Painting a fuller picture Innovative solutions Given the importance of accurate information on new displacements and the evolution of situations over time, we have begun to incorporate new approaches to our monitoring (see table 3.3). Taken together, our new “hybrid” approach combines event detection and data collection with the analysis of time-series data when it is available. For displacement associated with conflict, we have begun to identify and capture data about incidents of new displacement manually. In order to address the challenge of event detection on a global scale, we are also developing a new semiautomated process to identify potential displacements for human verification (see p.84). For disasters, we already capture several hundred incidents of new displacement a year – good but still not global. We tend to miss displacements associated with localised disasters that affect small numbers of people. The bigger gap, however, is in the systematic collection of timeseries data on people once they have become displaced. We have begun working with partners to collect and analyse more of this data so we can infer both the total number of people displaced by an event, and track the number and needs of displaced people as they evolve over time. One such method involves analysing satellite imagery to detect changes in human habitation in response to development projects such as dams, natural hazards and conflict. Based on the number of buildings destroyed or the extent of flooded land and population and settlement data, we will infer how many people may have been displaced, an approach already used by our sources to triangulate data obtained from the field. Another approach will transform our probabilistic risk model for displacement associated with disasters (described in part 1) into a real-time tool to support monitoring. When a hazard has been detected or is predicted to occur, we will simulate the amount of destruction and displacement expected to result. Using satellite imagery analysis and our displacement risk model as monitoring tools will help us extend the geographical coverage of our monitoring and address some of the factors responsible for the incomplete picture of displacement, notably language, reporting and selection biases. Table 3.3: Challenges and solutions for our “hybrid” monitoring approach Context Current situation Way forward Conflict Time-series data with a focus on end- Systematic event detection to inform of-year updates of stocks, but limited the collection of time-series data and event detection for new displacement more data points over time Disasters Event detection with a focus on the More comprehensive event detection occurrence of new displacement, but and systematic collection of data about limited time-series data how displacements evolve over time INSIDE THE GRID: Overcoming data shortfalls 83