Skip to content

README.md

Latest commit

 

History

History
79 lines (55 loc) · 2.5 KB

README.md

File metadata and controls

79 lines (55 loc) · 2.5 KB

3_Crane: Irrigated Site Calibration Tutorial

This example demonstrates the complete SWIM-RS calibration workflow for an irrigated alfalfa site at Crane, Oregon (S2).

OpenET Ensemble Members

This example uses the open source OpenET ensemble members for ETf observations:

  • SIMS (Satellite Irrigation Management Support)
  • geeSEBAL (Google Earth Engine Surface Energy Balance Algorithm for Land)
  • PT-JPL (Priestley-Taylor Jet Propulsion Laboratory)
  • SSEBop (Operational Simplified Surface Energy Balance)

To re-extract the remote sensing data from Google Earth Engine, install SWIM-RS with the OpenET optional dependencies:

pip install swimrs[openet]

Pre-extracted data is provided in data/ so you can run the tutorials without these dependencies.

Overview

The tutorial covers:

  1. Running an uncalibrated model with default parameters
  2. Calibrating model parameters using PEST++ with remote sensing observations
  3. Running the calibrated model and evaluating improvement

Workflow

Run the notebooks in order:

Notebook Description
01_uncalibrated_model.ipynb Load data, run uncalibrated model, compare with OpenET ensemble
02_calibration.ipynb Set up and run PEST++ calibration using OpenET ETf and SNODAS SWE
03_calibrated_model.ipynb Run calibrated model, visualize parameter evolution, evaluate improvement

Configuration

  • Config file: 3_Crane.toml
  • PEST++ worker script: custom_forward_run.py
  • ETf source: OpenET ensemble (SIMS, geeSEBAL, PT-JPL, SSEBop)
  • Date range: 2003-01-01 to 2007-12-31

Site Details

Property Value
Site ID S2
Location Crane, Oregon
Crop Irrigated alfalfa
Irrigation Active since ~1996 (per IrrMapper)

Data

Pre-built data is provided in data/:

File Description
prepped_input.zip Model input data (JSON format)

Key Differences from Fort Peck

Aspect 3_Crane 2_Fort_Peck
Land use Irrigated alfalfa Unirrigated grassland
SWB mode CN (curve number) IER
Date range 2003-2007 1987-2022

Expected Results

The uncalibrated model underestimates irrigation and shows poor agreement with the OpenET ensemble. After calibration:

  • RMSE reduced by ~50%
  • Model learns site-specific irrigation patterns and crop coefficients

Requirements

  • Python environment with SWIM-RS installed
  • PEST++ (pestpp-ies) for calibration