Quantifying Residual Risk in Land Release

By Dr Robert Keeley, RK Consulting (EOD) Ltd

 

INTRODUCTION

Land release allows mine action activities to focus on areas that are actually contaminated. Land release is therefore an economic process in that it helps mine action programs to optimise the use of scarce resources, particularly area clearance.

The paper first defines risk and ‘residual risk’ using established guidelines from other sectors. It then briefly explores what can cause residual risk in land release activities. It identifies the problem when survey techniques are not held to the same standard as clearance techniques.

The paper introduces the idea of an ‘accident profile’ and identifies the three dimensions along which various actions can be taken to reduce the likelihood of EO accidents.

The paper then introduces the principle of ‘Circular Error Probability’ (CEP) already used in military operations research as a way of quantifying accuracy of weapon systems such as aircraft bombs and artillery. The paper posits that CEP analysis could be used to identify the ‘six sigma’ (one in a million) point of contamination in cluster munition or dense minefield contamination. The paper makes the point that ‘One in a million’ is a widely accepted residual risk in other sectors, such as the nuclear power industry and the insurance sector. This would allow the sector to quantify the residual risk from using ‘fade-out’ techniques.

The paper then explains how this principle could be extended to help quantify the residual risk implicit in using dogs to conduct area reduction, supported by fade-out. It is potentially usable on all cluster munition strikes and densely mined areas, which speaks to its possible scalability. The paper also refers to alternative risk management techniques where it is not suitable.

Additionally, the paper posits the use of the six-sigma concept at a strategic level to identify the ‘end state’ for national mine action programs in terms of casualty numbers, thus providing an objective means to justify continued support for countries that are continuing to experience EO casualties.

Finally, the paper briefly summarises the practical research and data analysis that would be needed to establish and confirm the ‘ground truth’ of the ‘six sigma’ approach to quantifying residual risk from land release. The research and data analysis methods required are eminently feasible as they build on established techniques from other sectors (although they are not currently employed in humanitarian mine action).

BACKGROUND

Land release as a concept was first written about in a 2004 evaluation report on mine action in Cambodia1. It can be summarised as a means of determining where mines (and other items of explosive ordnance (EO) aren’t. Thus, land release allows mine action activities to focus on areas that are actually contaminated. Land release is therefore an economic process in that it helps mine action programs to optimise the use of scarce resources, particularly area clearance. Land release is now defined in International Mine Action Standards (IMAS)2 as:

“...the process of applying “all reasonable effort” to identify, define, and remove all presence and suspicion of Explosive Ordnance through non-technical survey [NTS], technical survey [TS] and/or clearance”.

AIM

The aim of this paper is to introduce a new approach to quantifying residual risk in humanitarian mine action (HMA), building on methods already used in other sectors, and proposing further study.

DEFINING RISK

In order to quantify residual risk, it is first necessary to define risk and set out what is meant by residual risk. People often use the terms ‘risk’, hazard’ and ‘threat’ interchangeably. However, risk is a mathematical concept and as such has a formal definition. IMAS 04.103 gives a version of that definition, which it takes from ISO Guide 51:1999:

A “combination of the probability of occurrence of harm and the severity of that harm”.

Expressed formally, this takes the following form:

R = f(ip, os)

Where R = risk, ip is the probability of the incidence and os is the severity of the outcome.

Practically speaking, risk can also be considered a function of hazard plus activity: it doesn’t matter how contaminated an area is if nobody goes there (from a risk perspective).

IMAS also suggests a minimum standard for clearance. It defines a cleared area as follows:

            “A defined area cleared through the removal and/or destruction of all specified EO hazards to a specified depth” 4.

Clearance does two things: it removes the EO hazard and allows the productive use of the land. If it is accepted that EO risk is a function of hazard and activity it follows that any form of ‘rough and ready’ clearance actually increases risk. People are at zero risk from EO if they don’t enter a contaminated area. Entering an area that has, for example, only had 80% of its contamination removed means the risk is greater than zero.

Quantitative definitions of residual risk exist in industry. There is a generally accepted rule of thumb that a tolerable residual risk is of 1:1,000,000 per year5. Two principles are understood in the concept of residual risk:

  • Even when the risk is tolerable, it must be reduced to a level which is as low as reasonably practicable (ALARP).
  • A point is reached at which the risk is, or has been made, so small that no further precaution is necessary.
What can cause there to be a residual EO risk?

