Biomass estimation from simulated GEDI, ICESat-2 and NISAR across environmental gradients in Sonoma County, California

Estimates of the magnitude and distribution of aboveground carbon in Earth's forests remain uncertain, yet knowledge of forest carbon content at a global scale is critical for forest management in support of climate mitigation. In light of this knowledge gap, several upcoming spaceborne missions aim to map forest aboveground biomass, and many new biomass products are expected from these datasets. As these new missions host different technologies, each with relative strengths and weaknesses for biomass retrieval, as well as different spatial resolutions, consistently comparing or combining biomass estimates from these new datasets will be challenging. This paper presents a demonstration of an inter-comparison of biomass estimates from simulations of three NASA missions (GEDI, ICESat-2 and NISAR) over Sonoma county in California, USA. We use a high resolution, locally calibrated airborne lidar map as our reference dataset, and emphasize the importance of considering uncertainties in both reference maps and spaceborne estimates when conducting biomass product validation. GEDI and ICESat-2 were simulated from airborne lidar point clouds, while UAVSAR's L-band backscatter was used as a proxy for NISAR. To estimate biomass for the lidar missions we used GEDI's footprint-level biomass algorithms, and also adapted these for application to ICESat-2. For UAVSAR, we developed a locally trained biomass model, calibrated against the ALS reference map. Each mission simulation was evaluated in comparison to the local reference map at its native product resolution (25 m, 100 m transect, and 1 ha) yielding RMSEs of 57%, 75%, and 89% for GEDI, NISAR, and ICESat-2 respectively. RMSE values increased for GEDI's power beam during simulated daytime conditions (64%), coverage beam during nighttime conditions (72%), and coverage beam daytime conditions (87%). We also test the application of GEDI's biomass modeling framework for estimation of biomass from ICESat-2, and find that ICESat-2 yields reasonable biomass estimates, particularly in relatively short, open canopies. Results suggest that while all three missions will produce datasets useful for biomass mapping, tall, dense canopies such as those found in Sonoma County present the greatest challenges for all three missions, while steep slopes also prove challenging for single-date SAR-based biomass retrievals. Our methods provide guidance for the inter-comparison and validation of spaceborne biomass estimates through the use of airborne lidar reference maps, and could be repeated with on-orbit estimates in any area with high quality field plot and ALS data. These methods allow for regional interpretations and filtering of multi-mission biomass estimates toward improved wall-to-wall biomass maps through data fusion.