README.md

    ______                    ______ __  __ _____  ____   _____
   / \  / \                  |  ____|  \/  |  __ \|  _ \ / ____|
  /   /\   \                 | |__  | \  / | |  | | |_) | |
 /   /  \   \                |  __| | |\/| | |  | |  _ <| |
 \--/----\--/                | |____| |  | | |__| | |_) | |____
  \/______\/                 |______|_|  |_|_____/|____/ \_____|

EMDBC - Empirical Mode Based Bias Correction

This repository contains the Python implementation for Empirical Mode Decomposition Based Bias Correction (EMDBC).

Please refer to the manuscript (citation below) for detailed methodology and theoretical background.

---

File Structure

├── emdbc.py
├── analysis.ipynb
└── validation_regions/
    ├── Midwest.nc
    ├── Northeast.nc
    ├── Northern.nc
    ├── Northwest.nc
    ├── Southeast.nc
    ├── Southern.nc
    └── Southwest.nc

---

Requirements

To run this code, ensure the following Python packages are installed:

• numpy
• scipy
• pandas
• matplotlib
• PyEMD
• pygam
• sklearn
• psutil
• scipy
• bottleneck
• cartopy

---

Usage

To apply bias correction using EMDBC, apply the Bias_Correction_EMD_disjoint method in emdbc.py.

Bias_Correction_EMD_disjoint(original_series, hist_series, fut_series, noise, filter_configs)

Perform bias correction on historical and future modeled time series using Empirical Mode Decomposition (EMD) and adaptive disjoint IMF selection.

Parameters:

• original_series (array-like):
  The observed time series to be used as the historical reference for bias correction.

• hist_series (array-like):
  The historical modeled time series to be used as the historical reference for bias correction.

• fut_series (array-like):
  The future modeled time series to be corrected.

• noise (float):
  The initial noise level to use for the ensemble EMD.

• filter_configs (dict):
  A dictionary that maps timescales (biweekly, seasonal, annual) to frequency bands (low and high cutoff values) for bandpass filtering:

filter_configs = {
    'Biweekly': (1/30, 1/3),
    'Seasonal': (1/180, 1/30),
    'Annual': (1/(365*2.5), 1/180)
}

Returns: