Assessment of maize nitrogen uptake from PRISMA hyperspectral data through hybrid modelling

Area of Science:

• Earth Observation
• Precision Agriculture
• Spectroscopy

Background:

• The Italian Space Agency's PRISMA mission provides new Earth Observation (EO) data streams.
• Estimating biophysical crop variables (BVs) requires advanced methods for this new data.
• This study focuses on canopy chlorophyll content (CCC) and canopy nitrogen content (CNC).

Purpose of the Study:

• To evaluate a hybrid approach for retrieving CCC and CNC from synthetic PRISMA data.
• To assess the correlation between estimated crop variables and plant nitrogen uptake (PNU).
• To apply the developed model to actual PRISMA images for PNU estimation.

Main Methods:

• A hybrid approach combining the PROSAIL-PRO radiative transfer model and machine learning (ML) regression algorithms was used.
• PRISMA-like data were simulated from HyPlant airborne sensor imagery.
• ML algorithms were employed to retrieve CCC and CNC, and their relationship with PNU was analyzed.

Main Results:

• The hybrid approach successfully retrieved CCC and CNC from simulated PRISMA data.
• Canopy nitrogen content (CNC) showed a slightly stronger correlation with plant nitrogen uptake (PNU) (R² = 0.82) than canopy chlorophyll content (CCC) (R² = 0.80).
• The CNC-PNU model applied to actual PRISMA data yielded PNU estimates within expected ranges and consistent temporal trends.

Conclusions:

• The developed hybrid approach is effective for estimating crop nitrogen and chlorophyll content using PRISMA data.
• The findings demonstrate the potential of PRISMA data and ML for precision farming and sustainable agriculture.
• The study validates the use of satellite-based estimations for monitoring plant nitrogen status throughout the growing season.