In order to quantify residual risk, it is important to understand what can cause it. There are three main circumstances which can make the residual risk in an area declared clear. These are discussed below.

  • Poor mapping. This covers a situation where the contaminated area has not been adequately defined. This will be explored in a separate paper.
  • Poor clearance. This covers a situation where clearance processes do not produce ‘cleared land’ within the definition set out in IMAS. One might expect accreditation and quality management processes to address this risk.
  • Deep buried items. Perhaps the main likely residual risk is where items of EO are buried deeper than the depth specified in national standards. This can happen for several reasons but includes deep buried projectiles, mortars or aircraft bombs that are outside the detection depth of standard mine detectors. This can be managed by careful scope of works on infrastructure and construction sites to use ‘bomb location’ techniques on areas such as building foundations that need to be excavated. Bombing data can help predict where such precautions might be necessary.

Table 1. Potential boundary setting errors in NTS/TS

Residual risk from NTS and TS

However, as set out above, there is more to land release than just clearance. Both NTS and TS are important ways to make land release efficient and achievable.

What sets clearance apart from both these survey techniques is that clearance is the only one of the three that always involves a 100% search of the target area. Survey thus faces a potential dilemma which is summarised in the matrix set out in Table 1 above.

The matrix set out in Table 1 includes two positive survey results: when the survey accurately reports an area as being clear, and when a survey accurately reports an area as being contaminated. The question of survey efficiency – reporting an area as contaminated when in fact it is clear, can, it is suggested, be addressed by training – particularly in boundary setting. Previous experience suggests that individual surveyors are so concerned that they will inadvertently declare contaminated areas clear, they will instead make the boundaries of suspected hazard areas so large as to make the survey results largely meaningless. Improving the efficacy of NTS is outside the scope of this paper.

This leaves the situation when a survey fails to report a contaminated area. It raises the following question: do we have any right to declare an area clear with different standards of risk? It is held that whatever the method used, the mine action practitioner must deliver land of the same standard of residual risk regardless of the method used.

Managing risk in land release

Taking EO risk as being a function of a function of hazard and activity, and adding a third element, that of behaviour, it is possible to understand EO risk in three dimensions, as shown in Figure 1 below. It is then possible to understand these three dimensions as preconditions for an accident to happen (the ‘accident profile’). Firstly, there must be contamination present (the hazard), secondly, there must be a reason for activity to take place on the contaminated land (usually an economic reason for using the land) and thirdly, the at-risk population must exhibit a form of risk-taking behaviour (as defined by UNICEF in 2005) 6.

Figure 1. The three dimensions of EO risk. Only where the three preconditions are in place will there be an EO accident.

In turn this model helps highlight that there are three axes of action that can be undertaken to reduce risk. Firstly, of course the hazard could be removed, but it is also possible to act on the other two axes: EORE can be used to help modify behaviour and livelihood programming (not always in the mine action toolbox) can be used to reduce the economic demand for the land.

Quantifying residual risk

Three methods for quantifying residual risk, and their likely use cases, are described below.

Fadeout and the principle of circular error probability (CEP)

This method is particularly useful in surveying cluster munition strikes or mined areas. The accuracy of ordnance, particularly air-delivered ordnance is described in terms of its circular error probability (CEP). If an aircraft aims a bomb at point X, the CEP will give a statistical measure of the chance of it hitting the target. See the description of CEP in Box 1 below.

Box 1. Definition of Circular Error Probability (CEP)

Figure 2. Distribution of ordnance predicted by CEP.

Mine action programs generate large amounts of data. It is very common to mark the location of every piece of EO found during area clearance, as shown in Figure 3 (below).

Figure 3. An example of a tactical minefield. The wooden stakes mark where mines were found.

Table 2. Worked examples of standard deviation

Putting these two elements together, it can be possible to plot the distribution of distances between any two pieces of EO. Once these distances are compiled, it is possible, using an Excel spreadsheet, to calculate the mean and the standard deviation (σ). The standard deviation is a measure of variance from the mean. Using a normal distribution, six standard deviations from the mean will capture 99.9999998% of the items: in other words, with a fadeout distance of 6 σ the chances of an item falling outside the fadeout area is less than 1:1,000,000 – i.e. within standard measures of residual risk..

