Climate nonstationarity amplifies inequality of urban flood adaptation debt

Directory Structure of Zenodo_Upload_Nature===========================================Code and Data for "Climate nonstationarity amplifies inequality of urban flood adaptation debt" Zenodo_Upload_Nature/│├── code/ # Scripts used to generate all figures and tables│ ││ ├── fig1/ # Main Figure 1: spatial distribution & macro-geographical trends of the RTF│ │ ├── Fig1_main_RFT_map.py # Fig. 1a — baseline RTF spatial map with count marginals│ │ ├── Fig1b_RFT_MAP_boxplot.py # Fig. 1b — RTF vs historical MAP boxplot (North/South)│ │ ├── Fig1c_d_twin_axis.py # Fig. 1c-d — longitudinal & latitudinal gradients (twin-axis)│ │ ├── 01_extract_data.py # extract raw annual drainage/investment records (2011-2024)│ │ ├── 02_compute_mean.py # compute multi-year mean pipe density & investment│ │ ├── trend_scatter.py # trend scatter & North-South comparison│ │ ├── fig1_spatial_map.py # spatial map variants│ │ ├── fig1_hbar.py # economic-circle stacked bar (also ED Fig. 1)│ │ ├── fig1_city_category_marginal.py # city-category marginal maps│ │ ├── investment_density_marginal.py # per-area investment vs pipe density (also ED Fig. 5b)│ │ ├── _quant_analysis.py # quantitative analysis helpers│ │ └── case_hangzhou.py # Hangzhou case study│ ││ ├── fig2/ # Main Figure 3: attribution of flood escalation│ │ ├── rainfall_trend_exposure_map.py # Fig. 3a — bivariate R95p-trend × population-exposure map│ │ ├── Fig3b_resilience_deficit.py # Fig. 3b — resilience score & infrastructure deficit map│ │ ├── Fig3c_SHAP_attribution.py # Fig. 3c — SHAP/Gini/Ridge integrated attribution│ │ ├── Fig2_FullPhysics_SHAP.py # full-physics SHAP panels│ │ ├── Fig2_SHAP_Attribution.py # SHAP attribution variants│ │ ├── Fig2_Master_Combined.py # combined attribution summary│ │ ├── Fig2_forest_plot.py # forest plot│ │ ├── Fig2d_Attribution_Panel.py # attribution panel (plan B)│ │ ├── Step_Resample_Precip_Green.py # resample precipitation & green-space to 0.1-degree grid│ │ └── generate_report.py # auto-generated attribution report│ ││ ├── fig3/ # Main Figure 2: spatiotemporal divergence of flood frequency│ │ ├── Fig2_flood_divergence.py # Fig. 2a-c — national map, time series, latitudinal profile│ │ ├── Fig3_main.py # Fig. 2a — flood-frequency trend map (CLEAN_no_tamper version)│ │ ├── compute_grid_impervious.py # compute 0.1-degree impervious-surface grids (CLCD)│ │ ├── preprocess_era5_impervious.py # preprocess ERA5 + impervious data│ │ ├── match_city.py # match city points to polygons│ │ ├── fig3_abcd_quantitative.py # quantitative panels a-d│ │ └── gen_report.py # auto-generated statistics report│ ││ ├── fig4/ # Main Figure 4: future adaptation debts under CMIP6│ │ ├── Fig4_main.py # Fig. 4a-g + Extended Data Fig. 7 (investment decomposition)│ │ └── Supplementary_Figures.py # Fig. 4 supplementary panels│ ││ ├── supplementary/ # Supplementary Figure S1-S14 scripts│ │ ├── FigS2_LLM_extraction.py # Fig. S2 — LLM extraction worked example│ │ ├── FigS10_dual_forcing.py # Fig. S10 — dual-forcing (antecedent window n, decay k)│ │ ├── FigS10_run_optimization.py # Fig. S10 — parameter optimization│ │ ├── FigS11_xgboost_performance.py # Fig. S11 — XGBoost performance│ │ ├── FigS12_SHAP_decoupling.py # Fig. S12 — SHAP causal decoupling│ │ ├── FigS13_data_completeness.py # Fig. S13 — crowdsourced data completeness│ │ ├── FigS13_reconstruct_flood_counts.py # Fig. S13 — reconstructed flood counts│ │ ├── FigS13_reconstruct_flood_plot.py # Fig. S13 — reconstruction plot│ │ ├── FigS14_sensitivity_analysis.py # Fig. S14 — sensitivity / robustness│ │ └── compute_etccdi_indices.py # compute 11 ETCCDI extreme-precipitation indices (upstream)│ ││ └── extended_data/ # Extended Data Fig. 1-8 and Table 1-2 scripts│ ├── Extended_Data_Fig_1_economic_circles.py # ED Fig. 1 — RTF classes across economic circles│ ├── Extended_Data_Fig_2_latitudinal_gradient.py # ED Fig. 2 — latitudinal gradient diagnostic│ ├── Extended_Data_Fig_3_bivariate_indices.py # ED Fig. 3a-d — extreme-precipitation bivariate maps│ ├── Extended_Data_Fig_4_correlation_matrix.py # ED Fig. 4 — pairwise correlation / scatter matrix│ ├── Extended_Data_Fig_5a_risk_profile.py # ED Fig. 5a — risk classes across agglomerations│ ├── Extended_Data_Fig_5b_pipe_density_investment.py # ED Fig. 5b — pipe density vs investment KDE│ ├── Extended_Data_Fig_6_EOF_rainbelt.py # ED Fig. 6a — EOF1 spatial pattern│ ├── Extended_Data_Fig_6_Hovmoller.py # ED Fig. 6b — Hovmoller diagram│ ├── Extended_Data_Fig_6_EOF_PC1.py # ED Fig. 6c — PC1 time series│ ├── Extended_Data_Fig_8ab_bivariate_maps.py # ED Fig. 8a-b — CMIP6 bivariate maps│ ├── Extended_Data_Fig_8_cde_trend_analysis.py # ED Fig. 8c-e — boxplot / CDF / latitudinal gradient│ └── Extended_Data_Table_2_cmip6_trends.py # ED Table 2 — projected R95p trends (CMIP6)│├── figures/ # Generated figures│ ├── main/ # Main-text figures (Fig. 1-4)│ │ ├── Fig1a_RTF_map.jpg # Fig. 1a│ │ ├── Fig1b_RFT_MAP_boxplot.jpg # Fig. 1b│ │ ├── Fig1c_d_gradients.jpg # Fig. 1c-d│ │ ├── Fig2a_freq_trend_map.png # Fig. 2a│ │ ├── Fig2b_freq_timeseries.png # Fig. 2b│ │ ├── Fig2c_latitudinal.png # Fig. 2c│ │ ├── Fig3a_bivariate_R95p.jpg # Fig. 3a│ │ ├── Fig3b_resilience_deficit.jpg # Fig. 3b│ │ ├── Fig3c_SHAP_attribution.jpg # Fig. 3c│ │ ├── Fig4a_b_maps.jpg # Fig. 4a-b│ │ ├── Fig4c_f_analysis.jpg # Fig. 4c-f│ │ └── Fig4g_latitude.jpg # Fig. 4g│ ││ ├── extended_data/ # Extended Data figures (Extended Data Fig. 1-8)│ │ ├── Extended_Data_Fig_1_economic_circles.jpg│ │ ├── Extended_Data_Fig_2_latitudinal_gradient.pdf│ │ ├── Extended_Data_Fig_3a-d_*.jpg # R99p / Rx1day / Rx3day / Rx7day│ │ ├── Extended_Data_Fig_4_correlation_matrix.jpg│ │ ├── Extended_Data_Fig_5a_risk_profile.jpg│ │ ├── Extended_Data_Fig_5b_pipe_density_investment.jpg│ │ ├── Extended_Data_Fig_6_EOF_rainbelt.png│ │ ├── Extended_Data_Fig_7_decomposition.jpg│ │ └── Extended_Data_Fig_8a/8b/8cde_*.jpg # bivariate maps + trend analysis│ ││ └── supplementary/ # Supplementary figures (Fig. S1-S14)│ ├── FigS1_*.pptx # Fig. S1 — LLM pipeline workflow (editable)│ ├── FigS2_LLM_extraction.png # Fig. S2 — LLM extraction example│ ├── FigS9_*.pptx # Fig. S9 — RTF calculation flowchart (editable)│ ├── FigS10_dual_forcing.jpg # Fig. S10│ ├── FigS11_xgboost_performance.jpg # Fig. S11│ ├── FigS12_SHAP_decoupling.jpg # Fig. S12│ ├── FigS13_data_completeness.jpg # Fig. S13│ └── FigS14_sensitivity.jpg # Fig. S14│├── data/ # Processed data (CSV / JSON)│ ├── fig1/ # RTF thresholds, city baseline (1980-1999), annual raw data (2020-2024)│ │ ├── Urban_CRTI_Thresholds_30th.csv # city RTF (mm)│ │ ├── City_Baseline_Data_1980_1999_Final.csv # RFT_mm, historical MAP, coordinates│ │ ├── Spatial_Map_Final_With_Coords.csv # city pipe density / investment + coordinates│ │ └── 20xx_yearly_raw_data.csv # annual raw records│ ├── fig2/ # Attribution & precipitation-index data│ │ ├── Attribution_Feature_Matrix.csv # driver feature matrix│ │ ├── China_Rainfall_All_Indices_1980_2025.csv # 11 ETCCDI indices (National/North/South)│ │ ├── Extended_Data_Table_1_indices_trends.csv # ED Table 1 (decadal trends)│ │ └── Table_S4_Trends_All_Lon.csv # all-longitude trend table│ ├── fig3/ # Flood-frequency & impervious-surface data│ │ ├── Final_City_Flood_Trends_1980_2025_Final.csv # city flood-frequency trends│ │ ├── Urban_Flood_Metrics_Yearly_1980_2025.csv # yearly flood metrics│ │ ├── Threshold_Analysis.csv / Threshold_Method_Comparison*.csv│ │ └── CityPoint_to_Polygon_Mapping.csv / Matched_Points_to_Polygons.csv│ ├── fig4/ # Adaptation-investment data│ │ ├── Plot_Data_Output.csv # adaptation investment plotting data│ │ ├── Future_Economic_Adaptation_Path.csv│ │ ├── City_Waterlogging_Trends_2015_2100.csv│ │ └── Summary_Regional_Changes_All_Models.csv│ ├── future_projection/ # CMIP6 model time series (14 GCMs + ensemble mean)│ │ ├── Data_TimeSeries_<model>.csv # regional R95p anomaly time series (SSP2-4.5 / SSP5-8.5)│ │ └── Summary_Regional_Changes_All_Models.csv│ └── supplementary/ # Station-level calibration data (SSVM-ANF)│ ├── National_Stations_Valid_Results_v2.csv│ ├── National_Stations_Optimal_Antecedent_Params_v2.csv│ ├── National_Stations_DecisionBoundary_Params.csv│ ├── Statistical_Summary_v2.json│ └── Paper_Key_Data.json│├── gis/ # Self-generated vector data (shapefiles, compressed as .zip)│ ├── City_Polygons_Aggregated_RFT.zip│ ├── Urban_Flood_Infrastructure.zip│ ├── Urban_Resilience_Elements.zip│ └── pipes_all.zip│├── README.md # Repository documentation, figure-code mapping, Data Availability├── Directory_Structure.txt # This file├── requirements.txt # Python dependencies└── LICENSE # MIT (code) + CC-BY-4.0 (data) Notes------ Boundary shapefiles (province / national border / nine-dash line / Qinling-Huaihe line / Hu Huanyong line) are NOT included in gis/; they must be obtained from the sources listed in the README "Data Availability" section.- Raw inputs (ERA5-Land, CMIP6, CLCD, LandScan, GDP, DEM, CMA rainfall) are external and not included; see README for acquisition URLs.- Supplementary figures S3-S8 and Supplementary Tables S1-S4 are not included in this release.