website: www.un.org/depts/dpko/mine

UNITED NATIONS OFFICE FOR PROJECT SERVICES
The Chrysler Building
405 Lexington Avenue, 4th Floor
New York, NY 10174, USA
Phone: (212) 457-1251
Fax: (212) 457-4049
Website: www.unops.org

SURVEY ACTION CENTER
United Nations Mine Action Service,
Department of Peacekeeping Operations
Somalia