Two worked examples are shown in Table 2. In the first example, the items of EO form a dense pattern; in the second, they are more widely spread. The optimal fadeout distance is calculated for each example. This is more accurate than the current ‘rule of thumb’ method used that can be found in many mine action programs.

Limitations7

This method can be applied to cluster munition strikes and to large, comparatively dense mined areas. Its suitability decreases with fewer data points. Thus, in areas where contamination is typically one or two mines placed to harass (sometimes referred to as ‘nuisance’ mining) it won’t be possible to use this method. However, as set out above there are three dimensions to EO risk, one of which is ‘activity’. It is possible to use NTS techniques to identify why a population wants to access a particular area and use a project-oriented approach to clearing land.

This project-oriented approach is discussed in more detail in a separate paper by the author8.

Residual risk for area reduction

The first tool described above is useful when dealing with a confirmed hazardous area (CHA). It is less helpful when a suspected hazardous area (SHA) needs to be investigated, or when the boundary of a CHA cannot be easily defined or is within a larger SHA. In such a case a mine action program may wish to use an area reduction methodology such as dogs (perhaps with machines as ground preparation so that the dogs can gain access). Numerous incidents in the past have shown that dogs and machines are not totally reliable as primary search/clearance techniques, but it may be less than desirable to use slow manual area clearance techniques to fix the boundaries of a CHA.

Nevertheless, it is possible to utilise area reduction tools using the same fadeout principle described above. Consider an area already identified by NTS (with indirect indicators of contamination) as ‘probably mined’, or an SHA which is believed to have a CHA within it whose boundaries are undefined. If dogs are used to search all that area (perhaps following machines) and reduce that area, they can be expected to define a boundary for the CHA. The area within the new boundary includes the ‘definitely mined’ CHA. The area outside the new boundary but within the area identified by the NTS can be described as having originally been hypothesised as being contaminated but now the evidence from the dog search. However, as stated above, there is the question of whether the efficacy of the dog search (its ‘sensitivity; in statistical terms) is sufficient to confirm that the reduced area is now safe to the same standard as might be established by manual demining. This can be resolved by using the same ‘fade-out’ process described above. The outer boundary of Area A in Figure 4 below represents an area identified as ‘possibly mined’ during an NTS. Area B represents an area identified as being ‘definitely mined’ by dogs. In this example. A centre line XX1 is established, and the area is then cleared to a distance established by the fade-out process described above. In this case area Y represents the total area searched by manual deminers. The residual risk can now be quantified as 1:1,000,000. Note that the entire area represented by A-B is searched: this is not a sampling technique.

Figure 4. Using ‘fade-out’ to check area reduction.

Casualty numbers and a strategic perspective on residual risk

Table 3. Lao PDR population and EO casualties by province, 2023

The two previous examples show how a quantified risk threshold, based on levels accepted in other sectors (i.e. 1:1,000,000) can be used on a site to identify the boundaries of a contaminated area. However, it is also possible to use this concept at a strategic level. Resources permitting, everyone in the mine action sector would like to see casualties from mines and other EO reaching zero. However, resources don’t always permit this and even where treaty obligations can be met, there remains the hazard posed by other forms of EO. Contaminated countries have competing requirements on funds. Even in non-contaminated countries, it isn’t possible to have an infinite number of ambulances or kidney machines. This question of the allocation of scarce resources is a core issue in health economics.

Consider the example of Lao People’s Democratic Republic (Lao PDR). Lao PDR has a population of just under 8 million people9. According to the Landmine Monitor, Lao PDR suffered 47 casualties in 202310. Using the same 1:1,000,000 risk threshold, 47 is 6 times greater than the tolerable risk. This supports the case for continued funding of the UXO program in Lao PDR. Achieving a national UXO risk threshold in Lao PDR of 8 casualties per year could be considered a ‘first order’ target. But not all provinces are contaminated. The table at Table 3 disaggregates both population and casualties using 2023 data. Thus, a second order target could be identified by applying the same risk threshold to the population of the seven provinces that had suffered casualties, i.e. 3 per year (which is nearly 1/16th of the current casualty rate). The calculation in column (e) of the table shows the relative risk exposure in each province and is found by dividing the casualty number by the population size. This controls for difference in population sizes, although it can be seen in the case of Lao PDR in 2023 that Xiengkhuang province had both the larger absolute casualty numbers and the largest relative risk exposure. Quantifying the relative risk exposure is suggested as a useful way of allocating resources in a national mine action program.

