Boundary-layer trapping, lagged accumulation, and conditional morphological effects on PM2.5 and NO2 across 5.5 heating seasons in Bishkek, Kyrgyzstan: a generalised additive model

Code and processed data accompanying the article "Boundary-layer trapping, lagged accumulation, and conditional morphological effects on PM2.5 and NO2 across 5.5 heating seasons in Bishkek, Kyrgyzstan: a generalised additive model" (Atmospheric Pollution Research). This repository contains the R analysis pipeline, Google Earth Engine extraction scripts, the processed analytical panels for PM2.5 (947,690 pixel-days) and NO2 (676,750 pixel-days), the derived daily inversion index, the nine-class urban morphology zones, spatial-robustness diagnostics, and the figure scripts needed to reproduce the generalised additive model (GAM) analysis of heating-season air pollution over Bishkek (October 2017 – March 2023). Raw third-party inputs (ERA5, Sentinel-5P TROPOMI, GHAP V6 PM2.5, MODIS, OpenStreetMap) are not redistributed; sources and access are documented in DATA_PROVENANCE.md. Meteorological station data are openly available from the CAIAG Sensor Data Storage System and NOAA.