In recent years, deep learning (DL), a rebranding of neural networks (NNs), has risen to the top in numerous areas, namely computer vision (CV), speech recognition, and natural language processing. Whereas remote sensing (RS) possesses a number of unique challenges, primarily related to sensors and applications, inevitably RS draws from many of the same theories as CV, e.g., statistics, fusion, and machine learning, to name a few. This means that the RS community should not only be aware of advancements such as DL, but also be leading researchers in this area. Herein, we provide the most comprehensive survey of state-of-the-art RS DL research. We also review recent new developments in the DL field that can be used in DL for RS. Namely, we focus on theories, tools, and challenges for the RS community. Specifically, we focus on unsolved challenges and opportunities as they relate to (i) inadequate data sets, (ii) human-understandable solutions for modeling physical phenomena, (iii) big data, (iv) nontraditional heterogeneous data sources, (v) DL architectures and learning algorithms for spectral, spatial, and temporal data, (vi) transfer learning, (vii) an improved theoretical understanding of DL systems, (viii) high barriers to entry, and (ix) training and optimizing the DL.

Comprehensive survey of deep learning in remote sensing: theories, tools, and challenges for the community

With the ubiquity of advanced web technologies and location-sensing hand held devices, citizens regardless of their knowledge or expertise, are able to produce spatial information. This phenomenon is known as volunteered geographic information VGI. During the past decade VGI has been used as a data source supporting a wide range of services, such as environmental monitoring, events reporting, human movement analysis, disaster management, etc. However, these volunteer-contributed data also come with varying quality. Reasons for this are: data is produced by heterogeneous contributors, using various technologies and tools, having different level of details and precision, serving heterogeneous purposes, and a lack of gatekeepers. Crowd-sourcing, social, and geographic approaches have been proposed and later followed to develop appropriate methods to assess the quality measures and indicators of VGI. In this article, we review various quality measures and indicators for selected types of VGI and existing quality assessment methods. As an outcome, the article presents a classification of VGI with current methods utilized to assess the quality of selected types of VGI. Through these findings, we introduce data mining as an additional approach for quality handling in VGI.

/pdf/a-review-of-volunteered-geographic-information-quality-4hz1a1k5v7.pdf

A review of volunteered geographic information quality assessment methods

The varied effects of recent extreme weather events around the world exemplify the uneven impacts of climate change on populations, even within relatively small geographic regions. Differential human vulnerability to environmental hazards results from a range of social, economic, historical, and political factors, all of which operate at multiple scales. While adaptation to climate change has been the dominant focus of policy and research agendas, it is essential to ask as well why some communities and peoples are disproportionately exposed to and affected by climate threats. The cases and synthesis presented here are organized around four key themes (resource access, governance, culture, and knowledge), which we approach from four social science fields (cultural anthropology, archaeology, human geography, and sociology). Social scientific approaches to human vulnerability draw vital attention to the root causes of climate change threats and the reasons that people are forced to adapt to them. Because vulnerability is a multidimensional process rather than an unchanging state, a dynamic social approach to vulnerability is most likely to improve mitigation and adaptation planning efforts. This article is categorized under:Vulnerability and Adaptation to Climate Change > Values-Based Approach to Vulnerability and Adaptation.

/pdf/explaining-differential-vulnerability-to-climate-change-a-ochvqdgufq.pdf

Explaining differential vulnerability to climate change: A social science review

Cascading effects and cascading disasters are emerging fields of scientific research. The widespread diffusion of functional networks increases the complexity of interdependent systems and their vulnerability to large-scale disruptions. Although in recent years studies of interconnections and chain effects have improved significantly, cascading phenomena are often associated with the “toppling domino metaphor”, or with high-impact, low-probability events. This paper aimed to support a paradigm shift in the state of the art by proposing a new theoretical approach to cascading events in terms of their root causes and lack of predictability. By means of interdisciplinary theory building, we demonstrate how cascades reflect the ways in which panarchies collapse. We suggest that the vulnerability of critical infrastructure may orientate the progress of events in relation to society’s feedback loops, rather than merely being an effect of natural triggers. Our conclusions point to a paradigm shift in the preparedness phase that could include escalation points and social nodes, but that also reveals a brand new field of research for disaster scholars.

/pdf/critical-infrastructure-panarchies-and-the-vulnerability-motgv4nq21.pdf

Critical infrastructure, panarchies and the vulnerability paths of cascading disasters

The analysis of data from social media and social networking sites may be instrumental in achieving a better understanding of human-environment interactions and in shaping future conservation and environmental management. In this study, we systematically map the application of social media data in environmental research. The quantitative review of 169 studies reveals that most studies focus on the analysis of people’s behavior and perceptions of the environment, followed by environmental monitoring and applications in environmental planning and governance. The literature testifies to a very rapid growth in the field, with Twitter (52 studies) and Flickr (34 studies) being most frequently used as data sources. A growing number of studies combine data from multiple sites and jointly investigates multiple types of media. A broader, more qualitative review of the insights provided by the investigated studies suggests that while social media data offer unprecedented opportunities in terms of data volume, scale of analysis, and real-time monitoring, researchers are only starting to cope with the challenges of data’s heterogeneity and noise levels, potential biases, ethics of data acquisition and use, and uncertainty about future data availability. Critical areas for the development of the field include integration of different types of information in data mashups, development of quality assurance procedures and ethical codes, improved integration with existing methods, and assurance of long-term, free and easy-to-access provision of public social media data for future environmental researchers.

