The 500-meter long-term winter wheat grain protein content dataset for China from multi-source data

Area of Science:

• Agricultural Science
• Remote Sensing
• Data Science

Background:

• Accurate wheat grain protein content (GPC) data is crucial for China's food security and market competitiveness.
• Previous GPC studies were limited by small scales, short durations, and lack of high-resolution data, especially across China's diverse terrain.
• A significant data gap exists for large-scale, long-term GPC estimation in China.

Purpose of the Study:

• To address the data gap and challenges in large-scale GPC estimation in China.
• To create the first 500-meter spatial resolution, long-term winter wheat dataset for China (CNWheatGPC-500).
• To develop and validate a robust model for estimating wheat GPC.

Main Methods:

• Integration of multi-source data from ERA5 and MODIS to construct the CNWheatGPC-500 dataset.
• Development of a GPC estimation model utilizing a hierarchical linear model approach.
• Rigorous validation using a dedicated dataset and cross-validation techniques (leave-one-year-out).

Main Results:

• The hierarchical linear model significantly outperformed conventional models for GPC estimation.
• The validation dataset achieved an R² of 0.45 and a Root Mean Square Error (RMSE) of 0.96%.
• Cross-validation demonstrated robust performance, with RMSEs ranging from 0.90% to 1.32% regionally and 0.77% to 1.11% for leave-one-year-out.

Conclusions:

• The CNWheatGPC-500 dataset provides unprecedented high-resolution, long-term GPC data for China's major winter wheat regions.
• The hierarchical linear model offers a superior approach for accurate GPC estimation.
• This work supports enhanced wheat production, quality control, and informed agricultural decision-making.