Unveiling Amazon Forest Vertical Dynamics through Multifrequency Microwave Radiometry

Monitoring the early stages of critical transitions in forest ecosystems is essential for improving our capacity to anticipate, adapt to, and mitigate their impacts. Such transitions may originate in the lower strata of forests, where structural and physiological changes can precede observable alterations in the upper canopy. Conventional satellite-based observations, largely reliant on optical vegetation indices and upper-frequency microwave bands (e.g., Ku-band), often fail to capture these early signals due to their limited penetration depth. In contrast, microwave remote sensing at low frequencies offers a unique capability to penetrate the canopy, providing valuable information on deeper forest layers. In our study, we use L- and C-band vegetation optical depth (VOD), a proxy of vegetation water content (VWC) to investigate and quantify the mean spatiotemporal patterns, the resilience, and stability of undisturbed regions of the Amazon Rainforest over the period 2010–2023. Our analysis reveals persistent interannual declines in VWC across several regions, with pronounced signals emerging around the 2015–2016 El Niño event onwards. These changes are particularly evident in the northeastern Amazon, a region characterized by the highest biomass across the basin. Further, quantitative estimates of resilience and stability indicate that this region exhibits the lowest stability within the Amazon Rainforests, with localized clusters showing declining resilience over the period analyzed. Notably, these emerging patterns remain largely undetectable using conventional optical vegetation indices but become apparent through the analysis of low-frequency microwave signals capable of penetrating the forest canopy. These findings highlight the critical role of multifrequency microwave satellite observations in understanding the dynamics of different forest layers and detecting early-warning signals of ecosystem stress. Integrating such approaches into forest monitoring frameworks could significantly enhance our ability to anticipate critical transitions and better inform conservation and management strategies aimed at mitigating large-scale forest dieback.