Passive crowdsourcing of social media in environmental research: A systematic map

. During and shortly after a disaster, data about the hazard and its consequences are scarce and not readily available. Information provided by eyewitnesses via social media is a valuable information source, which should be explored in a~more effective way. This research proposes a methodology that leverages social media content to support rapid inundation mapping, including inundation extent and water depth in the case of floods. The novelty of this approach is the utilization of quantitative data that are derived from photos from eyewitnesses extracted from social media posts and their integration with established data. Due to the rapid availability of these posts compared to traditional data sources such as remote sensing data, areas affected by a flood, for example, can be determined quickly. The challenge is to filter the large number of posts to a manageable amount of potentially useful inundation-related information, as well as to interpret and integrate the posts into mapping procedures in a timely manner. To support rapid inundation mapping we propose a methodology and develop "PostDistiller", a tool to filter geolocated posts from social media services which include links to photos. This spatial distributed contextualized in situ information is further explored manually. In an application case study during the June 2013 flood in central Europe we evaluate the utilization of this approach to infer spatial flood patterns and inundation depths in the city of Dresden.

/pdf/social-media-as-an-information-source-for-rapid-flood-3d8l7u0y3z.pdf

Social media as an information source for rapid flood inundation mapping

. At the end of October 2012, Hurricane Sandy moved from the Caribbean Sea into the Atlantic Ocean and entered the United States not far from New York. Along its track, Sandy caused more than 200 fatalities and severe losses in Jamaica, The Bahamas, Haiti, Cuba, and the US. This paper demonstrates the capability and potential for near-real-time analysis of catastrophes. It is shown that the impact of Sandy was driven by the superposition of different extremes (high wind speeds, storm surge, heavy precipitation) and by cascading effects. In particular the interaction between Sandy and an extra-tropical weather system created a huge storm that affected large areas in the US. It is examined how Sandy compares to historic hurricane events, both from a hydro-meteorological and impact perspective. The distribution of losses to different sectors of the economy is calculated with simple input-output models as well as government estimates. Direct economic losses are estimated about USD 4.2 billion in the Caribbean and between USD 78 and 97 billion in the US. Indirect economic losses from power outages is estimated in the order of USD 16.3 billion. Modelling sector-specific dependencies quantifies total business interruption losses between USD 10.8 and 15.5 billion. Thus, seven years after the record impact of Hurricane Katrina in 2005, Hurricane Sandy is the second costliest hurricane in the history of the United States.

/pdf/investigation-of-superstorm-sandy-2012-in-a-multi-7twkiaw2vm.pdf

Investigation of superstorm Sandy 2012 in a multi-disciplinary approach

The amount of information generated and provided by citizens via the World Wide Web is constantly growing. Citizens share information, thoughts, and experiences in blogs and contribute information to web-based content sharing platforms, collaboratively created data bases that are freely usable by everybody. Disaster management as one component of urban planning to decrease a society’s vulnerability can benefit from information provided by citizens. This chapter gives an overview of the application of information provided by citizens in disaster management. It points out the potential of using such information for the various phases of disaster management. Three main challenges, which affect the usefulness of information supplied by citizens, are presented in more detail: data collection, localization and quality assessment. For each of these challenges, various approaches to address them are discussed.

Volunteered Geographic Information for Disaster Management

For our research on chorematic diagrams and their automatised construction, we strive for a formal definition of chorematic diagrams. One approach for such a formal definition is to define generalisation constraints on their graphical, topological and geometric attributes. During our previous work, it became clear that the generalisation needs of chorematic diagrams are only partially covered by current cartographic constraint models. In this work in progress, we delineate the main schematisation tasks and show which can and which cannot be modelled with current constraint measures. We go on to make suggestions on how to extend existing constraint models correspondingly.

https://www.geog.uni-heidelberg.de/md/chemgeo/geog/lehrstuehle/gis/chorematic-schematisation.pdf

Joachim Fohringer

Papers

Social media as an information source for rapid flood inundation mapping

Investigation of superstorm Sandy 2012 in a multi-disciplinary approach

Volunteered Geographic Information for Disaster Management

Towards constraint formulation for chorematic schematisation tasks - work in progress

Super Typhoon Haiyan/Yolanda. CEDIM FDA Group Report, 13 Nov. 2013