Abstract: Decision-makers in humanitarian crisis need information to guide them in making critical decisions. Finding information in such environments is a challenging task. Therefore, decision-makers rely on domain experts who possess experience and knowledge from previous humanitarian crises to provide them with the information they need. In this paper, we explore the ability of the existing computing technologies to augment the capabilities of those experts and help decision-makers to make faster and better decisions. Among many computing technologies we have today, word embedding and the semantic web are able to support such augmentation of the domain expert. In this paper, we train a word embedding model using word2vec, transform words and terms from news archive to entities in domain ontology, annotate those entities with their equivalent concepts from upper ontologies, and reason about them using semantic similarity and semantic matching, to represent and retrieve knowledge, and answer questions of interest to decision-makers in humanitarian crises. The approach was evaluated by comparing the use of word embeddings with and without semantic classification for the retrieval of information about the current humanitarian crisis in Syria.

Abstract: In the Wellington Region, many lifelines are at risk, because they are in vulnerable narrow corridors close to active faults. In an earthquake, it is expected that these lifelines will be significantly damaged and unusable for extended periods of time. Because of this risk, many studies have been conducted to investigate the resulting downtimes. These studies, despite their usefulness, do not incorporate or make significant assumptions about localised factors. This paper summarises a thesis that aimed to improve the current predictive models, by including these local, and contextual influences. Multiple stakeholders who manage and repair the lifelines were interviewed to identify these factors which were then included into one of the current predictive models, and the influence on repair times was recorded. It was discovered that localised impacts such as staff logistics, land sliding, the land gradient, interdependency, and access doubled previous predicted repair times.

Abstract: There are growing concerns that future Atlantic hurricane seasons will be severe and unpredictable due to underlying factors such as climate change. The 2017 season may offer a range of lessons, especially to small island developing states (SIDS), who are looking to build community resilience and heighten community engagement to cope with disaster. While many SIDS utilise a range of media and technology for these purposes, there has been a recent uptake in the use of social media, which may have further potential to support their goals. This paper scopes the use and users of social media in the case of Antigua and Barbuda during the 2017 Atlantic hurricane season. Through a series of qualitative interviews it explains the role that social media currently has, and concludes with suggestions for its improvement in future seasons that are contextualized over the disaster lifecycle phases.

Abstract: Successful disaster relief operations call for responsive and flexible movements of goods in order to make the relief items available when and where they are needed by the affected communities. The objective of this explorative research is to investigate the applicability of the Physical Internet to emergency relief operations. The Physical Internet is a groundbreaking logistical system in which standardised, modular packages are automatically routed from origin to destination through a hyperconnected network of logistics providers and facilities. Although the concept is receiving growing attention from academics, practitioners and policy makers, research in the disaster management context is virtually non-existent. Based on a review of the relevant academic literature and publicly available information from emergency responding agencies and the media, we firstly use the Kaikoura earthquake that occurred in the South Island of New Zealand on 14th November 2016 to identify six key requirements that support the efficient movement of relief items in the aftermath of a disaster. We then identify six characteristics of the Physical Internet and explore how these characteristics can support the requirements of emergency response operations. We conclude that the Physical Internet principles have the potential to enhance the speed, flexibility and reliability of emergency responses. In other words, a fully integrated and collaborative logistics system in which relief items and information move seamlessly across a web of interconnected transport modes and operators can increase the efficient deployment of urgently needed relief items. This study extends the Physical Internet principles to emergency relief operations and identifies new ways of improving and optimising the logistics of emergency responses. In doing so, this research aims to stimulate debate within the disaster relief sector.

Abstract: The Spatial Futures Forum, held in Sydney in September 2017, highlighted issues that governments, emergency management organisations, academics and spatial systems research and development groups should carefully consider as they work towards a future that provides 'inter-connectedness with inclusiveness' for individuals and the societies in which they live. This is especially important when we consider the implications for disaster management when situational awareness and community resilience will be reliant on the: successful connection and integration of the 'islands' of spatial information generated by and stored in current systems; development of a real-time 'data on demand' approach to spatial systems; and the development and careful curation of an individual's 'virtual identity' from an ethical, legal, property ownership and risk perspective.

