pydaisi

Python Data Assimilation Informed model Structure Improvement (PyDAISI): Python code to run the PyDAISI method applied to the GR2M monthly rainfall-runoff model.

What is pydaisi?

This package implements the Data Assimilation Informed model Structure Improvement (DAISI) method described in the following paper: Lerat, J., Chiew, F., Robertson, D., Andreassian, V., Zheng, H. (2023), "Data Assimilation Informed model Structure Improvement (DAISI) for robust prediction under climate change: Application to 201 catchments in southeastern Australia", WRR, Submitted.

Installation

• Create a suitable python environment. We recommend using miniconda combined with the environment specification provided in the [env_mini.yml] (env_mini.yml) file in this repository.
• Git clone this repository and run pip install .

Basic use

To access the data:

from pydaisi import daisi_data

# Get the site meta data
sites = daisi_data.get_sites()

# Select a site id among the 201 catchments 
# For example the Jamieson River at Gerrang Bridge,
# (site ID 405218)
siteid = 405218
monthly_data = daisi_data.get_data(siteid)

print(monthly_data) 
# This command shows:
#                 Rain      Evap      Qobs
# 
# 1970-07-01  135.9429   33.5447  186.7124
# 1970-08-01  253.6129   46.9463  207.2174
# 1970-09-01   87.7752   69.9889  119.4157
# 1970-10-01   51.4173  114.8741   56.1140
# 1970-11-01  110.2354  145.4254   27.9042
# ...              ...       ...       ...
# 2019-02-01   34.0550  146.5809    3.1161
# 2019-03-01   62.1733  117.4158    2.5748
# 2019-04-01   27.0699   75.8672    2.8548
# 2019-05-01  138.3442   44.7638    8.6404
# 2019-06-01  141.6448   32.5938   43.1690

To run DAISI applied to the GR2M model:

# Run DAISI step 0 - calibration of GR2M rainfall runoff model
# by default, the script calibrates the model for the 
# 201 catchments. 
python scripts/STEP0_gr2m_calibration.py

# The scripts can be run over a subset of sites (batch) 
# using the -n (number of batch) and -t (taskid=0..n-1) options
# This is useful if one wants to run the script using 
# parallel computing. The same options are available for 
# all scripts mentioned below.

# Run DAISI step 1 - apply Ensemble Smoother algorithym to GR2M
python scripts/STEP1_data_assimilation.py

# Run DAISI step 2 - fitting of update coefficients
python scripts/STEP2_model_structure_update.py

# Run DAISI step 3 - Computation of diagnostic metrics
python scripts/STEP3_dianostic_compute_metrics.py

# Run DAISI step 3 - Distribution of performance
python scripts/STEP3_dianostic_plot_metrics.py

License

The source code and documentation of the pydaisi package is licensed under the BSD license.