GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 2.2 GNGTS 2024 Methodology The methods adopted in this project allowed us to combine the available exposure data at different spaCal resoluCons (global, regional, naConal and sub-naConal) and produced under several past projects based on cumng-edge technologies. In parCcular, remote sensing data are very important in order to derive exposure datasets that allow covering large areas, but also allow to assess the locaCon and specific characterisCcs of selected sites. These data were combined with local-scale informaCon provided by local partners in each of the 5 Central Asia countries that enabled us to grasp the differences among the naConal contexts. The collected informaCon was then homogenized in order to provide regionally-consistent aggregated results for the enCre Central Asia region. Exposure datasets are, by definiCon, spaCal datasets, that is, digital maps where the exposure informaCon is associated to spaCal coordinates. In fact, the locaCon of exposed assets (e.g., where different building types are located within a country) is required in order to perform the risk assessment. In absence of informaCon on the asset locaCon (e.g., address, coordinates), a common method to infer the buildings’ or faciliCes’ locaCon is to distribute them spaCally based on proxies such as populaCon or land use maps. This operaCon, also called ‘spaCal disaggregaCon’, and other spaCal operaCons of the kind (e.g., merging of databases, intersecCon of different maps) were performed using the QGIS open-source program ( hbps://www.qgis.org/en/site/ ) . PopulaHon In this project, we developed a populaCon dataset at 100m resoluCon that includes specific demographic abributes (age, gender) for the whole Central Asia. This dataset was based on data from several data sources, used as a starCng point for the development of the exposure layers. In parCcular, the consorCum used the Facebook high-resoluCon dataset (hbps:// data.humdata.org/organizaCon/facebook) , which provides populaCon, gender and age informaCon at approximately 20 m in Central Asia. The Facebook populaCon data, retrieved for 2020, was disCnguished into three age classes: younger than 5 years old, older than 60 years old or the intermediate age class. The populaCon layers were assembled as follows: First, the total populaCon in the Facebook dataset was compared with the WorldPop dataset (hbps://www.worldpop.org/ ), that provides total populaCon (but no informaCon on age and gender fracCons) for 2020. The comparison was performed aper aggregaCng the Facebook data at the resoluCon of the WorldPop layer (100 m) and showed a good agreement. This operaCon was performed directly on the spaCal layers using the QGIS open-source program.