Urban Drainage – Simulations and Data Extraction with SWMM

1. Context

This repository contains the files, scripts, and reports generated from hydrological and hydraulic simulations using the Storm Water Management Model (SWMM).
The objective is to analyze the behavior of the urban drainage system under different rainfall, imperviousness, and soil parameter scenarios, as well as to consolidate datasets for use in supervised machine learning models.

---

2. Execution Plan

(i) Simulation of urban scenarios in SWMM

A total of 8 distinct scenarios were simulated, varying:

• Rainfall duration: 5, 20, 60 minutes
• Imperviousness levels: 75%, 87.5%
• Curve Number (CN): 85 (lots), 98 (streets)

For each scenario, the following outputs were generated:

• .rpt and .out files (raw SWMM outputs)
• Individual plots of depth curves at selected nodes
• .csv files containing hydrological statistics per subcatchment
• Hydraulic classification of storm drains: Normal, Overloaded, or Overflow

---

(ii) Construction of the consolidated database

From the simulations, a table was created where each row represents a subcatchment, containing:

Geometric and hydrological explanatory variables (X):

• Subcatchment area (m²)
• Imperviousness (%)
• Curve Number (CN)
• Subcatchment type (lot or street)
• Overland flow path length (m)
• Average terrain slope (%)
• Rainfall duration (min)
• Total precipitation volume (mm)
• Maximum depth (m)
• Time to peak (min)
• Depth/inspection chamber height ratio (dimensionless)

Soil-related explanatory variables (when simulated via Green-Ampt):

• Saturated hydraulic conductivity (Ksat, mm/h)
• Suction head at wetting front (Ψ, mm)
• Initial soil moisture (θinit, fraction)
• Effective soil layer depth (m)
• Soil texture (categorical: sandy, loam, clay, etc.)

Target variable (y):

• Operational class of storm drain: Normal / Overloaded / Overflow

The final consolidated dataset is stored in df_final.csv, including all explanatory variables and the corresponding target classification.

---

3. Scripts and Outputs

Main scripts

• scenarios_peak_depth_analysis.py → analyzes depth in a single scenario
• scenarios_global_peak_analysis.py → consolidates statistics and curves across scenarios
• scenarios_input_audit.py → audits input .csv files
• scenarios_data_extractor.py → extracts and organizes hydrological, hydraulic, and soil variables

Generated outputs

• .csv statistics (maximum, minimum, mean, standard deviation)
• Time series plots in .png or .pdf
• Audit reports in .txt
• Structured datasets in .csv and .parquet

---

4. Folder Structure

/scenarios
   ├── scenarioXX.rpt                # Original SWMM report files
   ├── scenarioXX_peak_statistics.csv
   ├── scenarioXX_depth_timeseries.png / .pdf
   ├── scenarioXX.csv                 # Structured datasets
   ├── scenarioXX.parquet
   ├── scenarioXX_audit_report.txt
   ...
/outputs
   ├── all_scenarios_peak_statistics.csv
   ├── all_scenarios_max_curves.pdf
scenarios_peak_depth_analysis.py
scenarios_global_peak_analysis.py
scenarios_input_audit.py
scenarios_data_extractor.py

---

5. Tools and Technologies

• SWMM for hydrological and hydraulic simulations
• Python (pandas, matplotlib, pySWMM) for data extraction, analysis, and visualization
• Apache Parquet for optimized data storage
• Audit scripts to ensure transparency and traceability of the data

---

6. Notes

• All depths are expressed in meters (m)
• Rainfall duration in minutes (min)
• Volumes in m³ or mm, depending on the variable
• Data are suitable for scientific publications, reproducibility studies, and machine learning applications

---