Dataset for: Decoupling the Causal Effects of Vapor Pressure Deficit and Soil Moisture on Terrestrial Carbon Fluxes during Extreme Heatwaves

This dataset supplements the causal mediation analysis presented in "Decoupling the Causal Effects of Vapor Pressure Deficit and Soil Moisture on Terrestrial Carbon Fluxes during Extreme Heatwaves: A Double Machine Learning Approach." It contains processed daily AmeriFlux eddy covariance observations (NEE, TA, VPD, SW_IN) for 11 long-term sites spanning 2010–2024, merged with depth-weighted root-zone soil water content (SWC_ERA5, 0–100 cm) extracted from ECMWF ERA5-Land reanalysis data via Google Earth Engine. The dataset is structured to serve as direct input for the Double Machine Learning (DML) causal mediation pipeline, enabling the decoupling of direct thermal effects from indirect hydro-meteorological pathways across forest, grassland, and cropland biomes.