ReliefSource

Introduction: what Brahimi did

“Peace operations could benefit greatly from more extensive use of geographic information systems (GIS) technology, which quickly integrates operational information with electronic maps of the mission area, for applications as diverse as demobilization, civilian policing, voter registration, human rights monitoring and reconstruction.”  (Brahimi et al. 2000)

Although it was only a single paragraph in the 58-page Brahimi Report[1], this quote opened the floodgates for the introduction of GIS into humanitarian and peace operations.  A number of other factors encouraged this process, including the decreased cost of high-resolution commercial satellite imagery, the increased availability of GPS technology, and the wider information revolution that has seen the introduction of information and communications technology (ICT) at every level.  Since 2000, GIS has had an increasingly high profile in the humanitarian sector, through field operations such as the UN Humanitarian Information Centres (HICs) and UN Joint Logistics Centre (UNJLC), NGOs such as Vietnam Veterans of America Foundation or MapAction, and inter-agency bodies such as the Geographic Information Support Team (GIST).

The GIS Report Card: Could Do Better

Despite all this activity, it is hard to identify clear successes in the implementation of GIS in humanitarian work.  GIS has certainly been embraced by the mine action sector[2] but, despite general enthusiasm for the technology, other humanitarian sectors have not been so quick to adopt the technology.  This paper hopes to identify the reasons behind this, and to propose approaches that may help us to move forward. 

In the last five years, I have had the opportunity to be involved in helping to develop and implement GIS in a variety of (post-)conflict situations, largely through the Humanitarian Information Centres run by the United Nations Office for the Coordination of Humanitarian Affairs (UN OCHA).  GIS has been at the core of these HICs, and the quote from the Brahimi Report above shows that GIS has been recognized as a potentially powerful tool to support such operations.  The question is, how far has that potential been realised?

The HICs have been moderately successful in delivering GIS to the field environment, drawing on support from a range of organisations.  However, beyond the HICs and a handful of other organisations, my experience has been that GIS has not been widely implemented in humanitarian operations.  Where we do find GIS in the field, it seldom goes beyond thematic mapping, as a cursory examination of the maps produced by the various HICs will demonstrate.  Basic orientation, security, demographics – these are the standby HIC products, clearly very useful (if not essential) for aid workers, but with minimal added value. 

This observation is not intended to diminish the importance of the work that has been done, much of it behind the scenes, by GIS practitioners working in difficult conditions on steep learning curves.  Much of this work has been in the area of creating basic spatial data – particularly those laborious tasks of acquiring, cleaning and aggregating data for general use.

GIS experts tend to lay the blame for the “failure” of GIS at the door of those organisations they are working for, pointing out that these organisations need to change their working practices in order to use GIS properly.  Conversely, aid workers perceive that GIS (and, by extension, the GIS community) has failed to live up to its expansive promises, particularly in terms of delivering useful analysis that can inform programming decisions. 

Both sides would agree that the full analytical power of GIS is rarely, if ever, harnessed.  There is an element of truth on the accusations on both sides, but the state of GIS in the humanitarian sector suggests that both the humanitarian community and the GIS community are overlooking some basic truths about the successful management of information systems, particularly in the high-stress environments of humanitarian crises.

Indecision Making Processes

Shawn Messick, one of the most experienced GIS practitioners in the humanitarian sector, wrote that “It’s troubling that geographic accuracy continually appears to be only an afterthought in most decisions about aid programs.” (Messick 2004)  He was correct in identifying this as one of the key problems we have faced, but the reasons behind it bear further discussion.

It clearly is troubling for the GIS community, since geographic accuracy is a prerequisite for successful implementation of GIS.  So why is it not troubling for the humanitarian community?  It is partly because all humanitarian work is predicated on uncertainty – it is an integral part of the working environment for most aid workers.  It is partly because of a lack of accountability within the humanitarian sector, identified by numerous evaluations, which means that organisations do not take a systematic approach to their work. 

