PRINT: A semi-empirical canopy water removal approach for improving estimates of leaf nitrogen and grain protein contents in winter wheat
5 affiliations
- 1National Engineering and Technology Center for Information Agriculture (NETCIA), MARA Key Laboratory for Crop System Analysis and Decision Making, MOE Engineering Research Center of Smart Agriculture, Jiangsu Key Laboratory for Smart Agriculture, Nanjing Agricultural University, 666 Binjiang Avenue, Nanjing 211899, Jiangsu, PR China
- 2Hefei Meteorological Bureau, 16 Shihe, Hefei 230061, Anhui, PR China
- 3Precision Agriculture Lab, School of Life Sciences, Technical University of Munich, Freising, 85354, Germany
- 4State Key Laboratory of Wheat and Maize Crop Science, Agronomy College, Henan Agricultural University, 62 Nongye Road, Zhengzhou 450046, Henan, PR China
- 5National Engineering and Technology Center for Information Agriculture (NETCIA), MARA Key Laboratory for Crop System Analysis and Decision Making, MOE Engineering Research Center of Smart Agriculture, Jiangsu Key Laboratory for Smart Agriculture, Nanjing Agricultural University, 666 Binjiang Avenue, Nanjing 211899, Jiangsu, PR China; Corresponding author.
Abstract
Accurate estimation of leaf nitrogen content (LNC) and grain protein content (GPC) in winter wheat is crucial for making fertilization recommendations and assessing grain quality. Previous methods primarily relied on chlorophyll (Chl)-sensitive vegetation indices from the visible and near-infrared domain, while shortwave-infrared spectral features linked to dry matter and nitrogen remain underutilized due to strong leaf water masking. Although existing water removal methods can mitigate the absorption effects of leaf water, the reliability of the resulting simulated dry-canopy reflectance spectra (SDR) is limited by inaccurate leaf mass per area (LMA) retrieval. This study aimed to develop a semi-empirical water removal approach by integrating PRior knowledge and INitialization of leaf water conTent and LMA (PRINT) to simulate SDR, with the objective of improving the estimation performance of LNC and GPC. To achieve this goal, a Random Forest-based hybrid retrieval model was developed to increase LMA retrieval accuracy. Meanwhile, a visible and shortwave-based nitrogen index (VSNI) by combining the visible (proxy of Chl) and shortwave-infrared features was proposed for improving estimation performance of LNC and GPC. Moreover, experimental datasets covering multiple growth stages and geographic sites were collected to improve sample representativeness. These datasets were utilized to improve the retrieval of LMA and evaluate the performance of LNC and GPC estimations. The results demonstrated that the PRINT significantly improved the estimation accuracy of LMA over two standard physically-based inversion models (RMSE: 6.23 g/m 2 vs 8.25 g/m 2 and 45.73 g/m 2 ). Compared with existing VI-based models, the VSNI-based models yielded substantial improvements in LNC (R 2 : 0.62 ∼ 0.65, RMSE = 0.42 ∼ 0.54%) across sites, and enhanced GPC estimation accuracy up to R 2 = 0.66 (RMSE = 0.70%) at the peak filling stage across two years at one site. The proposed methodology advances the spectroscopic understanding of shortwave-infrared responses to nitrogen-related agronomic biochemicals, and holds significant potential for future applications in large-scale growth monitoring of wheat systems using satellite imagery.
Related
- Low-Salinity Rice-Straw Bioflocs Retain Nitrogen-Removal Capacity After Salinity Transfer and Improve Survival of Pacific White Shrimp.
- Contents list
- Shift in phospholipid and fatty acid contents accompanies brain myelination
- B-vitamin concentrations in raw milk from dairy cows receiving diets with different forages and grassland contents and the resulting fermented milk products (cheese and yogurt)
- Deep time dietary convergence evidenced by the stomach contents of a fossil seal from the Southwest Atlantic
BibTeX
@article{fang2026print,
title = {{PRINT: A semi-empirical canopy water removal approach for improving estimates of leaf nitrogen and grain protein contents in winter wheat}},
author = {Peng Fang and Yan Yan and Dong Li and Yapeng Wu and Wei Feng and Caili Guo and Xia Yao and Yan Zhu and Weixing Cao and Tao Cheng},
journal = {International Journal of Applied Earth Observations and Geoinformation},
year = {2026},
month = oct,
doi = {10.1016/j.jag.2026.105651},
url = {https://doi.org/10.1016/j.jag.2026.105651},
note = {CC BY},
}
RIS
TY - JOUR
TI - PRINT: A semi-empirical canopy water removal approach for improving estimates of leaf nitrogen and grain protein contents in winter wheat
AU - Peng Fang
AU - Yan Yan
AU - Dong Li
AU - Yapeng Wu
AU - Wei Feng
AU - Caili Guo
AU - Xia Yao
AU - Yan Zhu
AU - Weixing Cao
AU - Tao Cheng
PY - 2026
DA - 2026/10/30
JO - International Journal of Applied Earth Observations and Geoinformation
DO - 10.1016/j.jag.2026.105651
UR - https://doi.org/10.1016/j.jag.2026.105651
AB - Accurate estimation of leaf nitrogen content (LNC) and grain protein content (GPC) in winter wheat is crucial for making fertilization recommendations and assessing grain quality. Previous methods primarily relied on chlorophyll (Chl)-sensitive vegetation indices from the visible and near-infrared domain, while shortwave-infrared spectral features linked to dry matter and nitrogen remain underutilized due to strong leaf water masking. Although existing water removal methods can mitigate the absorption effects of leaf water, the reliability of the resulting simulated dry-canopy reflectance spectra (SDR) is limited by inaccurate leaf mass per area (LMA) retrieval. This study aimed to develop a semi-empirical water removal approach by integrating PRior knowledge and INitialization of leaf water conTent and LMA (PRINT) to simulate SDR, with the objective of improving the estimation performance of LNC and GPC. To achieve this goal, a Random Forest-based hybrid retrieval model was developed to increase LMA retrieval accuracy. Meanwhile, a visible and shortwave-based nitrogen index (VSNI) by combining the visible (proxy of Chl) and shortwave-infrared features was proposed for improving estimation performance of LNC and GPC. Moreover, experimental datasets covering multiple growth stages and geographic sites were collected to improve sample representativeness. These datasets were utilized to improve the retrieval of LMA and evaluate the performance of LNC and GPC estimations. The results demonstrated that the PRINT significantly improved the estimation accuracy of LMA over two standard physically-based inversion models (RMSE: 6.23 g/m 2 vs 8.25 g/m 2 and 45.73 g/m 2 ). Compared with existing VI-based models, the VSNI-based models yielded substantial improvements in LNC (R 2 : 0.62 ∼ 0.65, RMSE = 0.42 ∼ 0.54%) across sites, and enhanced GPC estimation accuracy up to R 2 = 0.66 (RMSE = 0.70%) at the peak filling stage across two years at one site. The proposed methodology advances the spectroscopic understanding of shortwave-infrared responses to nitrogen-related agronomic biochemicals, and holds significant potential for future applications in large-scale growth monitoring of wheat systems using satellite imagery.
SN - 1569-8432
SN - 1872-826X
ER -