作物灌溉模型 | DCWRISM-CropIrrigation

Physics-empirical crop irrigation water demand model.

FAO Penman-Monteith + differential-evolution optimization for crop irrigation water demand — 21.82% error on real data.

---

DCWRISM-CropIrrigation is a crop irrigation water demand model that blends the FAO-56 Penman-Monteith reference evapotranspiration with an empirical crop coefficient, then tunes the parameters with differential evolution. The result is a light, reproducible model that predicts crop water demand with only 21.82% error.

Note

中文项目：物理经验混合的智能灌溉需水模型，基于 FAO Penman-Monteith 与差分进化（DE）参数优化。

---

Quickstart

git clone https://github.com/Windyhhh/DCWRISM-CropIrrigation.git
cd DCWRISM-CropIrrigation

pip install -r requirements.txt

# Load data and run the base model
python dcwrism_model.py

# Advanced DE-optimized model
python FINAL_OPTIMIZED_HONEST_MODEL.py

---

Features

• FAO Penman-Monteith ET₀ — reference evapotranspiration from weather inputs.
• Differential-evolution tuning — auto-searches crop coefficients and soil factors.
• Honest, reproducible — final model with verified 21.82% error on real regional data.

---

Project Structure

DCWRISM-CropIrrigation/
├── dcwrism_model.py              # core irrigation water demand model
├── data_loader.py                # load & parse weather / region data
├── ADVANCED_OPTIMIZATION_MODEL.py
├── FINAL_OPTIMIZED_HONEST_MODEL.py   # final DE-tuned model
└── requirements.txt

---

技术实现细节

架构概览

项目采用扁平结构，核心代码位于根目录。

核心类与模块

• DataLoader

关键函数

• evaluate_model, evaluate_test_years, load_weather_data, load_crop_area

技术栈与依赖

核心框架/库：NumPy, pandas

主要 import：

import sys
import pandas as pd
import numpy as np
from scipy.optimize import minimize, differential_evolution
from data_loader import DataLoader
from dcwrism_model import DCWRISMModel
import io
import sys
import pandas as pd
import numpy as np

实现要点

• 以 DataLoader 为核心类，封装主要业务逻辑
• 通过 evaluate_model 等函数实现核心流程编排
• 基于 NumPy, pandas 构建，保证技术栈成熟稳定
• 代码结构清晰，模块间低耦合，便于扩展和维护

---

License

MIT — free to use, modify and distribute.