Moisture-Driven Phenology Regulates Seasonal Canopy Dynamics and Carbon Uptake in Tropical Deciduous Forests

This repository contains the near-surface imagery, co-located meteorological and soil observations, and the complete analysis code supporting the study "Moisture-driven phenology regulates seasonal canopy dynamics and carbon uptake in tropical deciduous forests" (submitted to Journal of Geophysical Research: Biogeosciences). The study uses daily PhenoCam imagery from a pheno-meteorological tower in a tropical moist deciduous forest in the Similipal Biosphere Reserve, eastern India (21.96° N, 86.49° E), over January 2023 – May 2024. Daily Green Chromatic Coordinate (GCC) is derived from a fixed canopy region of interest and analysed against co-located air temperature, relative humidity, rainfall, photosynthetically active radiation, and soil moisture and temperature at 5 cm and 20 cm depths. The analysis delineates five canopy phenophases, quantifies interannual differences, attributes phase-specific environmental controls using Random Forest and XGBoost models, characterises lead–lag structure with Granger-precedence analysis, and compares PhenoCam GCC against Sentinel-2 NDVI, EVI, and S2-GCC. Contents PhenoCam imagery and the extracted daily GCC time series for the study period. Co-located meteorological and soil observations (air temperature, relative humidity, rainfall, PAR, and soil moisture and temperature at 5 cm and 20 cm). Python notebooks reproducing all results and figures: GCC time-series smoothing and phenophase delineation; ΔGCC bootstrap confidence intervals; Mann–Whitney U comparisons; Random Forest and XGBoost training with permutation importance; Granger-precedence (lead–lag) analysis; and Sentinel-2–PhenoCam regression. requirements.txt specifying all package versions, and a README describing the file structure and execution order.