Dear Colleagues,

There has been a significant increase in the severity and frequency of crises and hazards worldwide, which are defined as “an interruption in the reproduction of economic, cultural, social and/or political life (Johnston, R.J. (2002). Dictionary of human geography. (4th ed.). Oxford, UK: Blackwell.)”. While extreme weather events are usually the causes of crisis, 2020 has become an expensive and deadly year due to another type of crisis, i.e., the COVID-19 pandemic. Whatever the cause of a crisis, though, technologies like cloud computing, location-based services, network science, web applications, and artificial intelligence (AI) are being used for crisis informatics to aid with crisis management and resilience efforts. 
Similarly, data obtained from both static and dynamic sources, such as remote sensing, unmanned aerial systems, and social media, enable the development of new approaches to charaterize and predict disaster situations at different locations and scales. Human dynamics data in both physical and virtual spaces are big, spatial, temporal, dynamic, and unstructured. The proliferation of data and interactive mapping technologies has also significantly enhanced access to and utility of spatial decision support systems, helping communities to better prepare for, respond to, and recover from crises and hazards. Understanding human dynamics can help to more efficiently deal with natural or man-made disasters. Significant advancements have also been made in developing statistical as well as data-driven models to integrate these heterogeneous data for real-time and off-time informatics. Because of the heterogeneous nature of these data in terms of data structure, content, data sources, and the spatial and temporal resolutions at which they are being obtained, these data suffer from uncertainties associated with positional accuracy, reliability, and completeness, thereby impacting the quality of the models being generated and their reproducibility.
Due to the spatiotemporal nature of a crisis, geospatial data sets and spatiotemporal models integrating various data sources are being developed. In addition to the uncertainties associated with the data, the developed models rarely account for scale, which influences not only the mechanisms used to aggregate and integrate data sets, but also the final outputs of the model. The end result is the development of models for crisis informatics that produce varying results and hence may not be useful in real-time decision making.
In this Special Issue in ISPRS International Journal of Geo-Information, we solicit articles that advance theories and methods and/or applications integrating spatial and temporal datasets at varying scales for crisis informatics. The articles should leverage existing theories and/or develop new theories of scaling and spatiotemporal modeling while taking advantage of big data theories and technologies to aid with crisis/disaster preparedness, mitigation, recovery, and resilience.

 Potential topics include (but are not limited to) the following:

1. Uncertainty in data and spatiotemporal models;
2. Data fusion methods and accuracies;
3. Data quality and impact on decision making;
4. Role of scale and reproducibility of models;
5. Human dynamics in crises and hazards;
6. Open knowledge network and convergence research;
7. Spatial decision support systsem for crisis management;
8. Geo-visualization and geo-computation techniques for real-time applications;
9. Models and analytics for crisis, human movement and behaviors, interaction of natural and built environments.

This Special Issue is scheduled to be published by 30 April 2021. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the Special Issue website.

Dr. Bandana Kar
Dr. Xinyue Ye
Dr. Zhenlong Li
Dr. Qunying Huang
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All papers will be peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. ISPRS International Journal of Geo-Information is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1400 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Title: Disaster image recognition by fusing multi-modality social networks data
Authors: Zhiqiang Zou1,2*, Hongyu Gan1 and Qunying Huang3
Affiliation: 1. College of Computer, Nanjing University of Posts and Telecommunications, Nanjing 210023, China; 2. Jiangsu Key Laboratory of Big Data Security and Intelligent Processing, Nanjing 210023, China 3. Department of Geography, University of Wisconsin–Madison, 550 N. Park St., Science Hall, Madison, 53706 United States * Correspondence: Zhiqiang Zou ([email protected])
Abstract: Abstract: The application of social media in disaster assessment has received widespread attention. When a disaster occurs, many users will post messages in a variety of formats, e.g., photo and text, on social media platforms. Useful information may be mined from these multi-modality data, which can provide decision support to disaster assessors. However, the multi-modality data collected from social media contain much irrelevant content which need to be filtered out. Existing work mostly combined the information of image classification and text keyword extraction in a very simple way; in other words, these methods treated the image and text features separately. This paper will seamlessly integrate pictures and text information based on the multi-modality fusion method to complete the recognition of useful disaster pictures. A CNN architecture was designed to extract the visual features from photos and a RNN framework was then used to extract the textual features. Next, a novel data fusion model will be developed to combine both visual and textual features to complete disaster image recognition. Using different disaster events (e.g., Earthquakes) as case studies, the proposed method can largely improve the identification of disaster pictures by comparing with the state-of-the-art work. Keywords: feature extraction、image recognition、text classification 、CNN、RNN、multi-modality fusion

Title: Integrating high-resolution imagery and points of interest (POIs) for urban scene recognition
Authors: Wenjing Denga, Qian Shia*, Qiang Qiub a Sun Yat-sen University, Guangzhou 510220, China b Institute of Computing Technology, Chinese Academy of Science, Beijing 100190, China *Corresponding: [email protected]
Affiliation: a Sun Yat-sen University, Guangzhou 510220, China b Institute of Computing Technology, Chinese Academy of Science, Beijing 100190, China *Corresponding: [email protected]
Abstract: Urban scenes play an important role in urban planning and management. As a result, high spatial resolution remote sensing imagery is commonly considered an ideal choice to extract information about urban land use. However, since the recognition of urban scenes require socio-economic knowledge, the information extracted from remotely sensed imagery cannot accurately reflect human activities and urban functions. However, social-related data (e.g., social network data) can compensate for this shortcoming and provide semantic labeling of some urban targets. Current methods mostly use either remote sensing data or social data to describe urban land use at a pixel or object level. The combination of features derived from high-resolution remote sensing imagery and socioeconomic features derived from crowdsourced data –such as points of interest (POIs) in block level– can help to understand better urban scenes. In this paper, we use a convolutional neural network (CNN) to extract spatial and spectral features in functional zones using both high-resolution images and parcels in open street maps. Then, kernel density estimation is adopted to describe socio-economic features using POIs. Finally, a random forest classifier is used to delineate both spectral characteristics and spatial object distributions, as well as semantic information in urban scenes. Our experimental results demonstrate that remote sensing data combined with POIs can significant help to better understand complicated urban scenes at a block level. Key words: urban scene recognition; high-resolution imagery; points of interest (POIs); block level.