Processed data and analysis code for: Apparent loss of urban vegetation cooling depends on the spatial scale of measurement, Greater Dhaka, Bangladesh, 2000–2023

This deposit contains the processed datasets and analysis code supporting the manuscript "Apparent loss of urban vegetation cooling depends on the spatial scale of measurement: evidence from two decades of urbanization in Greater Dhaka, Bangladesh". The study evaluates whether the relationship between vegetation greenness (NDVI) and land surface temperature (LST), widely used to measure the cooling function of urban vegetation, is consistent across the spatial units on which it is computed. The same relationship was calculated from the same harmonised Landsat record over 2000-2023 using five spatial supports: 46 thana administrative units, and regular grids at 2 km, 1 km, 500 m and 250 m resolution. The deposit includes annual dry-season NDVI and LST composites summarised to each spatial unit, GHSL multi-epoch built-up fractions, ESA WorldCover land-cover composition, a single-sensor Landsat 8 extraction used for sensor sensitivity testing, a year-by-year Landsat scene inventory, and derived result tables. Analysis code reproduces the annual regressions, Mann-Kendall and Sen's slope trend tests, land-cover stratification, the temporally consistent dense-urban analysis, and the spatial-block bootstrap. All spatial data use EPSG:32646. LST is in degrees Celsius. Underlying satellite imagery is not redistributed; source products are openly available and documented in the README.