Assessing the Accuracy of GEDI Data for Canopy Height andAboveground Biomass Estimates in Mediterranean Forests

• Autor(a) principal: Dorado-Roda, Iván
• Data de Publicação: 2021
• Outros Autores: Pascual, Adrián, Godinho, Sérgio, Silva, Carlos A., Botequim, Brigite, Rodríguez-Gonzálvez, Pablo, González-Ferreiro, Eduardo, Guerra-Hernández, Juan
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10174/32263; https://doi.org/doi.org/10.3390/rs13122279

Global Ecosystem Dynamics Investigation (GEDI) satellite mission is expanding the spatialbounds and temporal resolution of large-scale mapping applications. Integrating the recent GEDIdata into Airborne Laser Scanning (ALS)-derived estimations represents a global opportunity toupdate and extend forest models based on area based approaches (ABA) considering temporal andspatial dynamics. This study evaluates the effect of combining ALS-based aboveground biomass(AGB) estimates with GEDI-derived models by using temporally coincident datasets. A gradient offorest ecosystems, distributed through 21,766 km2 in the province of Badajoz (Spain), with differentspecies and structural complexity, was used to: (i) assess the accuracy of GEDI canopy height in fiveMediterranean Ecosystems and (ii) develop GEDI-based AGB models when using ALS-derived AGBestimates at GEDI footprint level. In terms of Pearson’s correlation (r) and rRMSE, the agreementbetween ALS and GEDI statistics on canopy height was stronger in the denser and homogeneousconiferous forest of P. pinaster and P. pinea than in sparse Quercus-dominated forests. The GEDI-derived AGB models using relative height and vertical canopy metrics yielded a model efficiency(Mef) ranging from 0.31 to 0.46, with a RMSE ranging from 14.13 to 32.16 Mg/ha and rRMSE from38.17 to 84.74%, at GEDI footprint level by forest type. The impact of forest structure confirmedprevious studies achievements, since GEDI data showed higher uncertainty in highly multilayeredforests. In general, GEDI-derived models (GEDI-like Level4A) underestimated AGB over lower andhigher ALS-derived AGB intervals. The proposed models could also be used to monitor biomassstocks at large-scale by using GEDI footprint level in Mediterranean areas, especially in remote andhard-to-reach areas for forest inventory. The findings from this study serve to provide an initialevaluation of GEDI data for estimating AGB in Mediterranean forest.