The main reason is that the humanitarian community is not used to making decisions based on evidence, relying instead on experience of past events as their guide to the future.  This is entirely understandable, particularly since data in emergency situations will always be incomplete, but it leaves decision-makers vulnerable to errors when situations change.  In Afghanistan in 2002, we were tasked by the UN Humanitarian Coordinator to identify the most vulnerable districts, through District Vulnerability Mapping.  Discovering that there was little data, of poor quality and with patchy coverage, we developed proxy indicators from available sources and put together the maps that had been requested.  When these maps were presented to senior management, the immediate response from long-serving Afghan hands was “These maps are wrong!”

We tried to make the case that either the maps were wrong, which may well have been true – in which case, better data collection by the humanitarian community was needed – or the assumptions of the humanitarian community were wrong – in which case, the humanitarian community needed to review its preconceptions about the situation in Afghanistan.  We failed to make much impact with these arguments; aid agencies preferred to go on as they had been for the last 20 years, holding on to what they knew to be “true” about Afghanistan.  It was a clear illustration of the reluctance of humanitarian actors to take decisions that appear to be inconsistent with past experience – even if the context of that experience has changed completely. 

It was also an illustration of the problems that we faced when we could not acquire useful data that might have made our case more effectively.  The U.S. Geological Survey (USGS) defines GIS as “a computer system capable of storing, manipulating, and displaying geographically referenced information… Practitioners also regard the total GIS as including operating personnel and the data that go into the system.”  This definition identifies the three essential components of a GIS: technology, people and data.

Acquiring the first two is relatively simple, given the necessary financing – technology can be purchased and staff can be hired – but data, the third component of GIS, is difficult to get right.  To some extent, lack of data can also be addressed through spending – satellite images can be purchased, historical data can be collated, data processing services can be engaged – but there will always be barriers to acquiring timely and accurate data during a humanitarian emergency. 

External data acquisition (e.g. from remote sensing, whether governmental or commercial) suffers from large-scale errors in classification (i.e. features are misnamed or mislocated) or time (i.e. data was correct at the time of collection, but has since changed – a particular problem where there are gaps in the metadata and it is difficult to establish when the data was collected).  In either case, extensive additional work in the field is frequently required to correct these errors in order to make the data useable.  The availability of remote imagery and processing from providers such as the US government is not necessarily the solution, for two reasons.  Firstly, historical map sources may not have been entirely correct to begin with, whether because of lack of coverage or poor methology in previous cartographic efforts, because there are variant names, or simply for political reasons .  Secondly, in (post-)conflict situations, names may change frequently, entire villages may disappear or relocate, rendering the existing data unreliable.

We faced exactly this problem in Liberia.  A key dataset for humanitarian work is the location and status of health facilities, and in Liberia we were fortunate to find a list of facilities (albeit with no metadata, so no chance of finding out who had created it).  The list had a few problems: district boundaries were continually reassigned locally; district, village and clinic names had changed; facilities had been damaged or destroyed during the war; and there was no way of knowing how complete the list was. 

In these cases, it would have been futile for us to protest that only official boundaries be used for reporting, or that a nationwide survey be carried out to update the information; and so we faced a difficult decision.  Should we publish maps that we know to be inaccurate, ensuring our disclaimers are in place?  From my experience, all the disclaimers in the world do not stop the complaints from coming in that our maps are “wrong” – and that of course affects the credibility of our GIS with the humanitarian community.

Internal data collection (i.e. field based surveys) can provide the essential dynamic data required for humanitarian programming, as well as ground truthing for externally collected data.  However, such exercises have been and will continue to be hindered by the constraints of the working environment, with a recent study by the Humanitarian Policy Group of the Overseas Development Institute (ODI) in London identifying in particular “problems of access, the fast-changing nature of the environment, and the extreme variations in the type and quality of information available.” (Darcy and Hoffmann 2003) 

In addition, both types of data collection will always take time.  In his paper on the use of GIS in Kosovo, David G Smith of the US State Department noted that “Prior planning is key for getting all the actors on board and in agreement on what data needs to be collected; once the crisis is too far advanced, it may be too late because no organization wants to retrofit its already collected data into another format.” (Smith 2000)  Of course, GIS projects have long lead times while staff are placed, data collected and systems put in place.

