JN-LUT: station-area land use, metro ridership and co-optimisation data and code for Jinan, Qingdao and Shanghai

Data and code accompanying the manuscript Capacity-constrained low-carbon co-planning of land-use intensity and travel structure in Jinan, China, with a land-use-conditioned spatio-temporal network and differentiable-surrogate multi-objective optimisation (Lu L. and Yu T.). The record contains two multi-volume 7-Zip archives (25 MB volumes; put all volumes of an archive in one folder and open the .001 volume with 7-Zip, or run 7z x on it): JN-LUT-package.7z (.001–.030): station-area features of Shanghai (302 stations), Jinan and Qingdao; metro adjacency graphs; EULUC-China 2.0 parcels clipped to each city; the Shanghai MetroFlow 10-min entry/exit tensor; 516 city-level transport statistics with sources; all system parameters with sources; zero-history cross-validation, short-term forecasting, cross-city transfer, elasticity and optimisation results including every run, final populations and convergence histories; trained model checkpoints; the complete code (feature construction, LUC-STNet models, mode-structure/emission system, DS-NSGA-II/NSGA-III/SMS-EMOA optimisation, figures, manuscript generators); all manuscript figures (PNG 600 dpi and PDF). JN-LUT-raw-inputs.7z (.001–.042): the raw public input data read by the code (Shanghai MetroFlow, CNBH-10m building-height tiles, 100 m population grid of China 2020, CPTOND-2025 metro and bus networks of the three cities, CMAB Jinan buildings, city statistical communiqués/yearbooks and other public statistics with source URLs), redistributed under their original licences. The national EULUC-China 2.0 geodatabase is openly available as Zenodo record 16794007 and is not duplicated; the clipped parcels used by the analysis are in the derived package. See README.md for the file list, the original data sources (with DOIs and licences) and step-by-step reproduction commands. Software: Python 3.12, PyTorch 2.11, pymoo 0.6.2, geopandas, rasterio, scikit-learn, LightGBM. Licence: derived data and results CC BY 4.0 (please also cite the original data sources); code MIT.