Dynamic of above ground biomass variation on a global scale over the last three decades.

Global vegetation plays a major role in the Earth carbon budget, storing the largest carbon stock on land. Both direct human activities and natural evolution under a changing climate impact the state of global forests, leading to regional decreases or vegetation growth. Monitoring these variations over long time periods can help better constrain estimates of the land carbon sink, and understand the driving forces of the cyclical and long term variations. This enables a refined understanding of climate effects and policies impact on current and future global vegetation carbon uptake.

Satellite records now span multiple decades, with microwave-based remote sensing providing complementary insights to optical observations. The lowest microwave frequencies are less affected by atmospheric perturbations and enable deeper penetration into the surface cover, with canopy penetration depth increasing with decreasing frequencies. However, achieving multi-decadal records requires the use of multiple instruments over time. These changes in instruments and observation types necessitate careful calibration and harmonization to produce consistent long-term time series of observations. The combination of different observation sources and different frequencies can be used as proxy to monitor geophysical variables variations such as the above ground biomass.

In this work we used a statistical model to combine observations of the Special Sensor Microwave - Imager, Special Sensor Microwave Imager Sounder and the C-band ERS/Advanced Scatterometer and Ku-band QSCAT to estimate above ground biomass on a global scale.  These models are applied to create a ~30 years time series of above ground biomass with R²>0.85 and RMSE<40 Mg/ha compared to the reference data from the CCI Biomass map. The retrievals are performed at different timescale highlighting the seasonality of vegetation cover and its impact on the microwave observations. The yearly estimates of AGB enable new insight into the dynamic of vegetation across different regions. The afforestation and deforestation effect can be evaluated across biomes, providing new estimates of the changes in carbon stocks at large scale.