Executed Notebooks

The full pipeline rendered with real outputs - data preparation (any AOI) and the Bali example.

The pages in this section are the actual Jupyter notebooks rendered with all outputs intact. They were executed end-to-end so you can read them as static walkthroughs - nothing is regenerated when the docs site builds.

Use them when you want to see:

Exactly which cells produced a given figure.
The full sequence of src/ calls per stage.
What the printed diagnostics look like (shapes, warnings, station counts, accumulator sizes).

Data preparation (Step 0, any AOI)

These two run once when you adapt the framework to a new region. The Bali bundle ships pre-prepared, so you can skip them for the Bali example.

Notebook	What it does
00 - Define AOI	Build the land-sea mask for your AOI: country boundary, bounding box, or user polygon. Output: `data/mask/aoi/{prefix}_mask.nc`.
01 - Data Acquisition	Download IMERG-L, IMERG-F, CPC-UNI (both 0.1° and native 0.5°) for your AOI and time span. Stitches multi-year NetCDFs.

The conceptual walkthrough for these two lives in the Data Preparation tutorial.

Bali pipeline (Step 2 onwards)

These five are pre-pinned to config_bali.yml and operate on the 11 MB Bali bundle (data/example_bali/). They live in notebooks/example_bali/ in the repo.

Notebook	What it does
02 - LSEQM+DL bias correction	LS → LSEQM → LSEQM+DL for all 36 dekads, with CNN training and confidence-mask blending.
03 - Measuring performances	Computes the 31 WMO-style metrics per dekad for each method.
04 - QA framework	Continuous Quality Index combining basic / distribution / temporal scores.
05 - Station validation	Independent BMKG-station validation, including per-station scatter and time series.
06 - Visualisation hub	Compilation maps, Taylor diagrams, threshold curves, and regional box plots.

Note

Re-running: the Colab badge inside each notebook points at the unexecuted template in notebooks/ (data prep) or notebooks/example_bali/ (Bali pipeline). The pages here are the rendered output of a prior local / Colab run and won’t re-execute as you read them.

Adapting to your own AOI

The _bali notebooks are the same code as the un-suffixed ones in notebooks/, with CONFIG_FILE = 'config_bali.yml' pinned. To run the framework on a different region: run nb00 + nb01 for that AOI, edit config.yml (or a copy), and point the un-suffixed notebooks at it - see the Configuration guide.

--- title: "Executed Notebooks" description: "The full pipeline rendered with real outputs - data preparation (any AOI) and the Bali example." --- The pages in this section are the actual Jupyter notebooks rendered with all outputs intact. They were executed end-to-end so you can read them as static walkthroughs - nothing is regenerated when the docs site builds. Use them when you want to see: - Exactly which cells produced a given figure. - The full sequence of `src/` calls per stage. - What the printed diagnostics look like (shapes, warnings, station counts, accumulator sizes). ## Data preparation (Step 0, any AOI) These two run once when you adapt the framework to a new region. The Bali bundle ships pre-prepared, so you can **skip them** for the Bali example. | Notebook | What it does | |----------|--------------| | [00 - Define AOI](00_define_aoi.ipynb) | Build the land-sea mask for your AOI: country boundary, bounding box, or user polygon. Output: `data/mask/aoi/{prefix}_mask.nc`. | | [01 - Data Acquisition](01_data_acquisition.ipynb) | Download IMERG-L, IMERG-F, CPC-UNI (both 0.1° and native 0.5°) for your AOI and time span. Stitches multi-year NetCDFs. | The conceptual walkthrough for these two lives in the [Data Preparation tutorial](../tutorials/data-preparation.qmd). ## Bali pipeline (Step 2 onwards) These five are pre-pinned to `config_bali.yml` and operate on the 11 MB Bali bundle (`data/example_bali/`). They live in [`notebooks/example_bali/`](https://github.com/bennyistanto/hybrid-bias-correction/tree/main/notebooks/example_bali) in the repo. | Notebook | What it does | |----------|--------------| | [02 - LSEQM+DL bias correction](02_lseqmdl_bias_correction_bali.ipynb) | LS → LSEQM → LSEQM+DL for all 36 dekads, with CNN training and confidence-mask blending. | | [03 - Measuring performances](03_measuring_performances_bali.ipynb) | Computes the 31 WMO-style metrics per dekad for each method. | | [04 - QA framework](04_qa_framework_bali.ipynb) | Continuous Quality Index combining basic / distribution / temporal scores. | | [05 - Station validation](05_station_validation_bali.ipynb) | Independent BMKG-station validation, including per-station scatter and time series. | | [06 - Visualisation hub](06_visualisation_hub_bali.ipynb) | Compilation maps, Taylor diagrams, threshold curves, and regional box plots. | ::: {.callout-note} **Re-running**: the Colab badge inside each notebook points at the *unexecuted* template in [`notebooks/`](https://github.com/bennyistanto/hybrid-bias-correction/tree/main/notebooks) (data prep) or [`notebooks/example_bali/`](https://github.com/bennyistanto/hybrid-bias-correction/tree/main/notebooks/example_bali) (Bali pipeline). The pages here are the rendered output of a prior local / Colab run and won't re-execute as you read them. ::: ::: {.callout-tip title="Adapting to your own AOI"} The `_bali` notebooks are the same code as the un-suffixed ones in `notebooks/`, with `CONFIG_FILE = 'config_bali.yml'` pinned. To run the framework on a different region: run nb00 + nb01 for that AOI, edit `config.yml` (or a copy), and point the un-suffixed notebooks at it - see the [Configuration guide](../get-started/configuration.qmd). :::