Overview¶

Bakaano-Hydro combines three layers:

Data preparation Static and dynamic raster inputs are downloaded, clipped, and aligned to a common DEM grid.
Hydrology VegET generates runoff and routes it through the river network.
Neural network The TCN-based model learns streamflow dynamics from routed hydrologic predictors and static catchment descriptors.

Pipeline diagram¶

Study area + working_dir
         |
         v
+-----------------------+
|  bakaano.data.*       |
|  DEM / Soil / NDVI    |
|  Tree cover / Meteo   |
|  AlphaEarth           |
+-----------------------+
         |
         v
+-----------------------+
|  bakaano.hydrology    |
|  VegET runoff/routing |
|  Rainfall features    |
+-----------------------+
         |
         v
+-----------------------+
|  bakaano.neuralnet    |
|  train / evaluate /   |
|  simulate             |
+-----------------------+
         |
         v
+-----------------------+
|  Outputs              |
|  model + CSVs         |
+-----------------------+

Package diagram¶

bakaano/
  core/         shared project helpers and utilities
  data/         preprocessing and downloads
  hydrology/    runoff, routing, rainfall products, plotting
  neuralnet/    training, evaluation, simulation
  extensions/   scenario and flood-mapping add-ons

Artifact flow¶

elevation/dem_clipped.tif
soil/*.tif
vcf/*.tif
ndvi/daily_ndvi_climatology.pkl
ERA5|CHIRPS|CHELSA/{prep,tasmax,tasmin,tmean}/*.nc
alpha_earth/band_A*.tif
     |
     v
runoff_output/wacc_sparse_arrays.pkl
runoff_output/rainfall_sparse_arrays.pkl   [optional]
     |
     v
models/bakaano_model.keras
models/alpha_earth_scaler.pkl
     |
     v
predicted_streamflow_data/*.csv

Which function should I use?¶

Task	Canonical entry point
Inspect project layout and readiness	`bakaano.core.project.ProjectContext`
Train with GRDC or CSV observations	`bakaano.neuralnet.train.train_streamflow_model`
Validate a trained model interactively	`bakaano.neuralnet.simulate.evaluate_streamflow_model_interactively`
Simulate official station sets	`bakaano.neuralnet.simulate.simulate_grdc_csv_stations`
Simulate arbitrary lat/lon points	`bakaano.neuralnet.simulate.simulate_streamflow`
Land-cover-change scenarios	`bakaano.extensions.scenario.ScenarioManager`
Flood mapping	`bakaano.extensions.flood_mapper.FloodMapper`