RECENT UXO/MINE CLEARANCE FINDINGS
Over recent years, the community working towards the aims of the Anti-Personnel Mine Ban Convention1 (APMBC) has begun to struggle with fundamental questions related to the efficiency of clearance efforts and the need to release land in countries facing strong demographic and social pressures (Jordan Times, 2004).
In a series of studies on Lao PDR it was found that physically cleared ground more than 292 km2 was less than 2% contaminated with Explosive Remnants of War (ERW), denoting a rampant and chronic waste of resources and life-saving efficiency in an industry that lives on the good will of international donors. Very costly mistakes are by and large made at tactical (local/community) decision making level, i.e. when deciding whether to clear, sample or release a Suspected Hazardous Area.
This appears to be indicative of operators clearing ground where, in all likelihood there may be no ERW, instead of undertaking a reasonable, transparent and sustainable analysis of available information and then allocating resources that would maximize the result.
VARIOUS ENTITIES WORLDWIDE ARE INTERESTED IN ENHANCING PERFORMANCES OF HUMANITARIAN ACTIONS
Various entities worldwide are of course interested in releasing land and allowing clearance resources to be deployed to areas where mines and UXO presence is most likely.
However, unless the consequences of an ERW initiation are included in the information analysis (impact on population, development etc.), i.e. unless “Risk Based Decision Making (RBDM) for Land Clearing and Releasing” is performed, then rational, transparent and sustainable portfolio prioritization at local, regional or national scale cannot be achieved.
A preliminary version of the RBDM developed by Riskope (C.+F. Oboni) when consulting for GICHD and calibrated by them at programming using examples from Europe and using Whitman’s acceptability criteria reported 30% less mistakes made in selecting properties to clear than usual procedures ( Riskope Presentation at GICHD )
MODELS CAN BE MORE OBJECTIVE THAN HUMANS
The results confirmed the nowadays generally accepted concept that Humans seem to be overly confident in their decision-making skills and human misjudgment is a source of wide-spread budget and resource wastes. Those who resist this type of model’s findings, generally a majority at first, are typically quite willing to admit that models will be right more often than human experts “on average”, but love to point out special cases (Financial Times, How computers killed the experts, FT, London, Saturday September 1, 2007; P. E. Meehl, Clinical versus statistical prediction: A theoretical analysis and a review of the evidence. Northvale, NJ: Jason Aronson. (Original work published 1954), 1996.) and the inevitable exceptions. To the authors’ dismay the models were tampered with shortly after their release, most likely as a result of this train of thoughts.
WHAT CAN BE DONE TO ENHANCE RBDMs PRELIMINARY MODELS
The preliminary models were based on a priori probabilities to overcome the lack of data and the siloed data syndrome. Despite their limitations, such probabilistic models are perfectly acceptable, and are the only possible way to proceed in many industries. Possible enhancements would come from pertinent use of the available data and development of proper statistics.
The 2006 GICHD models showed it was possible to significantly reduce inefficiencies due to poor judgment in a UXO contaminated area with a probabilistic model using limited information. The authors believe a statistical model based on extant data, once the data silos syndrome is overcome and factual available data are pertinently processed, would certainly outperform the former and bring great advantage to the world of humanitarian demining. Moreover such an approach, pertinently modified could work for the landmine contamination case as well.
Over the last few months the meager set of results derived from the field trial of the preliminary RBDM method enabled the authors to pursue under their own financing and free of any contractual obligation the development of such an enhanced model for the evaluation of the probability of existence of an ERW on a considered property.
Interestingly the analysis of the limited field trial data lead to preliminary conclusions which either “confirm” the general experts’ beliefs on UXO presence or go in a counter intuitive direction, as exemplified below.
For example, Riskope found that:
• The probability to find a generic UXO (pGU) while clearing a piece of land where someone has declared seeing a UXO is 76%.
• The pGU while clearing a piece of land where no one saw a UXO is 53%.
These two statements “confirm” the general experts’ belief that an eye witness is important to guide decisions, but interestingly the probabilities are not so dramatically different.
Riskope also found that:
• The pGU on a parcel of land where an accident has been reported is 38%.
• The pGU where an accident has not been reported is 75%.
This second set says that one has better chances to find UXO on a piece of land where an accident was not reported than where an accident was reported. So, if these findings are confirmed, the “feeling” that if a site has a “history” of accidents, then it is a high priority will be proven wrong.
These examples show that measurable relationships exist between extant data. Models using these relationships, that may sometimes be counter-intuitive, would certainly outperform what has been produced to date and be self-adapting to any ERW, any environment, in any country, leading to a “universal model”.
Under the Riskope’s R&D program analyses of the data relationships allowed to derive two pioneering models of the second generation.
By obtaining more data, further data analyses can be performed and relationship can be studied, proving or disproving good old “experience based” flair and methodologies based on non-scientific agglomeration of data under the form of tables, flowcharts etc.
Riskope pledges time for the development of a predictive method for UXO/ Landmine contamination based on the data that the public/NGOs/other entities will return to them for treatment (contact us via Riskope web site to receive instruction on formats and data requirements). The results of this endeavor will remain available to all the contributors and will not be the object of any commercial activity by Riskope.
If we all work together we will be able to deliver a tool that will save lives by allowing optimum deployment of demining/clearing efforts in entire countries.