DSM Workflow Documentation

Author

ISRIC - World Soil Information

Published

2026-05-26

Welcome

DSM Workflow Documentation

This is a Digital Soil Mapping (DSM) workflow developed at ISRIC - World Soil Information.

This workflow starts from standardised inputs, it organises data for modelling, performs modelling with Random Forest and creates soil maps as final outputs, including pixel based uncertainty with Quantile Regression Forests.

It is a file-based workflow with each step needing specific files as inputs and creating specific files as outputs. Files are named and organised according to the main variables/options chosen by the user. A configuration file allows the user to keep track of the chosen options.

Start with Chapter 1 Getting started and follow the instructions:

In 1.1 Installation we give installation instructions for the workflow and required dependencies.
In 1.2 Inputs we explain how to prepare the input data.
In 1.3 Running the workflow we explain how to run the workflow. This process requires two files: 1) the configuration file a YAML script, that we will simply call config file, and 2) an R script, to which we will refer as workflow script.

In Chapter 2 Configuration file we introduce the config file.

In 2.1 Get the configuration file we present the template config file.
In 2.2 Config file options, the options that can be set in the config file for running the seedlings are explained.

The workflow script in 3 Workflow script compiles the commands that trigger the workflow to run.

4 Modeling Choices describes main conceptual aspects that are important for understanding the model.

5 Design Choices mentions key points about how data and functions are handled in the Seedling workflow.

Note that 4 Modeling Choices and 5 Design Choices provide a deeper look at the options in the config file.

Frequently Asked Questions (FAQ) can be found in 6 FAQ.

The licence for this software is in Chapter 7 LICENCE:

The code in this repository is released under the GPL3 license according to the ISRIC data policy.

Disclaimer

The code is provided without warranty. ISRIC is not obliged to provide user support, updates or “bug fixes” of any kind.

Feedback, bugs and questions

Feedback and bug reports are welcome. Please contact us at dsm at isric dot org

Citation

Genova, G., Poggio, L., Kempen, B., & Colman, B. (2024). DSM Workflow Seedling. ISRIC - World Soil Information. https://doi.org/10.17027/ISRIC-FSX2-2691