Radiative transfer model inversion using high-resolution hyperspectral airborne imagery - Retrieving maize LAI to...

Area of Science:

• Agricultural Remote Sensing
• Crop Physiology
• Precision Agriculture

Background:

• Mapping crop yield variability is crucial for precision agriculture.
• Leaf Area Index (LAI) is a key indicator of maize growth, biomass, and grain yield (GY).
• Hyperspectral imagery offers detailed vegetation information for crop monitoring.

Purpose of the Study:

• To estimate maize biomass and GY using LAI retrieved from hyperspectral aerial images via PROSAIL model inversion.
• To compare PROSAIL-based estimations with those from simple vegetation index approaches.
• To identify optimal growth stages for maize yield prediction.

Main Methods:

• Acquired hyperspectral aerial images over two maize fields with different irrigation systems.
• Collected ground LAI, biomass, and GY data.
• Calibrated and validated the PROSAIL model for LAI retrieval.
• Estimated biomass and GY using PROSAIL-retrieved LAI and various vegetation indices (NDRE, GNDVI, NDVI).

Main Results:

• PROSAIL model validation showed R² of 0.5 for LAI retrieval against ground data.
• NDRE vegetation index achieved the highest accuracy for biomass (R²=0.81) and GY (R²=0.83) estimation.
• The V16 late vegetative growth stage was optimal for maize yield prediction across all indices.

Conclusions:

• PROSAIL model inversion and vegetation indices derived from hyperspectral imagery are effective for estimating maize biomass and yield.
• The NDRE index at the V16 stage shows significant potential for accurate in-season maize yield prediction.
• This approach supports precision agriculture by providing detailed within-field yield variability information.