Execution Pipeline

This page describes the high-level runtime flow from user input to final outputs.

At a glance

flowchart LR
    A[Input collection] --> B[Argument normalization]
    B --> C[Preprocessing and validation]
    C --> D[Core algorithm execution]
    D --> E[Post-processing and output writing]

Pipeline stages

1) Input collection

Purpose

Capture tool parameters from the selected runtime interface.

Main actions

GUI mode: parameters are collected from form widgets.
CLI mode: parameters are parsed from command-line arguments.

Inputs

User-provided tool parameters
Runtime context (GUI/CLI)

Outputs

Raw, mode-specific parameter payload

2) Argument normalization

Purpose

Build a consistent runtime argument set independent of call mode.

Main actions

Tool metadata defines required/optional parameters.
Framework options (processes, call_mode, log_level) are applied consistently.

Inputs

Raw parameter payload
Tool metadata/schema

Outputs

Normalized tool kwargs ready for execution

3) Preprocessing and validation

Purpose

Verify that inputs are usable and compatible before expensive computation starts.

Main actions

Check that the input folder exists and holds LAS/LAZ files, and create the output folder.
Check that the R packages the tool needs are installed, and install any that are missing.

Inputs

Normalized kwargs
Source data files

Outputs

Validated inputs and a ready R session
Early validation errors (if any)

4) Core algorithm execution

Purpose

Run the selected tool’s core processing logic.

Main actions

The tool wrapper sources r_scripts_collection.r and calls the tool's R function.
The R function processes the LAScatalog tile by tile with parallel R workers.

Inputs

Validated inputs
Execution options (including processes)

Outputs

One output per input tile (raster or point cloud)

5) Post-processing and output writing

Purpose

Finalize results and persist outputs for downstream use.

Main actions

Merge per-tile rasters into a Merged_*.tif mosaic when there is more than one tile.
Write tables and plots for the raster analysis tools.
Emit logs/progress information.

Inputs

Core algorithm outputs

Outputs

Saved output datasets
Runtime logs and progress signals

Typical workflow chain

flowchart LR
    A[LiDAR data info] --> B[Ground Classify Using CSF]
    B --> C[Generate DTM]
    B --> D[Normalize LiDAR Data]
    D --> E[Generate CHM or DSM]
    B --> F[Detect Hummock and Hollow]
    E --> G[Surface Water Mapping]
    C --> H[Depth-to-Water]
    G --> H
    C --> I[Seismic Line Disturbance Analysis]
    F --> I
    H --> I

Failure points to monitor

Failure signal	Likely cause	Pipeline stage	Mitigation
Error finding R, or rpy2 fails to import	R not installed in the environment, or `R_HOME` points elsewhere	Preprocessing and validation	See How R is located; run from an activated environment
`R packages could not be installed`	No internet access, or missing system libraries on Linux	Preprocessing and validation	Run `tests/test_r_environment.py`; install the GDAL/GEOS/PROJ/udunits development libraries
Hummock/hollow stops without output, or a CHM comes out as a DSM	Input not ground-classified, or not height-normalized	Core algorithm execution	Run Ground Classify Using CSF, then Normalize LiDAR Data; check with LiDAR data info
Out of memory or very slow processing	Too many R workers for a small cell size	Core algorithm execution	Lower `processes`; the TIN and kriging DTM tools already halve the workers below a 1 m cell size
Missing CRS reported as `Unknown`	LAS files without a CRS	Preprocessing and validation	Assign the CRS to the LAS files before processing