EskisehirUHI_code_data: GEE and Python code and grid dataset for urban heat island risk assessment in Eskişehir, Turkey
Description
This dataset provides the code and derived data supporting the manuscript "A Grid-Based Urban Heat Island Potential Assessment Framework Integrating Building Height and Tree Cover: A Google Earth Engine Application in Eskişehir, Turkey" (Talha Aksoy). It contains: (1) Google Earth Engine scripts that generate the 250 m analysis grid from GHSL Built-H 2018 (building height), Street Tree Layer 2018 (tree canopy cover), and Landsat 8 Collection 2 Level-2 land surface temperature composites (2018–2020); (2) Python scripts reproducing all reported statistics — descriptive statistics, Pearson correlations, one-way ANOVA with Tukey HSD and eta-squared, Moran's I spatial autocorrelation, spatial error and lag regression models, block-permutation ANOVA, and the OpenStreetMap distance-to-industry analysis; (3) the derived 6,420-cell urban grid dataset (eskisehir_uhi_ghsl2018_250m.csv) with per-cell building height (m), tree cover fraction (0–1), land surface temperature (°C), and UHI risk class (1–5); and (4) a README with the data dictionary and run instructions. The variable "tree" denotes fractional canopy cover (share of cell area shaded by tree crowns), not tree height. All scripts run in Python with pandas, numpy, scipy, statsmodels, scikit-learn, and the PySAL ecosystem (libpysal, spreg, esda). Note: the Street Tree Layer 2018 input is a user asset in Google Earth Engine and is not redistributed here; the derived grid CSV is included so the statistical analyses are fully reproducible.