The problem is that time is the one resource not readily available in a humanitarian emergency, where there is generally a direct inverse correlation between the amount of resources available and the amount of time available for investment.  This is another basic economic truth in humanitarian work; when taken in conjunction with the problem of opportunity costs, it is easy to see why many humanitarian actors have yet to be convinced that GIS can really help their work.

Opportunity Costs in Humanitarian Work

Opportunity cost is a term used in economics to identify the cost of something in terms of other opportunities that must be foregone – one of the very few “eternal truths” of economics.  Humanitarian organisations are economic actors like any other, and application of this concept to our activities might offer useful new perspectives on our work.  With limited resources, humanitarian organisations are regularly required to choose between different courses of action – where to locate medical clinics, or to distribute food aid – and, in each of these situations, there are other opportunities open to them which they must give up once they have made their choice.

The example that I gave above – of how GIS work has been focused on spatial data issues, rather than on GIS implementation – is a case in point.  I would argue that this data processing work should not be done in the time-hungry environment of the emergency response, and that placing resources behind these activities comes at the expense of value-added GIS work (from the point of view of end users in the humanitarian community).  Indeed, investment in GIS has actually been criticised as investment in an unproven new technology at the expense of more basic humanitarian activities – both in terms of diverting funds, and also in terms of the additional burden that GIS implementation is perceived to impose on aid workers in the field. 

If that is the case, why have we devoted so much time to these activities?  It is simply because the baseline data that we have to work with is frequently so poor.  Despite the recent adoption of key information technologies such as GIS, the ODI cited above found that there continues to be “a critical shortage of essential management information in certain key areas, most strikingly in the areas of primary concern in the humanitarian sector: mortality rates, morbidity patterns, levels of acute malnutrition” (Darcy and Hoffmann 2003).

Lack of baseline data makes any serious analysis difficult, and it particularly affects GIS.  There have been attempts to collect baseline data, with some success in locations such as Kosovo and Sierra Leone.[3]  Not every effort has been successful, however, and a useful illustration comes from the HIC Iraq in 2003, where a large-scale Rapid Assessment Process was planned for our re-entry to Iraq, to provide the humanitarian community with village-level baseline data, geocoded in order to map humanitarian needs to village level. 

Although supported by the Humanitarian Coordinator, the UN agencies and a number of NGOs, the Rapid Assessment Process never took off – partly due to inter-agency politics, but also partly due to access problems.  In light of the poor security situation in Iraq, the potential cost of sending out assessment teams was simply too high, not just in financial resources but also in terms of human life.  Given the available resources (which can charitably be characterised as poor), there was also a trade-off implied in the Assessment, between the rapidity of the process and the level of coverage that we could expect in a country the size of Iraq (Benini et al. 2004).

It was clear that assessments in Kosovo and Sierra Leone were successful partly because the conflict situation had stabilised, making access possible for a range of agencies; in Iraq (and most other places where the exercise was attempted), this simply never happened.  As a result, questions are now being raised about how useful and cost-effective these types of exercises are in providing the humanitarian community with the information it needs.  The ODI report points out that “the benefits of assessment, like every other sphere of activity, have to be weighed against its costs.”   

This is certainly true, but I have argued in another paper that, while the development of better information management systems – particularly data collection efforts – may be an additional burden on humanitarian organisations, if those systems make our work more effective, then we have an obligation to support them (Currion 2003).

I have frequently heard the argument, from senior level decision makers and field workers alike, that the costs of ensuring baseline data is available – and, more importantly, that data is collected and shared in a systematic manner by their own organisation – are too high, and that those funds could be better spent in actually responding to humanitarian needs (even if those needs are poorly understood due to lack of data). 

