Global spatially-distributed sectoral GDP map for disaster risk analysis

Abstract

Global risk assessments of economic losses by natural disasters while considering various land uses is essential. However, sector-specific, high-resolution pixel-level economic data are not yet available globally to assess exposure to local disasters such as floods. In this study, we employed new land-use data to construct a global, spatially distributed map of sector-specific gross domestic product (GDP). We developed three global GDP maps, SectGDP30, in 2010, 2015, and 2020 for the service, industry, and agriculture sector with 30 arcsec resolution. The map (SectGDP30) demonstrates strong consistency (R2 > 0.9) with actual sub-national statistical data, exhibiting superior alignment compared to conventional GDP maps (PB-method) reliant solely on gridded population information. The methodology refined GDP distribution for specific sectors. Industry GDP was more accurately mapped using non-residential land areas as a proxy, effectively capturing its localized concentrations. Agriculture GDP's accuracy improved by incorporating cropland data and a distance-based distribution assumption from population agglomeration. Application of this dataset in estimating flood-induced business interruption (BI) losses confirmed the map's capacity to represent inter-sectoral differences in estimated losses, reflecting varied hazard spatial distributions. This underscores the importance of considering sector-specific spatial patterns for accurate disaster damage assessment. These maps serve as a foundational tool for estimating detailed, sector-classified economic losses, enabling precise calculation of sector-specific impacts from diverse natural disasters worldwide. These global sectoral GDP maps (SectGDP30) are available at 10.5281/zenodo.15774017 (Shoji et al., 2025).