Wheat LAI Inversion and Yield Estimation under T-MLP
International Journal of Advanced Engineering and Technology Research · 02.04.2026 · 1 (3) · 1-14
DOI: 10.54097/ghhm1r17 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Advanced Engineering and Technology Research, online erschienen am 02.04.2026. Verfasst von Huan Li. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
To address the limited accuracy and poor generalization capabilities of existing winter wheat Leaf Area Index (LAI) inversion methods, such as the NDVI vegetation index model and support vector machine (SVM), this paper proposes a hybrid modeling approach that integrates a radiative transfer model with deep learning. This method uses the PROSAIL model to generate simulated samples, optimizes the input parameter configuration through local and global sensitivity analysis, constructs a lookup table (LUT) to assist in the generation of training sets, and designs a T-MLP network that integrates Transformer and Multilayer Perceptron (MLP). This approach leverages the cross-band global dependencies in Landsat 8multispectral data and enhances LAI modeling capabilities through nonlinear regression. Using ground-truth data from the main wheat-producing region of Shangqiu City, Henan Province, this model's performance is validated and compared with the NDVI, PROSAIL and CNN models. Winter wheat yield estimation is then performed using the LAI inverted by the T-MLP model in combination with the EC-LUE model. Results showed that T-MLP achieved superior inversion accuracy across multiple growth stages (R²=0.83, RMSE=1.04), outperforming the NDVI model (R²=0.55, RMSE=1.58), the PROSAIL model (R²=0.76, RMSE=1.36) and CNN model (R²=0.78, RMSE=1.12). Comparison of the inversion results with the MODIS LAI product revealed that T-MLP demonstrated greater robustness in terms of temporal consistency and cross-plot stability, accurately reflecting the dynamic trends of wheat LAI. The relative error of yield predictions was below 10%. This study provides a practical approach for high-precision LAI inversion and yield estimation that integrates physical constraints with the advantages of deep learning.
Zitieren
APA
Li, H. (2026). Wheat LAI Inversion and Yield Estimation under T-MLP. International Journal of Advanced Engineering and Technology Research, 1(3), 1-14. https://doi.org/10.54097/ghhm1r17
Vancouver
Li H. Wheat LAI Inversion and Yield Estimation under T-MLP. International Journal of Advanced Engineering and Technology Research. 2026;1(3):1-14. doi:10.54097/ghhm1r17
BibTeX
@article{Li2026,
author = {Li, Huan},
title = {Wheat LAI Inversion and Yield Estimation under T-MLP},
journal = {International Journal of Advanced Engineering and Technology Research},
year = {2026},
volume = {1},
number = {3},
pages = {1--14},
doi = {10.54097/ghhm1r17}
}
EUVIMED