"Closing-the-Design-Ground-Truth-Overlap-Compliance-Gap"

"SUPPLEMENTARY MATERIALS PACKAGE \u2014 ABSTRACT==========================================Paper: Closing the Design-Ground-Truth Overlap Compliance Gap in      Multibeam Bathymetric Survey Line Planning via Prior-Data-Driven      Depth InterpolationAuthors: Jiahao Hu, Mindong Liu (corresponding), Zhi Cheng, Weipeng Zhou,        He Bai, Xiaojie ZhaoAffiliation: Jiangsu University of Science and Technology, Zhenjiang, ChinaContact: [e-mail] OVERVIEW----------------This supplementary materials package provides the complete source code,input datasets, and pre-computed result files required to reproduce allexperiments and figures presented in the paper. The package totalsapproximately 3.3 MB and is organized into three subdirectories: code\/,data\/, and results\/, with a README.md describing setup and usage.CONTENTS--------1. SOURCE CODE (code\/, ~280 KB, 15 Python files)  Core algorithm modules:  - run_gebco_case5_v3.py    Main pipeline implementing hierarchical                             terrain decomposition, plane and grid depth                             models, adaptive line spacing, and K-means                             terrain clustering.  - nsga2_optimizer.py       NSGA-II multi-objective optimizer (M7).  - dubins_planner.py        Dubins-curve transit path planning with                             greedy nearest-neighbor and 2-opt.  - isobath_planner.py       Auxiliary isobath-based line planner.  Experiment runners (each reproduces a specific table or figure in the paper):  - run_literature_comparison.py   Seven-method comparison (Table on M1-M6).  - run_nsga2_grid.py              M7 NSGA-II + grid depth experiment.  - run_monte_carlo.py             Monte Carlo robustness analysis under                                   Gaussian depth noise.  - run_sensitivity.py             Parameter sensitivity (RMSE threshold                                   tau and maximum region count N_max).  - run_cross_validation.py        NOAA cross-validation experiment using                                   independent high-resolution ground truth.  - run_multi_area_8.py            Eight-area generalization experiment                                   across diverse terrain types (RMSE                                   range 2.9-27.8 m).  Figure generation:  - generate_paper_figures.py      Main paper figures (study area,                                   decomposition, line layouts, metrics,                                   overlap distribution, Monte Carlo).  - gen_cross_val_figs.py          Cross-validation overview map and                                   eight-area bar chart.  - fix_fig2.py \/ fix_fig7.py \/ fix_fig8.py   Standalone scripts for the                                   decomposition, RMSE-vs-compliance,                                   and M3-vs-M5 comparison figures.2. DATASETS (data\/, ~3.0 MB, 4 files)  - real_bathymetry_subregion_v2.npz       Source: GEBCO 2023 global bathymetric grid.       Coverage: South China Sea sub-region, 4.8 x 4.4 nautical miles.       Grid: 19 x 24 raw points, resampled to 239 x 227 (~37 m spacing).       Depth range: 75 to 254 m.       Used as the primary validation dataset.       Arrays: lat, lon, depth, x_m, y_m.  - noaa_crm_subregion_1as.npz       Source: NOAA Coastal Relief Model (1 arc-second).       Coverage: Santa Barbara Channel, California.       Grid: 576 x 576 points at approximately 30 m resolution.       Depth range: 60 to 544 m.       Used as independent ground truth for the cross-validation       experiment and for the eight-area generalization test.       Arrays: lat, lon, depth, x_m, y_m.  - GEBCO_Grid_documentation.pdf       Official GEBCO grid documentation.  - GEBCO_Grid_terms_of_use.pdf        GEBCO data terms of use.3. PRE-COMPUTED RESULTS (results\/, ~14 KB, 4 CSV files)  - literature_comparison.csv       Seven-method comparison data (M1 through M6) including survey       length, transit length, total path, GT compliance, mean overlap,       coverage, and computation time.  - monte_carlo_summary.csv       Monte Carlo robustness results under depth noise levels       sigma = 0, 1, 2, 5, 10 m (50 trials per level), reporting mean       compliance, standard deviation, 5th\/95th percentiles, mean overlap,       and coverage.  - cross_validation_results.csv       Single sub-region cross-validation results for all seven methods,       comparing same-data evaluation against independent NOAA ground truth.  - multi_area_8_results.csv       Eight-area generalization experiment data for M3 (plane model) and       M5 (grid depth) across areas S1-S8, including depth range,       plane-fit RMSE, line counts, same-data compliance, cross-validated       compliance, circular bias, and coverage.REPRODUCIBILITY---------------Software requirements: Python 3.10+, with packages numpy, scipy, shapely,matplotlib, scikit-learn, and pymoo. Each experiment script can be runindependently from the code\/ directory after placing the data files atthe location expected by the scripts. Typical runtime is a few minutesper script on a standard workstation.All numerical results reported in the paper tables can be regeneratedfrom the code and data provided in this package, and each result filein the results\/ directory directly corresponds to one or more tables inthe manuscript.DATA SOURCES AND LICENSES-------------------------- GEBCO 2023 Grid: Provided by the GEBCO Compilation Group under the GEBCO Grid Terms of Use. Available at https:\/\/www.gebco.net\/- NOAA Coastal Relief Model: Provided by the NOAA National Centers for Environmental Information (public domain). Available at https:\/\/www.ngdc.noaa.gov\/mgg\/coastal\/crm.html "