Statistical Analysis and Tectonic Implications of Drainage Network in Southern Haiti

This dataset is associated with the paper intitled "Systematic tectonic control of drainage networks in southern Haiti" by Newdeskarl Saint Fleur. - Saint_Fleur_catchment_data.csv contains all the relative to the data relative to the watersheds extracted from ASTER DEM. The column names are intruitive and informative. Examples: ws_location_n_s stands for location of watershed (ws) either the broader north or broader south (n_s); 'ws_location_n_s_only' corresponds to strictly north and south; ws_geology identifies the geological units underlying the watersheds; main_river_aster_topo_mean stands for the mean topography (elevation) of the main river of the watersheds; hi stands for hypsometric integral. - Saint_Fleur_channel_data.csv retakes the column names in Saint_Fleur_catchment_data.csv with a few specifications to rivers.- Saint_Fleur_correlation_matrix.xlsx: correlation matrix table of the numerical variables (columns).-Saint_Fleur_boxplot_for_outliers.png : Boxplots for checking outliers.- Saint_Fleur_Chi2_Test_All_CategVar.csv contains table of the Chi-2 test of caterogical variables-Saint_Fleur_ShapiroWilkTest_UntransformedData.csv , Saint_Fleur_ShapiroWilkTest_Log10TransformedData.csv , Saint_Fleur_ShapiroWilkTest_Log1pTransformedData.csv correspond to results of Shapiro-Wilk test with untransformed and transformed data (Log10 and Log1).-Saint_Fleur_ShapiroTests.png : Shapiro-Wilk test visualization -Saint_Fleur_anova_results.csv : Anova test results-Saint_Fleur_anova_fig.png : Anova test visualization -Saint_Fleur_kruskal_wallis_results.csv : Kruskal-Wallis results-Saint_Fleur_kruskal_wallis_fig.png: Kruskal-Wallis visualization -Saint_Fleur_variance_inflation_factor.csv : Variace inflation faction for feature selection.-Saint_Fleur_n_clusters_different_metrics.png : Different metrics for choosing the number of clusters.-Saint_Fleur_visualization_of_clusters_for_different_k.png : Cluster visualization for different k (number of clusters).-Saint_Fleur_figures_correlation_for_fillna.png: Scatter plots demonstrating strong correlations between watershed and channel morphometric parameters used for missing value imputation. High correlation coefficients (R² > 0.7) between independently measured parameters validate the morphometric relationships and enable robust statistical analysis of tectonic signals. (a-l) Progressive correlation chains used to fill missing values through regression relationships: (a) watershed perimeter vs. main river length, (b) main river length vs. elevation count, (c) elevation count vs. topographic variety, (d) watershed mean elevation vs. river maximum elevation, (e) river maximum vs. elevation range, (f) elevation range vs. elevation variance, (g) watershed elevation range vs. river elevation sum, (h,i) river minimum elevation vs. minority/majority elevations, (j) river maximum vs. mean elevation, (k) river mean vs. median elevation, (l) steepness index vs. watershed median elevation. R²: coefficient of determination; MAE: mean absolute error. The consistency of these relationships across the entire dataset demonstrates the morphometric coherence necessary for detecting systematic tectonic signals. -Saint_Fleur_detailed_channel_profiles_1.png: Detailed channel profiles -Saint_Fleur_detailed_channel_profiles_2.png: Detailed channel profiles -Saint_Fleur_detailed_channel_profiles_3.png: Detailed channel profiles -Saint_Fleur_detailed_channel_profiles_4.png: Detailed channel profiles -Saint_Fleur_detailed_hypsometric_curves_1.png: Detailed hypsometric curves -Saint_Fleur_figure_VIF_correlation_graph.png: Multicollinearity assessment and correlation structure analysis guiding feature selection for watershed classification. (a) Variance Inflation Factor (VIF) analysis identifying redundant variables. Variables with VIF > 10 (red threshold) exhibit problematic multicollinearity requiring removal or combination. Extremely high VIF values (>1000) in size-related parameters (watershed area, river length) and elevation statistics indicate strong interdependence requiring careful variable selection. (b) Correlation network graph (threshold = 0.9) revealing clusters of highly correlated variables. Node size represents variable importance; connections indicate correlations >0.9. Distinct clustering patterns justify representative variable selection: watershed size metrics cluster, elevation variability cluster, and topographic statistical measures group. Strategy: From each cluster, the variable with highest variance (red) and lowest multicollinearity was selected as representative, ensuring independent features for clustering analysis while maintaining morphometric information content. This rigorous feature selection enables robust statistical classification by eliminating redundant information while preserving essential morphometric signals related to tectonic and lithological controls. The systematic correlation structure demonstrates the internal consistency of morphometric relationships necessary for detecting meaningful geological patterns. -Saint_Fleur_relationship_watershed_main_river.png: Relationship between watershed area and main river length revealing systematic spatial patterns related to structural controls. Strong positive correlation (R² = 0.93) validates morphometric relationships across four orders of magnitude in watershed size. Points are colored by: (a,b) watershed location relative to EPGF trace, (c) main river flow direction, and (d) main river orientation. (b) Simplified version treating N/S and S/N categories as primarily N and S respectively. Key observations: (1) Largest watersheds (N/S, S/N categories) successfully cross the topographic barrier created by the EPGF, indicating their pre-existing establishment or sufficient drainage power to defeat active tectonism; (2) Northern watersheds (N=84) are systematically fewer than southern watersheds (N=150), reflecting asymmetric landscape development across the fault zone; (3) East-west flowing rivers are predominantly small, suggesting capture by the dominant N-S drainage pattern controlled by regional structural grain. This size distribution provides first-order evidence for systematic tectonic control on drainage network organization. R²: coefficient of determination; MAE: mean absolute error. -Saint_Fleur_hypsometric_curves_1.png: Hypsometric curves for representative watersheds across southern Haiti, revealing systematic patterns related to tectonic activity and lithological controls. Curves plot normalized cumulative area (y-axis) versus normalized elevation (x-axis), standardizing watersheds for direct comparison regardless of size or absolute elevation. Curve interpretation: Convex curves (high area at high elevations) indicate youthful or tectonically rejuvenated landscapes; S-shaped curves suggest mature equilibrium conditions; concave curves reflect advanced erosional development. Key patterns: (1) Northern watersheds often display convex upper segments and abrupt slope transitions, particularly those intersecting E-W fault zones (e.g., Catchments 8, 51, 52); (2) Watersheds spanning the EPGF zone show complex, multi-inflection curves reflecting structural complexity (e.g., Catchments 60-62); (3) Western region catchments exhibit relatively smooth, regular curve progressions consistent with stable geological conditions (e.g., Catchments 73-76); (4) Mixed lithological assemblages create distinctive curve irregularities at elevation transitions corresponding to geological boundaries. The diversity of curve shapes reflects the complex interaction between active tectonics, lithological resistance, and landscape evolution processes operating across the transpressional EPGF system. -Saint_Fleur_hypsometric_curves_2.png: Continued hypsometric curves for representative watersheds. Curves shown here emphasize southern and eastern peninsula watersheds, demonstrating the most extreme hypsometric signatures. Notable examples include eastern catchments (93-106) displaying highly irregular curves with pronounced inflection points, consistent with complex structural and lithological controls characteristic of the Massif de la Selle region. Southern catchments (1, 21, 23-24, 47-48) show variable curve shapes reflecting the transition from stable central regions to tectonically active eastern zones.