SWMM Params (MVP mapping layer)

Part of Agentic SWMM — install the project first for the executable toolchain (aiswmm CLI, SWMM solver, MCP servers).

What this skill provides

• Transparent CSV-to-JSON mapping for:
  • land use class -> SWMM [SUBCATCHMENTS] + [SUBAREAS] defaults
  • soil texture/type -> SWMM [INFILTRATION] (Green-Ampt) defaults
• Deterministic, auditable outputs with explicit fallback usage and unmatched-key reporting.
• Optional merge step that emits one builder-ready JSON artifact.

Scripts

• scripts/landuse_to_swmm_params.py
  • maps subcatchment_id + landuse_class to runoff/subarea parameters
• scripts/soil_to_greenampt.py
  • maps subcatchment_id + soil_texture to Green-Ampt infiltration parameters
• scripts/merge_swmm_params.py
  • merges outputs from the two mapping scripts into one JSON package for future swmm-builder

Default lookup tables

By default, scripts read bundled lookup CSVs:

• skills/swmm-params/references/landuse_class_to_subcatch_params.csv
• skills/swmm-params/references/soil_texture_to_greenampt.csv

You can override lookup paths with CLI flags.

Minimal input format

Land use input CSV:

• required columns: subcatchment_id, landuse_class

Soil input CSV:

• required columns: subcatchment_id, soil_texture

Example files are provided under examples/.

Outputs

Each mapper writes explicit JSON containing:

• records (row-level audit trail)
• sections (SWMM-oriented lists keyed by subcatchment)
• unmatched_* lists (rows that used fallback)
• counts summary

The merge script writes:

• sections (subcatchments, subareas, infiltration)
• by_subcatchment (combined record per subcatchment ID)
• incomplete_ids (IDs missing one or more sections)

CLI flags

All three scripts share these optional flags:

• --strict — fail instead of using the DEFAULT fallback row when an input key is missing from the lookup table. Useful for auditable production runs where silent fallback would mask a data gap.

landuse_to_swmm_params.py also accepts:

• --subcatchment-column \x3Ccol> — override the CSV column used as the subcatchment ID (default: subcatchment_id).
• --landuse-column \x3Ccol> — override the CSV column used as the land use class (default: landuse_class).

soil_to_greenampt.py also accepts:

• --subcatchment-column \x3Ccol> — override the CSV column used as the subcatchment ID (default: subcatchment_id).
• --soil-column \x3Ccol> — override the CSV column used as the soil texture/type (default: soil_texture).

MCP

MCP wrapper location:

• mcp/swmm-params/server.js

Exposed tools:

• map_landuse (inputCsvPath, optional lookupCsvPath, outputPath)
• map_soil (inputCsvPath, optional lookupCsvPath, outputPath)
• merge_params (landuseJsonPath, soilJsonPath, outputPath)

Quick start:

npm --prefix mcp/swmm-params install
npm --prefix mcp/swmm-params run start

MVP limitations

• Lookup mapping is key-based only (no spatial interpolation or fuzzy matching).
• A fallback row is expected in lookup tables (DEFAULT for land use, - or DEFAULT for soil).
• No unit conversion or calibration logic is included here.
• This skill only maps parameters; it does not write a full SWMM .inp.

Example commands

python3 skills/swmm-params/scripts/landuse_to_swmm_params.py \
  --input skills/swmm-params/examples/landuse_input.csv \
  --output runs/swmm-params/example_landuse.json

python3 skills/swmm-params/scripts/soil_to_greenampt.py \
  --input skills/swmm-params/examples/soil_input.csv \
  --output runs/swmm-params/example_soil.json

python3 skills/swmm-params/scripts/merge_swmm_params.py \
  --landuse-json runs/swmm-params/example_landuse.json \
  --soil-json runs/swmm-params/example_soil.json \
  --output runs/swmm-params/example_builder_params.json