An Examination of the Amount of GEDI Data Required to Reliably Characterize Central Africa Tropical Forest Aboveground Biomass at REDD+ Project Scale

The Global Ecosystem Dynamics Investigation (GEDI) is the first spaceborne LiDAR designed to improve quantification of vegetation structure and forest aboveground biomass (AGB, units Mg ha-1) including in the tropics where forest AGB inventories are sparse. Reducing Emissions from Deforestation and Degradation (REDD+) projects require forest AGB inventories to quantify avoided carbon emissions achieved by preserving the forest. In tropical Africa the average REDD+ project size is 286,876 ha (equivalent to 50 × 50 km). This paper examines the amount of GEDI data (number orbits, days, and months) needed to reliably characterize tropical forest AGB at REDD+ project scale. Thirty one months of GEDI footprint-level AGB product data acquired over 15 50 × 50 km sites across Mai Ndombe province (12.3 million ha) in the Democratic Republic of the Congo were considered. The GEDI footprints are 25 m diameter and sampled every 600 m across track and 60 m along track from variable ascending and descending orbits. A 30 m intact forest map was derived from the global 2010 percent tree cover and 2000-2021 forest cover loss products. The province was stratified into 5 AGB strata considering the GEDI good quality forest AGB values, and 3 sites with low, medium, and high sill values were selected from each stratum by inspection of the directional semivariogram. The average of the good quality forest GEDI footprint AGB values over the 31 months, i.e., the overall mean GEDI AGB (OMGA), was the target parameter used to assess how many months of GEDI data are needed to characterize REDD+ project scale AGB. At each site, the OMGA was compared with the mean AGB derived considering different combinations of 1, 2, 3 ... n GEDI orbits, where n is the total number of GEDI orbits. The number of orbits when all the mean AGB estimates fell within ±p% of the OMGA was indicative of the minimum number of orbits required to reliably characterize the site AGB to within ±p%. Of the 15 sites considered the maximum period of GEDI data needed to characterize forest AGB within ±5%, ±10%, and ±20% were 2.1, 1.1, and 0.5 years, respectively. Sites with high AGB spatial variation needed more orbits of GEDI data to characterize the forest AGB. The results of this study are illustrated and the implications for GEDI based tropical forest AGB characterization are discussed.