In effect, these people are arguing that GIS implementation has an opportunity cost in terms of the rapidity of our response.  What this means in practice is that it is incumbent on the GIS community to demonstrate that GIS will, in fact, make our work more effective in the midst of a humanitarian emergency.  The value of GIS will rely on whether relevant, useable information of sufficient quality is being generated, despite the constraints in the field that I have outlined above; only this will ensure that the necessary resources are available to actually make GIS work in the humanitarian environment.

Two Models for the GIS revolution

Questions of cost are not the only barrier to the entry of GIS into the humanitarian market; questions of culture also have an impact.  Earlier I mentioned the comparative success of GIS in the field of mine action, and it is worth exploring a possible reason for this particular success story.

Robin Schofield of Accenture has identified two approaches to information management in the humanitarian sector, based on differing ‘mental models’ (Schofield 2002).   The ’systems’ model presents a hierarchical system, derived from structures found in governments and the military, relying on organisations buying in to the system completely.  In contrast, the ’service’ model is drawn from commercial news services (such as Reuters), allowing for greater independence on the part of individual organisations as they decide whether or not to participate in the system. 

GIS has been very successful with government and military organisations, because GIS tends more towards a systems model – strongly hierarchical, with a central authority acting as a guardian for data standards and maintenance, requiring regular reporting in a set format.  The mine action sector (specifically in mine clearance) draws deeply from the military, with many staff in the sector coming from military backgrounds.  It is not a huge leap of imagination to suggest that the success of GIS here is not due simply to the utility of GIS itself, but the characteristics of the mine action sector and the personalities working within it.

This is not to say that GIS can not be used in other sectors, but a similar argument applies.  The World Health Organisation (WHO) has also introduced GIS as a key tool in developing Health Information Systems (even going so far as to develop a dedicated GIS application, HealthMapper) – but this is for large-scale public health programmes that are either managed by governments or are intended for future handover to governments. 

Schofield goes on to suggest that “Experience in the commercial world casts doubt on whether greater systems integration across such a fragmented industry as humanitarian assistance is feasible.”  This raises the question of whether the type of GIS project we have seen so far is, in fact, a suitable model for the humanitarian sector, where a large amount of the field work is carried out by UN agencies and NGOs – who most definitely do not buy in to the systems model.

As I pointed out above, decision-making processes in the humanitarian sector are often very weak; management structures are badly formed and resource flows frequently prevent organisations from investing consistently in information management.  The humanitarian community does not manage information assets well, and humanitarian operations frequently suffer as a result.  However GIS is not the answer to these problems and, in some cases, GIS may distract humanitarian organisations from the real issue of poor management in general.  Having said that, it is clear that a more strategic approach is required to ensure that the implementation of GIS is more successful in future.

A Strategic Approach to GIS in Humanitarian Operations

In light of the above, what is the role of GIS in the humanitarian sector?  To begin with, we should be clear in our definition: there is no such thing as “humanitarian GIS”, no uniquely humanitarian aspect of GIS that can be identified; there is only GIS used to support humanitarian activities.  We obviously need to begin from a position that GIS is a tool to be used in this support capacity, and not the end goal of humanitarian work.  That said, what practical steps can we take to promote a more effective use of GIS in the humanitarian sector?

A sustained and integrated approach is a prerequisite for success; it is no use hoping that a series of unconnected projects, such as we have currently, will yield a coherent approach.  This is particularly true if we wish to ensure that GIS in an emergency response can successfully be carried over into reconstruction and recovery after a crisis.  Almost all successful GIS programmes have required a long-term commitment to creating and sustaining the skills, data and technology necessary to enjoy solid returns – with a corresponding investment.  Although basic cartography can be done relatively cheaply, spatial analysis takes more resources – particularly in terms of time – and there should be serious discussion about whether this is justifiable in the humanitarian sector.

