Neinavaz et al. (2025) Evaluating Airborne Thermal Infrared Hyperspectral Data for Leaf Area Index Retrieval in Temperate Forests

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Remote Sensing
• Year: 2025
• Date: 2025-11-26
• Authors: Elnaz Neinavaz, Roshanak Darvishzadeh, Andrew K. Skidmore, Marco Heurich, Xi Zhu
• DOI: 10.3390/rs17233820

Research Groups

Specific research groups, labs, or departments are not explicitly detailed in the provided text. The study was conducted in the Bavarian Forest National Park in southeastern Germany, involving the EUFAR-TIR flight campaign.

Short Summary

This study evaluates the reliability of retrieving Leaf Area Index (LAI) using in situ and airborne thermal infrared (TIR) hyperspectral data, moving beyond controlled laboratory conditions. It found that Partial Least Squares Regression (PLSR) and Artificial Neural Network (ANN) models, particularly with the Scaled Conjugate Gradient algorithm, effectively predict LAI, identifying specific TIR wavebands that are robust across varying environmental conditions.

Objective

• To evaluate the reliability of retrieving Leaf Area Index (LAI) using in situ and airborne thermal infrared (TIR) hyperspectral data, building upon previous findings from controlled laboratory conditions.

Study Configuration

• Spatial Scale: 36 plots, each 30 m × 30 m in size, with airborne sensor data at 3 m spatial resolution.
• Temporal Scale: Field data collection and airborne flight campaign conducted on 6 July 2017.

Methodology and Data

• Models used: Partial Least Squares Regression (PLSR), Artificial Neural Network (ANN) models (using Levenberg–Marquardt and Scaled Conjugate Gradient algorithms).
• Data sources: In situ and airborne thermal infrared (TIR) hyperspectral data collected using an AISA Owl TIR hyperspectral sensor; emissivity data derived from six narrowband indices computed from wavebands between 8 µm and 12.3 µm.

Main Results

• TIR narrowband indices demonstrated poor performance in estimating in situ LAI (R² = 0.28, RMSECV = 0.02) compared to controlled conditions.
• The PLSR model achieved high prediction accuracy for LAI retrieval (R² = 0.86, RMSECV = 0.36).
• The ANN approach using the Levenberg–Marquardt algorithm showed lower accuracy (R² = 0.56, RMSECV = 0.71).
• The ANN approach using the Scaled Conjugate Gradient algorithm outperformed the Levenberg–Marquardt algorithm (R² = 0.83, RMSECV = 0.18), demonstrating the lowest RMSECV among the tested models.
• Key wavebands located at 8.1 µm, 9.1 µm, 9.85–9.95 µm, and 9.99–10.27 µm were identified as equally effective in predicting LAI, regardless of sensor or measurement/environmental conditions.

Contributions

• First evaluation of TIR hyperspectral data for LAI retrieval under in situ and airborne conditions, validating previous laboratory findings.
• Identification of specific TIR wavebands (8.1 µm, 9.1 µm, 9.85–9.95 µm, and 9.99–10.27 µm) that are robust for LAI prediction across varying environmental conditions.
• Provides important implications for upscaling LAI predictions and informs the capabilities of upcoming thermal satellite missions (e.g., Landsat Next, Copernicus LSTM).

Funding

No specific funding projects, programs, or reference codes are provided in the paper text.

Citation

@article{Neinavaz2025Evaluating,
  author = {Neinavaz, Elnaz and Darvishzadeh, Roshanak and Skidmore, Andrew K. and Heurich, Marco and Zhu, Xi},
  title = {Evaluating Airborne Thermal Infrared Hyperspectral Data for Leaf Area Index Retrieval in Temperate Forests},
  journal = {Remote Sensing},
  year = {2025},
  doi = {10.3390/rs17233820},
  url = {https://doi.org/10.3390/rs17233820}
}