Quantifying key vegetation parameters from Sentinel-3 and MODIS over the eastern Eurasian steppe with a Bayesian...

Area of Science:

• Earth Observation
• Remote Sensing
• Ecology

Background:

• Accurate grassland monitoring is crucial for ecological studies.
• Sentinel-3 OLCI offers new potential for vegetation variable estimation.
• Previous evaluations of OLCI for grassland variables are limited.

Purpose of the Study:

• Evaluate Sentinel-3 OLCI and MODIS data for grassland LAI, FVC, and AGB estimation.
• Compare the accuracy of OLCI and MODIS data in the eastern Eurasian steppe.
• Assess the utility of red edge bands and Bayesian spatial modeling for improved estimations.

Main Methods:

• Utilized Sentinel-3 OLCI and MODIS satellite imagery.
• Employed a Bayesian spatial model (INLA-SPDE) to handle spatial autocorrelation.
• Compared model performance using red edge bands versus conventional visible and NIR bands.

Main Results:

• Sentinel-3 OLCI models showed higher accuracy than MODIS models for LAI, FVC, and AGB.
• Red edge bands improved estimation accuracy compared to visible and NIR bands.
• The INLA-SPDE model outperformed random forest and random forest kriging methods.

Conclusions:

• Sentinel-3 OLCI data is a valuable tool for grassland variable estimation.
• Red edge bands and advanced spatial modeling enhance remote sensing-based ecological monitoring.
• The findings support the use of OLCI for improved grassland management and research.