WHAT NEXT?

Firstly, it would be necessary to conduct data analysis of finds to confirm the applicability of the CEP principle to contamination. Ideally this would be done by secondary data analysis, in programs where the locations of the ‘yellow pickets’ were adequately recorded. Where there are insufficient records, it would be necessary to work with willing clearance organisations to facilitate the recording of adequate data.

Secondly, a similar data analysis could be conducted on the combination of area reduction by dogs supported by fade-out clearance to increase the understanding of dog’s ability to establish a boundary of contaminated areas. This should allow the residual risk of using dogs for area reduction, and any necessary buffer zone, to be quantified.

Finally, the concept of residual casualties could be an important way to help quantify an ‘end state’ for national mine action programs, especially in situations of scarce resources. It is suggested that this could be a subject for a discussion between donors and national mine action programs. It may therefore be useful to facilitate a workshop to allow such discussions.

CONCLUSIONS

It is possible to quantify residual risk in HMA, borrowing approaches from other sectors, such as the nuclear industry. In these other sectors it is done by identifying the tolerable risk threshold (which is commonly set at 1:1,000,000) and then analysing processes to be sure that this standard can be met. Borrowing again from the established CEP technique and applying statistical analysis to EO distribution patterns it should be possible to establish a quantifiable ‘fade-out’ when the risk of an item being outside that pattern is less than 1,000,000. This technique cannot reliably be applied to areas with very low-density contamination but understanding risk as a function of hazard, activity and behaviour allows the use of a project-oriented clearance approach to minimise risk in such circumstances.

More research could be done on identifying minimum data sets and on independent research of dogs to establish a replicable means of measuring their ‘sensitivity’ (in the statistical rather than the olfactory sense).

Finally, it is also possible to use the concept of tolerable risk threshold in the same manner as used in health economics to measure the relative risk of EO accidents compared to other causes of injury or death. In this way it can help justify continued expenditure on HMA programs and also help with allocation of mine action resources at a provincial level. ■

REFERENCES
  1. Joint Evaluation of Mine Action in Cambodia for the Donor Working Group on Mine Action. R Griffin and R Keeley, December 2004. See https://commons.lib.jmu.edu/cisr-globalcwd/1154/?utm_source=commons.lib.jmu.edu%2Fcisr-globalcwd%2F1154&utm_ medium=PDF&utm_campaign=PDFCoverPages
  2. IMAS 04.10 Second Edition (Amendment 12, October 2024)
  3. Ibid.
  4. Ibid.
  5. See for example: https://nucleus.iaea.org/sites/graphiteknowledgebase/wiki/Guide_to_Graphite/Tolerability%20of%20Risk%20the%20ALARP%20Philosophy.aspx
  6. https://reliefweb.int/report/world/imas-mine-risk-education-best-practice-guidebook-1
  7. Thanks to Dr Russell Gasser for discussion on statistical analysis.
  8. ‘Infrastructure and commercial projects: EO clearance and risk management’ by Dr R Keeley. Published in ‘Counter-IED Report’, April 2025.
  9. Source: Lao Statistics Bureau https://laosis.lsb.gov.la/statHtml/statHtml.do?orgId=856&tblId=DT_YEARBOOK_C001&conn_path =I2&language=en
  10. https://the-monitor.org/country-profile/lao-pdr/impact?year=2023

 

ABOUT THE AUTHOR

Dr. Robert Keeley is a former British Army Bomb Disposal Officer active in humanitarian and commercial mine action and explosive ordnance disposal (EOD) since 1991. He has worked in numerous countries and for several governments and international organisations. He specializes in project design, evaluation, and quality assurance of all aspects of mine action and has helped shape the emerging humanitarian improvised explosive device (IED) sector. Dr. Keeley is a member of the Institute of Explosives Engineers, a Member of the International Association of Bomb Technicians and Investigators, and is a Fellow of the Royal Geographical Society. He holds a PhD in Applied Environmental Economics; his thesis was on “the Economics of Landmine Clearance.”

Contact Information
Dr. Robert Keeley, Director, RK Consulting (EOD) Ltd Ashford, Kent, United Kingdom
[email protected]
www.rk-consulting.net

Download PDF: 33-40 Robert Keeley article – COUNTER-IED REPORT, Winter 2026