 Thank you all for being here. Thank you MSF London for organizing this event. My name is Joanna Luce. I work for MSF in Brussels. I have a background in statistics and geographic information systems. I would like to mention my co-author Ansa Chaud who also works for MSF and is a student in International Humanitarian Law. In the following nine minutes I will briefly describe some of the challenges we faced with estimating populations under-seaged in specific areas around the capital city of Syria, Damascus, and go over a number of the solutions we developed for defining siege and providing rough population estimates for these areas. For those who are not yet aware of the situation, Syria has been has been at war for the past five years causing tremendous suffering and massive displacement of the population. The targeting of medical assistance and the besiegement of civilian populations have been characteristic of a conflict with little regard for rules. In this insecure and politicized context, direct humanitarian assistance to victims has been highly restricted limiting most MSF operations to support or remote programs in the country. MSF currently supports a number of medical facilities in besieged areas but doesn't have staff working directly in these areas. A particular challenge of this type of support model is the inability to rely on direct human experience and the problem of depending on various sources of information. For MSF this has led to some complications regarding the estimation of populations under siege. According to the United Nations there would be under 500,000 people under siege in Syria whereas other sources including stakeholders working with MSF in besieged areas count up to 3 million people. For organizational and advocacy purposes this alarming difference had to be verified but it had to be done rapidly with minimal data collection, minimal resources and limited contact with the field. To verify these numbers we set two distinct objectives. The first to define siege and the second to estimate populations. We narrowed our focus down to the most contested areas around the capital city of Damascus and we input a number of existing research methods to respond with flexibility to the needs of the project. I don't know why this is, sorry, there seems to be a problem. A siege is a method of war in which there is a military encirclement of an area. In Syria the siegement is also characterized by the restriction of humanitarian assistance to these areas and the containment of civilian populations. The concept definition is rather straightforward but actually defining areas under siege in Syria isn't so easy. We found that the UN's framework based on 17 conditions for classifying areas under siege was too restrictive for documenting realities on the ground. For example, according to the UN's framework, minor inflows of commercial goods in areas that were otherwise entirely besieged were automatically disqualified from the UN's official list. This resulted in awkward geographies of siege as illustrated by Heze, the green area on the screen, an area not listed as besieged by the UN but entirely surrounded by areas besieged. This type of problem wasn't unique to the UN. A common challenge we faced when defining siege was dealing with fuzzy or indeterminate geographies as illustrated by the differently colored polygons on the left of the screen. We dealt with this problem by calculating spatial averages and focusing in on core areas of besiegement where there was a greater concentration of the population as indicated by the presence of settlements he or she didn't block. In our approach to defining besieged areas we decided to simply adjust the more problematic aspects of the UN's framework and collect our own data. We sent a survey structured around indicators of siege that were to be graded on a liquid scale by respondents in besieged areas over a period of one month. We gave respondents the choice of completing the survey online or offline through mobile devices or on paper as they felt safest and most comfortable. We also included maps in the surveys so that respondents could draw in additional information if they wished to do so. With this data collected it was possible to calculate a siege score from 0 to 1 and from this siege score it was possible to identify broad zones around the capital city of Damascus that had various severities of siege and distinct dynamics. In our final analysis we found that the vast majority of areas where we were providing medical support were in fact severely besieged with high levels of restriction and destruction. Once the areas under siege had been identified we compared population estimates of different datasets for similarly sized communities. As illustrated in the box spot on the screen the average estimate for similarly sized communities was actually similar across all datasets at about 20,000 people per community. Only two communities seemed to have disproportionate population estimates in the dataset of other sources which weighed heavily on the overall estimate of 3 million people under siege. For one specific example the besieged population was estimated at around 700,000 people for an area that contained about 50,000 people before the crisis. Such an increase in this area would imply record-breaking densities of over 100,000 people per square kilometer. To verify the plausibility of this particular estimate we used a method for rapid population estimates. First we defined the geographic area as I discussed earlier in the presentation. In this specific case the average surface area was 4.6 square kilometers. Second we counted the number of structures in the area by using digitalized maps based on satellite imagery provided by missing maps and MSF GIS units in Geneva. And we paid particular attention to the presence of camp-like or slum-like structures in addition to multi-story complex and simple buildings. Third we determined the number of households per structure based on surveys completed by focal points in these areas and photo documentation when possible. Note on the screen that we broke down the building structures in four different types according to the typical number of households. Fourth we determined the number of people per household by asking doctors to question patients on the number of people they lived with and the type of structures they lived in. And finally we calculated the rough population estimates and mapped the densities for the specific areas based on the assumption of maximum occupancy rates. This exercise allowed us to discard the outstanding population estimate of 700,000 people and replace it by more plausible although rough estimates of 250,000 people. Overall this project allowed us to more accurately define siege and estimate populations under siege. We were able to determine that there was likely somewhere between 1 and 1.5 million people under siege in Syria. In the midst of a highly politicized debate we were able to make clear arguments and influence stakeholders as illustrated by the press release on the screen. We were able to demonstrate the relevance of using mixed and flexible methodologies for supporting operations in complex and insecure environments. Unfortunately we were unable to review population estimates for all besieged areas in Syria and much of the data collected was limited and not based on random sampling. Over the coming months we will continue to collect small data by providing training to local organizations so they can take the lead in collecting demographic data in besieged areas. And we will start a new project that looks into bigger data by turning towards social media monitoring to support our work in Syria. Thank you very much for your attention.