A parametric approach for global estimation of forest above-ground biomass with SMOS and SMAP L-band radiometer data

L-band radiometer data collected by the Soil Moisture Ocean Salinity (SMOS) and Soil Moisture Active Passive (SMAP) missions have shown potential for mapping the spatial distribution and temporal changes of the above-ground biomass (AGB) of forests. Most studies focussed on the relationships observed between AGB and estimates of the vegetation optical depth (VOD) derived from L-band radiometer data. We here present an approach for retrieving AGB from SMOS and SMAP brightness temperatures which builds upon existing AGB retrieval frameworks developed for active microwave data. A physically-based model was adapted to relate brightness temperatures to the percent canopy cover and height available from space-borne optical and LiDAR missions and, via modelled relationships between canopy cover, height, and AGB, to AGB. An initial set of 36 global AGB maps was produced from 10-days composites of a polarimetric index calculated from H and V polarization SMOS and SMAP brightness temperatures acquired in 2016. When compared to an ESA Climate Change Initiative Biomass AGB map, the AGB estimates produced from SMOS and SMAP presented a reasonable agreement with low systematic biases and explained, dependent on the type of forest, between 30 % and 80 % of the AGB variability in the reference map. A comparison with AGB reference information derived from plot-level inventory data for a limited number of sites across the major forest biomes indicated the merit of the suggested retrieval approach but also revealed a need for improving the retrieval algorithm locally. •New approach for biomass retrieval from SMOS/SMAP data independent of in-situ data.•Brightness temperatures modelled as function of forest biomass with parametric model.•Comparable retrieval performance across different seasons.•Global retrieval reveals retrieval errors dependent on forest type and region.•Suggested algorithm a flexible approach for optimizing retrieval performance locally.