With this question in mind, the following steps will lay the foundations for improving GIS implementation by humanitarian actors: 

a.   Prepare Systems

GIS has not proven itself in humanitarian response, but perhaps it can play a greater role in the ‘before’ and ‘after’, the preparedness and recovery phases of an emergency.  This would fit more easily into supporting capacity development for governments, an area where GIS has already been proven its effectiveness.  Humanitarian organisations would still be able to plug into those information systems when they deploy in a country, enabling emergency response to cohere more quickly and providing a framework for handover when the organisation withdraws. 

Recommendations

• Develop lighter GIS applications; quicker and easier to deploy in environments characterised by resource constraints, low levels of computer literacy and weak government structures.

• Adopt a modular approach and develop targeted applications that can support to individual government ministries such as health, education and security. 

b.   Prepare Data

Another requirement for supporting pre- and post-conflict countries is the problem of data scarcity in those countries.  Frequently governments in developing countries do not possess the resources to systematically collect and store data in digital formats.  A concrete project that would yield immense value is development of core datasets in countries prone to natural disasters or complex emergencies; a Global Data Preparedness Project, a clearing house for a wide range of datasets, specifically geared to supporting countries where data is poor or nonexistent. 

Recommendations

• Develop guidelines, templates and tools for governments and other organisations in developing countries to address data management issues.

• Encourage state and non-state institutions to create their own information management policies and strategies to ensure a more coherent, interoperable approach.

• Invest in data preparedness for countries at risk from natural or complex emergencies,- collating existing data and investing in new data (for instance, remote sensing imagery).

• Promote improved data collection during field operations, beginning with a focus on adding spatial data to existing data collection methodologies and initiatives.

• Lobby for more open sharing of that data as an obligation, lobbying against the use of proprietary information in emergency response, possibly integrating this obligation into existing standards (such as SPHERE) or agreements (such as donor contracts). 

c.    Market Honestly

A recent article on the use of GIS in malaria campaigns notes that “GIS users have not done a very good job of selling their applications to decision-makers… [they tend] to get caught up in technical jargon and not in the fact that a GIS can quickly make maps, and that maps are much easier to understand than tables.” (Sipe and Dale, 2003)  In order to sell GIS to the end users, we need to focus on the practical outputs that they will benefit from, and not on the technical details of data management. 

Recommendations

• Develop a body of case studies of GIS in the humanitarian sector that can be used to demonstrate practical value.

• Create opportunities for GIS practitioners to become involved in humanitarian work to create wider understanding of operational realities.

• Create realistic expectations based on existing capabilities, rather than create unrealistic expectations that will lead to clients becoming  

d.   Embrace Difference

Expecting humanitarian actors to rapidly, wholeheartedly and without reservation embrace GIS in their work is potentially damaging to both sides.  With this in mind, the GIS community needs to work harder to meet the needs of the humanitarian community, not to impose upon them business processes that may not be appropriate.  In my most recent work with WFP in the tsunami response, we had the greatest success when we focused on existing processes and developed tools to strengthen and improve them, rather than trying to force change (Currion 2005). 

Recommendations

• Identify existing business processes within the humanitarian community and develop GIS around those, engaging with end users on the ground (and not just at headquarters).

• Develop new ways of working in organisations with weak management structures, using a network-based service model that retains the creative autonomy of humanitarian organisations while loosely linking them together to enable co-operation and co-ordination.

(This is in contrast with a hierarchy-based systems model based on an integrated command-and-control approach.) 

e.   Accept Imperfection

Expecting all countries to achieve the level of detail in their geospatial data as the US is unrealistic, and raised expectations will ensure that both GIS practitioners and their clients come away disappointed.  In these situations, ‘perfect’ is definitely the enemy of ‘good enough’; and we need to become more comfortable working with data that may be incomplete or inaccurate while maintaining a clear focus on continuous improvement of that data.  As per point b, above, we need to invest in better data when we have the opportunity, but we must be realistic in what we expect and what we promise.

Recommendations 

• In humanitarian operations, focus on the key data that is important for decision-making – and ignore the rest.  Of course, this first means defining critical data for humanitarian operations so it can be focused on!

