An Artificial Intelligence-Based Framework for Modeling Vegetation Photosynthesis Anomalies in India Corresponding to...

Area of Science:

• Earth System Science
• Climate Science
• Remote Sensing

Background:

• Interannual climate variability, driven by oceanic perturbations like ENSO, significantly impacts the land carbon cycle and global food production.
• Terrestrial vegetation's response to large-scale climate patterns is crucial for predicting food anomalies and understanding carbon cycle dynamics.

Purpose of the Study:

• To investigate the temporal variability of vegetation photosynthesis in India using solar-induced chlorophyll fluorescence (SIF) anomalies.
• To evaluate the response of Indian vegetation activity to sea surface temperature (SST) variations in the tropical Pacific.
• To develop an AI-driven model for predicting vegetation photosynthetic variability based on SST teleconnections.

Main Methods:

• Utilized solar-induced chlorophyll fluorescence (SIF) anomalies to assess vegetation photosynthesis variability over India.
• Analyzed the correlation between Indian SIF anomalies and Niño 3 SST anomalies during the June-August (JJA) period (2001-2020).
• Employed a non-linear framework with a multi-layer perceptron (MLP) architecture, integrating tropical SSTs and local hydroclimate data, to model SIF anomalies. Incorporated data augmentation with seasonal lags and an ensemble approach for model refinement.

Main Results:

• A robust inverse relationship (R = -0.68) was found between Indian SIF anomalies and Niño 3 SST anomalies during JJA.
• The developed MLP model, especially the ensemble at lags 1-3 (Ensemble JJA,1-3), effectively captured SIF variability (R² = 0.9).
• Explainable AI identified Niño 3 and western Indian Ocean SST anomalies as significant contributors (~45%) to vegetation photosynthetic variability.

Conclusions:

• Oceanic SST anomalies, particularly from the tropical Pacific, exert a substantial influence on Indian vegetation photosynthesis.
• The AI-enabled non-linear modeling approach successfully captures complex SST-vegetation teleconnections.
• This research enhances the capacity for early warnings of vegetation stress in India, crucial for climate adaptation and food security.