Refine
Has Fulltext
- yes (2)
Is part of the Bibliography
- yes (2) (remove)
Year of publication
- 2023 (2) (remove)
Document Type
- Journal article (2)
Language
- English (2) (remove)
Keywords
- Deep learning (1)
- West African refugee crisis (1)
- multiscale encoder (1)
- news media reporting (1)
- sentinel (1)
- spatio-thematic coverage (1)
- surface model (1)
The fastest growing regional crisis is happening in West Africa today, with over 8 million people considered persons of concern. A culmination of identity politics, climate-driven disasters, and extreme poverty has led to this humanitarian crisis in the region and is exacerbated by a lack of political will and misplaced media attention. The current state of the art does not present sufficient investigations of the thematic and spatial coverage of news media of this crisis in this region. This paper studies the spatial coverage of this crisis as reported in the media, and the themes associated with those locations, based on a curated dataset. For the time frame 12 March to 15 September 2021, 2017 news articles related to the refugee crisis in West Africa were examined and manually coded based on (1) the geographical locations mentioned in each article; (2) the themes found in the articles in reference to a location (e.g., Relocation of people in Abuja). The dataset introduces a thematic dimension, as never achieved before, to the conflict-ridden areas in West Africa. A comparative analysis with UNHCR (United Nations High Commissioner for Refugees) data showed that 96.8% of refugee-related locations in West Africa were not covered by news during the considered time frame. Contrastingly, 80.4% of locations mentioned in the news do not appear in the UNHCR repository. Most news articles published during this time frame reported on Development aid or Political statements. Linear multiple regression analysis showed GDP per capita and political stability to be among the most influential determinants of news coverage.
In recent history, normalized digital surface models (nDSMs) have been constantly gaining importance as a means to solve large-scale geographic problems. High-resolution surface models are precious, as they can provide detailed information for a specific area. However, measurements with a high resolution are time consuming and costly. Only a few approaches exist to create high-resolution nDSMs for extensive areas. This article explores approaches to extract high-resolution nDSMs from low-resolution Sentinel-2 data, allowing us to derive large-scale models. We thereby utilize the advantages of Sentinel 2 being open access, having global coverage, and providing steady updates through a high repetition rate. Several deep learning models are trained to overcome the gap in producing high-resolution surface maps from low-resolution input data. With U-Net as a base architecture, we extend the capabilities of our model by integrating tailored multiscale encoders with differently sized kernels in the convolution as well as conformed self-attention inside the skip connection gates. Using pixelwise regression, our U-Net base models can achieve a mean height error of approximately 2 m. Moreover, through our enhancements to the model architecture, we reduce the model error by more than 7%.