Abstract: We estimate the causal effect of the Indian Ocean tsunami in Sri Lanka on household income and consumption eight years after the event, using a quasi-experimental method. A strong association between area-wide tsunami disaster shock and increases in household income and consumption in the long-term emerged from our empirical investigation. Deviating from the common observation on short-term impacts, these results are suggestive of an optimistic potential for some long-lasting potentially successful recovery scenarios. Still, Sri Lanka received a very large amount of external transfers post-tsunami, much larger than is typical for disaster events and one which may not be replicable in other cases. Our findings suggest a more nuanced picture with respect to household consumption impacts. We observe a reduction of food consumption and only find an increase in non-food consumption. The increase in non-food consumption is much smaller than the observed increase in income. We also find that households in high-income regions experienced much better recovery from the disaster. Keywords Sri Lanka, Tsunami, disaster, household survey, long-run impact

Abstract: This paper outlines practitioner work-in-progress in Canterbury, New Zealand, to develop a regional approach for geographical information systems (GIS) for emergency management. This is based upon recent events in Canterbury including earthquakes, floods, and fire; as well as New Zealand-wide work that is being done under the NZ GIS4EM banner. It introduces our approach, discusses a mind map that is being used to track desired data sets, plans to develop applications to support response functions in emergency operations centres, and the goal of using the common data sets as the basis of a common operating picture for Canterbury. Risks and issues associated with this work are highlighted, and then the draft strategy is introduced with desired outcomes and principles to achieve this goal. While initial work is primarily focused on GIS, the expectation is that the approach will be expanded to take a broader information management perspective in future.

Abstract: Many machine learning and natural language processing techniques, including supervised and domain adaptation algorithms, have been proposed and studied in the context of filtering crisis tweets. However, applying these approaches in real-time is still challenging because of time-critical requirements of emergency response operations and also diversities and unique characteristics of emergency events. In this paper, we explore the idea of building “generalized” classifiers for filtering crisis tweets that can be pre-trained, and are thus ready to use in real-time, while generalizing well on future disasters/crises data. We propose to achieve this using simple feature based adaptation with tweet representations based on word embeddings and also sentence-level embeddings, representations which do not rely on unlabeled data to achieve domain adaptations and can be easily implemented. Given that there are different types of word/sentence embeddings that are widely used, we propose to compare them to get a general idea about which type works better with crisis tweets classification tasks. Our experimental results show that GloVe embeddings in general work better with the datasets used in our evaluation, and that the supervised algorithms used in our experiments benefit from GloVe embeddings trained specifically on crisis data. Furthermore, our experimental results show that following GloVe, the sentence embeddings have great potential in crisis tweet tasks.

Abstract: Is your organisation ready to cope with underground infrastructure condition assessment data collected after an earthquake? Drawing on lessons from the 2010-2011 Canterbury and 2016 Kaikoura earthquakes, we provide guidance on how to make small differences in how your organisation currently collects and stores the necessary condition data to prepare for emergencies, especially for small- and medium-size councils without sophisticated asset management systems. Key questions to address include: Are you receiving condition assessment data in electronic format? Are your contractors providing XY coordinates when repairs are undertaken, or when providing photographs as part of visual assessment? Do you have an asset management system able to prioritise critically damaged underground infrastructure? Do you have easy access to your current network condition for insurance purposes? Simple business-as-usual improvements will provide enhanced preparedness and resilience capability in the event of an earthquake. In addition, we provide a framework for future data collection processes.