• Develop confidence levels for datasets (based on accuracy, completeness and timeliness of data, and reliability of source) that can be applied across organisations and countries, so that there is a common understanding of the limitations of the data. 

f.      Build Community

In general this has been one of the great successes of GIS, particularly through the efforts of ESRI – the creation of GIS communities in specific sectors.  However in the humanitarian community, GIS is still largely disjointed, with organisations failing to capitalise on the experience of others.  This can be seen in a number of areas, particularly in setting standards – for instance, the failure to create and promote a common humanitarian symbol set that can be used by all GIS practitioners in the field.  The pool of available specialists in humanitarian GIS has grown rapidly since the Brahimi Report, but it is still very limited and  there is frequently a lack of knowledge transfer between more and less experienced staff.

Recommendations 

• Create more tools to connect GIS practitioners in a community of practice that can reach across organisations.

• Approach the academic and private sectors to develop further and higher education courses that deal with humanitarian GIS (and development GIS, another area poorly served in terms of training).  These could be specialist courses, or components of broader courses in humanitarian work.

• Identify and agree GIS standards for the humanitarian sector, based on user need.  These could range from cartographic basics such as symbology and map formats, to more technical issues such as data models.[4]

g.   Educate End Users 

The failure of GIS (and to a lesser extent remote sensing) to meet expectations in the field is partially due to the lack of familiarity of decision-makers with the function of GIS and its potential to support humanitarian operations.  GIS professionals, supported by the wider GIS community, need to develop better awareness of their work, demonstrating the analytical power of GIS rather than being limited to visual representation.  The best way to achieve this is to attach such demonstrations to key humanitarian activities with clearly defined parameters – the success of mine action is an example in this respect.

Conclusion: what Brahimi didn’t do

The Brahimi Report correctly identified GIS as a technology that would be critical for the future management of peace (and humanitarian operations), but it could not lay out a framework for how GIS should be developed and implemented in those operations.  We still do not have a coherent approach to GIS in the humanitarian sector; I hope that this paper has provided some indication of the steps that need to be taken in order to build that approach.  I’m well aware that none of the recommendations laid out above are revolutionary; they are just the first steps on a long path towards successful application of an important new technology.  The challenge has been clearly set out for us; in the humanitarian sector, GIS might in future truly be the difference between life and death, and it is our responsibility to ensure that the technology is applied appropriately.

Bibliography

Benini, Aldo, et al., “Rapid Humanitarian Assessments – How Rational?  A Value-of-Information Study of Two Assessments in Iraq”, Report, 2004, Vietnam Veterans of America Foundation

Brahimi, Lakhdar, et al., Report of the Panel on United Nations Peace Operations, 21 August 2000, United Nations A/55/305–S/2000/809 

Currion, Paul, “Surviving ‘Droughts’ and ‘Floods’: Stretching the Metaphor for Humanitarian Information Management”, Symposium Paper, September 2003, Public Entity Risk Institute.

Currion, Paul, “Trailing the Tsunami: Information Systems in the World Food Programme”, Presentation, April 2005, Crisis Response Executive Advisory Team (CREATE) meeting. 

Darcy, James and Hoffman, Charles-Antoine, “According to need? Needs assessment and decision-making in the humanitarian sector”, HPG Report 15, September 2003, ODI.

Messick, Shawn, “Humanitarian Organizations Use Mapping To Save Lives”, June 2004, Geoworld Magazine, GeoTec Media. 

Schofield, Robin, “New technologies, new challenges: information management, coordination and agency independence”, Humanitarian Exchange 21, July 2002, ODI Humanitarian Practice Network.

Sipe, Neil G, and Dale, Pat, “Challenges in using geographic information systems (GIS) to understand and control malaria in Indonesia”, November 2003, Malaria Journal. 

Smith, David G, “Kosovo: Applying Geographic Information Systems in an International Humanitarian Crisis”, ESRI User Conference Paper, 2000