Vegetation optical depth reveals changes in ecosystem-level water stress for global forests

Plant water stress due to climate change is posing a threat to various ecosystem services such as carbon sequestration, food and wood production, and climate regulation. To address this issue, methods are needed to assess and monitor plant water stress at various spatial and temporal scales. Passive microwave emission observations from satellites have proven useful in monitoring changes in vegetation water content and assessing plant water stress at a low spatial resolution (> 9 km). In this study, we used vegetation optical depth (VOD) and measurements of hydraulic vulnerability to create a novel model for assessing ecosystem-level water stress. We used L-band VOD and global measurements of xylem water potential at 88% loss of stem hydraulic conductivity (P88) from the TRY database (including 1103 measurements of P88 from 463 species and nine different vegetation biomes) to create a linear regression model between L-band VOD and biome-level P88. We used monthly mean values of L-band VOD and calculated ratios of yearly minimum and maximum VOD (L-VODmin/max) for each pixel to describe average variability in ecosystem-level water content. The developed L-VODmin/max metric explained 75% of the variation in P88 at the biome level (R2=0.75) indicating that the novel L-VODmin/max metric is capable of capturing changes in plant water status. We then used the L-VODmin/max metric and daily climate data from the ERA5 to see if water stress has increased over time in the world's forests that are more water limited (aridity index below 1.5). For these areas, we found a positive trend in maximum daily vapour pressure deficit, which correlated negatively (p<0.05) with L-VODmin/max trend for the same time period further confirming that L-VODmin/max is capable of explaining differences in plant water status. Additionally, we examined the trend in L-VODmin/max for global forests for the same 2011-2020 period and found a significant negative trend (increasing water stress, p<0.05) for forests in central Africa, southeast Asia, and eastern Australia, and a positive trend (decreasing water stress) for boreal forests in North America and rainforests in Indonesia. Further studies are required to confirm our results suggesting that some of the world's largest carbon sinks are experiencing rapid changes in water stress as a result of climate change.