Evaluating and comparing remote sensing terrestrial GPP models for their response to climate variability and CO₂...

Area of Science:

• Earth and Environmental Sciences
• Ecology
• Remote Sensing

Background:

• Remote sensing (RS)-based models are crucial for estimating and monitoring terrestrial ecosystem gross primary productivity (GPP).
• Existing RS-based GPP models exhibit varying sensitivities to environmental factors, necessitating comparative analysis for accurate interpretation and model selection.
• Global GPP estimation is vital for understanding carbon cycling and climate change impacts.

Purpose of the Study:

• To globally evaluate and compare the sensitivities of 14 diverse RS-based GPP models.
• To benchmark these models against flux site measurements and Free-Air CO2 Enrichment (FACE) experiments.
• To identify how different model algorithms account for environmental factors, particularly CO2, temperature, radiation, and precipitation.

Main Methods:

• Evaluated 14 RS-based GPP models: 2 process-based, 4 vegetation-index-based, 5 light-use-efficiency-based, and 3 machine-learning-based.
• Benchmarked model responses against climatic factors at flux sites and elevated CO2 concentrations at FACE sites.
• Analyzed model sensitivity to CO2 (via greenness indices or photosynthesis), temperature, radiation, and precipitation across different spatial regions.

Main Results:

• Models sensitive to increasing atmospheric CO2 showed higher GPP trends, with CO2 effects implemented either via greenness indices or direct photosynthesis influences.
• Model responses to temperature and radiation varied significantly in magnitude and direction, while precipitation responses were more consistent.
• Model sensitivity differences were most pronounced in the tropics; high latitudes showed positive responses to temperature and radiation; mid-latitudes benefited from precipitation.

Conclusions:

• RS-based GPP models exhibit substantial differences in their representation of environmental controls, especially CO2 fertilization and water stress.
• No single model demonstrated consistent superior performance across diverse ecosystems and varying environmental conditions.
• Understanding model-specific sensitivities is critical for interpreting GPP estimates and selecting appropriate models for specific research questions.