A 1km-Resolution Dataset of Lower Respiratory Infection Mortality Probability Distribution in China (CHLRIMP) (2015–2022)

Overview This dataset provides a high-resolution (1km), annual spatial distribution of Lower Respiratory Infection (LRI) mortality probabilities across China spanning from 2015 to 2022. Anchored to the LandScan population distribution dataset as its fundamental spatial baseline, it is designed to support advanced spatial epidemiology, environmental health assessments, and precise public health resource allocation by mapping LRI vulnerabilities at a highly granular micro-spatial scale. Methodology and Data Sources The dataset was developed utilizing a novel spatial downscaling framework driven by Graph Neural Networks (GNN). The modeling process involves: Spatial Baseline Definition: The 1km-resolution LandScan global population distribution dataset was employed as the unified underlying grid. This ensures that the spatial allocation of mortality probabilities strictly aligns with real-world demographic densities and avoids mapping errors in unpopulated areas. Target Data Extraction: Baseline LRI mortality counts across multiple administrative levels in China were systematically collected through comprehensive literature retrieval and rigorous data extraction protocols. Covariate Integration: The mortality data was integrated with diverse socio-ecological factors. A key driver in this model is the SCOH index—a composite pollution and thermal exposure indicator synthesized from the Multiple Air Pollutants (MuAP) dataset and land surface temperature data. GNN-Based Downscaling: The multi-source data was first transformed into a complex graph structure. A customized Graph Neural Network (GNN), explicitly encoding administrative boundary containment and logical topological relationships, was then employed. This approach effectively downscaled the multi-level, aggregated LRI mortality counts into a continuous, 1km-resolution LRI mortality probability distribution. Technical Specifications Spatial Extent: China Temporal Coverage: 2015 – 2022 Temporal Resolution: Annual (1 Year) Spatial Resolution: 1km (approx. 0.008° × 0.01°) Coordinate Reference System (CRS): WGS84 (EPSG:4326) Data Format: GeoTIFF (.tif) Usability The dataset is provided in standard GeoTIFF format. It is highly accessible and ready for immediate, out-of-the-box use in mainstream Geographic Information Systems (e.g., QGIS, ArcGIS) and spatial data science environments (e.g., Python's Rasterio, R's terra) upon download, requiring no further spatial projection or format conversion.