Abstract: The standard framework for undertaking a risk assessment of a natural hazard involves analyzing the interaction of three components: Hazard data (in the form of maps), the elements exposed to the hazard (exposure), and measures of these elements' vulnerability (understood as the susceptibility to harm or damage). In New Zealand, national flood risk remains unquantified due to the absence of national flood inundation hazard map coverage. In this paper, we develop a methodology that aims to fill this gap by estimating instead the likelihood of a flood insurance claim for a stock of residential buildings. We estimate a non-linear limited-dependent variable model and using a set of fragility functions (also known as damage curves), we calculate the expected monetary losses under plausible flood depth scenarios. The outcome of this research could inform insurers of their potential liabilities and threats to their financial sustainability in the face of flood and storms.

Abstract: With the Hurunui (HDC), Mackenzie (MDC) and Kaikoura (KDC) District Councils sharing their Geographic Information Systems (GIS) and Information Technology (IT) resources since 2017 it was decided to work on a GIS strategy for Emergency Management (EM) that would be applicable for all three councils as in the past geospatial skills and tools did not get equally utilised at all three councils during emergency and training events. ArcGIS Online (AGOL) was chosen as a common platform for a fully cloud based approach to the new Emergency Management Applications. The core modules of these applications are Story maps, WebApp Builder, Survey123, Operations Dashboard, Workforce and AppStudio. The development of these applications is a work-in-progress situation which is driven by the constant conversation and testing between the GIS person and the Emergency Management (EM) officers and a work flow is being developed to integrate these applications into the existing Emergency Operations Centre (EOC) picture. We believe once finalised this set of applications will add great functionality to New Zealand's Coordinated Incident Management System (CIMS) by providing interactive and cloud based visual geospatial information, situational awareness, forecasting, task management and task tracking.

Abstract: Crisis informatics researchers have taken great interest in methods to identify information relevant to crisis events posted by digital bystanders on social media. This work codifies the information needs of emergency dispatchers and first responders as a method to identify actionable information on social media. Through a design workshop with public safety professionals at a Public-Safety Answering Point (PSAP) in the United States, we develop a set of information requirements that must be satisfied to dispatch first responders and meet their immediate situational awareness needs. We then present a manual coding scheme to identify information satisfying these requirements in social media posts and apply this scheme to fictitious tweets professionals propose as actionable information to better assess ways that this information may be communicated. Finally, we propose automated methods from previous literature in the field that can be used to implement these methods in the future.

Abstract: Inter-organisational collaboration is often cited as a factor in effective emergency response to mitigate risks and impacts of extreme events, such as natural disasters (James, 2011; Power, 2017). However, collaboration during times of emergencies is difficult as diverse roles and authorities of other organisations (Kapucu, 2008), differences in organisational cultures and systems in place (Bharosa, Lee, & Janssen, 2010; Mendonça, Jefferson, & Harrald, 2007) and difficulty in coordination and information sharing (Bharosa et al., 2010; Huxham, 1996) can hinder collaborative development. Moreover, previous research have not paid much attention to the enablers and barriers of such collaborations in disaster settings. This research aims to investigate how collaborative systems deal with uncertainty during disaster aftermath in Malaysia, including both barriers and enablers which influence inter-organisational collaboration in disaster environments. The study employs a qualitative methodology using a multiple-case study design. Malaysia National Disaster Management Agency (NADMA) was the focal organisation for this research, and units of analysis were two major floods that occurred in Malaysia in 2014 and 2017. Thirty semi-structured interviews were conducted with strategic and operational decision makers (from 11 different organisations) and disaster aid recipients. Qualitative research tools such as coding, memoing, constant comparison and theoretical saturation were applied to analyse the data (Charmaz, 2008). The initial findings to date detect salient enablers: proficiency of stakeholders in managing relationships, competent leaderships, and early preparedness; barriers include ambiguous organisational roles, weak legitimate framework and different perceptions on trust. This research has both academic and practical implications. The findings uncover how enablers and barriers influence collaboration between public agencies and its partners in emergencies. This research recommends solutions to collaboration barriers, and discusses how multiple partners can cooperate to resolve problems too complex for any one single organisation to manage on their own. Khiam LEE