Multi‐Season Configuration

Passive acoustic monitoring projects are typically repeated over multiple seasons or years. Maintaining a reproducible analysis workflow is therefore essential. pamflow addresses this need by leveraging the Kedro framework, which supports season-specific configurations through environments while allowing shared settings, such as a global output path, to be defined once.

Directory Structure

conf/
├── base/
│   ├── catalog.yml       ← shared catalog with ${globals:output_dir} references
│   ├── parameters.yml    ← shared default parameters
│   └── globals.yml       ← default output path (single-season projects stop here)
├── dry_2023/
│   ├── parameters.yml
│   ├── catalog.yml
│   └── globals.yml       ← overrides output_dir for this season
└── wet_2024/
    ├── parameters.yml
    ├── catalog.yml
    └── globals.yml

1. Set the Default Output Path

For projects with a single season, define the output path in conf/base/globals.yml:

# conf/base/globals.yml
output_dir: data/output

This is enough. No season-specific files are needed and all runs write to data/output/.

2. Override per Season

For multi-season projects, create a configuration folder inside conf/ and add a globals.yml to each season’s configuration folder:

# conf/dry_2023/globals.yml
output_dir: data/output_dry_2023

# conf/wet_2024/globals.yml
output_dir: data/output_wet_2024

3. Adjust Season Configuration

Here’s a version that is a bit more explicit about the relationship between each season, its deployment sheet, and the corresponding Kedro configuration.

3. Configure Each Season

Each seasonal environment under conf/ must have:

A field deployment spreadsheet describing the deployments for that season.
A catalog entry referencing that spreadsheet.
An audio_root_directory pointing to the root directory containing the audio recordings for that season.

For example, if your project contains the environments dry_2023 and wet_2023:

Field deployment spreadsheet

Create one deployment spreadsheet for each season, for example:

data/input/field_deployments/
├── field_deployments_sheet_dry_2023.xlsx
└── field_deployments_sheet_wet_2023.xlsx

Update the catalog

In each seasonal catalog (conf/<season>/catalog.yml), update the filepath to reference the corresponding deployment spreadsheet.

For example, in conf/dry_2023/catalog.yml:

field_deployments_sheet@pandas:
  type: pamflow.datasets.pamDP.field_deployments_sheet.FieldDeployments
  filepath: data/input/field_deployments/field_deployments_sheet_dry_2023.xlsx

and in conf/wet_2023/catalog.yml:

field_deployments_sheet@pandas:
  type: pamflow.datasets.pamDP.field_deployments_sheet.FieldDeployments
  filepath: data/input/field_deployments/field_deployments_sheet_wet_2023.xlsx

Update the audio directory

In each environment’s parameters.yml, set the root directory containing that season’s audio files.

For conf/dry_2023/parameters.yml:

audio_root_directory: /data/audio_dry_2023

For conf/wet_2023/parameters.yml:

audio_root_directory: /data/audio_wet_2023

The audio_root_directory should point to the directory containing all audio recordings associated with the corresponding field deployment spreadsheet. This ensures that each Kedro environment is fully configured to process a single monitoring season independently.

4. Run using environment

Single season (uses conf/base/globals.yml):

pamflow run

Specific season:

pamflow run --env=dry_2023
pamflow run --env=wet_2024

Kedro merges conf/base/ with the selected environment, with the environment values taking precedence. Each season writes to its own output folder, making reruns safe without overwriting previous results.