FOREST ECOLOGY AND MANAGEMENT | 卷:491 |
Beyond trees: Mapping total aboveground biomass density in the Brazilian savanna using high-density UAV-lidar data | |
Article | |
Teixeira da Costa, Maira Beatriz1  Silva, Carlos Alberto2,3  Broadbent, Eben North4  Leite, Rodrigo Vieira5  Mohan, Midhun6  Liesenberg, Veraldo7  Stoddart, Jaz8  do Amaral, Cibele Hummel5  Alves de Almeida, Danilo Roberti9  da Silva, Anne Laura10  Goya, Lucas Ruggeri Re Y.10  Cordeiro, Victor Almeida10  Rex, Franciel11  Hirsch, Andre10  Marcatti, Gustavo Eduardo10  Cardil, Adrian12,13,14  Furtado de Mendonca, Bruno Araujo15  Hamamura, Caio16  Dalla Corte, Ana Paula11  Trondoli Matricardi, Eraldo Aparecido1  Hudak, Andrew T.17  Almeyda Zambrano, Angelica Maria18  Valbuena, Ruben8  de Faria, Bruno Lopes19,20  Silva Junior, Celso H. L.21,22  Aragao, Luiz21  Ferreira, Manuel Eduardo23  Liang, Jingjing24  Chaves e Carvalho, Samuel de Padua25  Klauberg, Carine10  | |
[1] Univ Brasilia, Dept Forestry, Campus Darcy Ribeiro, BR-70910900 Brasilia, DF, Brazil | |
[2] Univ Florida, Sch Forest Fisheries & Geomat Sci, POB 110410, Gainesville, FL 32611 USA | |
[3] Univ Maryland, Dept Geog Sci, College Pk, MD 20740 USA | |
[4] Univ Florida, Sch Forest Fisheries & Geomat Sci, Spatial Ecol & Conservat SPEC Lab, Gainesville, FL 32611 USA | |
[5] Fed Univ Vicosa UFV, Dept Forest Engn, Vicosa, MG, Brazil | |
[6] Univ Calif Berkeley, Dept Geog, Berkeley, CA 94709 USA | |
[7] Santa Catarina State Univ UDESC, Coll Agr & Vet, Dept Forest Engn, Lages, SC, Brazil | |
[8] Bangor Univ, Sch Nat Sci, Bangor LL57 2W, Gwynedd, Wales | |
[9] Univ Sao Paulo USP ESALQ, Luiz de Queiroz Coll Agr, Dept Forest Sci, Piracicaba, SP, Brazil | |
[10] Fed Univ Sao Joao del Rei UFSJ, Sete Lagoas, BR-35701970 Sao Joao Del Rei, MG, Brazil | |
[11] Fed Univ Parana UFPR, Dept Forest Engn, Curitiba, Parana, Brazil | |
[12] Technosylva Inc, La Jolla, CA USA | |
[13] Univ Lleida, Dept Crop & Forest Sci, Lleida, Spain | |
[14] Joint Res Unit CTFC AGROTECNIO, Solsona, Spain | |
[15] Univ Fed Rural Rio de Janeiro, Dept Silvicultura, Rua Floresta, BR-23897005 Seropedica, RJ, Brazil | |
[16] Fed Inst Educ Sci & Technol Sao Paulo, BR-11533160 Sao Paulo, SP, Brazil | |
[17] US Forest Serv, USDA, Rocky Mt Res Stn, 1221 South Main St, Moscow, ID 83843 USA | |
[18] Univ Florida, Ctr Latin Amer Studies, Spatial Ecol & Conservat SPEC Lab, Gainesville, FL 32611 USA | |
[19] Fed Inst Technol North Minas Gerais IFNMG, BR-39100000 Diamantina, MG, Brazil | |
[20] Fed Univ Vales Jequitinhonha & Mucuri, Dept Forest Sci, UFVJM, Campus JK, Diamantina, MG, Brazil | |
[21] Natl Inst Space Res INPE, Earth Observat & Geoinformat Div, Av Astronautas 1758, BR-12227010 Sao Jose Dos Campos, SP, Brazil | |
[22] Univ Fed Goias, Image Proc & GIS Lab LAPIG, BR-74001970 Goiania, Go, Brazil | |
[23] Univ Estadual Maranhao UEMA, Dept Engn Agr, BR-65055310 Sao Luis, MA, Brazil | |
[24] Purdue Univ, Dept Forestry & Nat Resources, W Lafayette, IN 47907 USA | |
[25] Univ Fed Mato Grosso, Coll Forestry, Av Fernando Correa da Costa 2367, BR-78060900 Cuiaba, MT, Brazil | |
关键词: Biomass; Vegetation; Tropical savanna; Remote sensing; Cerrado; Mapping; GatorEye; | |
DOI : 10.1016/j.foreco.2021.119155 | |
来源: Elsevier | |
【 摘 要 】
Tropical savanna ecosystems play a major role in the seasonality of the global carbon cycle. However, their ability to store and sequester carbon is uncertain due to combined and intermingling effects of anthropogenic activities and climate change, which impact wildfire regimes and vegetation dynamics. Accurate measurements of tropical savanna vegetation aboveground biomass (AGB) over broad spatial scales are crucial to achieve effective carbon emission mitigation strategies. UAV-lidar is a new remote sensing technology that can enable rapid 3-D mapping of structure and related AGB in tropical savanna ecosystems. This study aimed to assess the capability of high-density UAV-lidar to estimate and map total (tree, shrubs, and surface layers) aboveground biomass density (AGBt) in the Brazilian Savanna (Cerrado). Five ordinary least square regression models estimating AGBt were adjusted using 50 field sample plots (30 m x 30 m). The best model was selected under Akaike Information Criterion, adjusted coefficient of determination (adj.R-2), absolute and relative mot mean square error (RMSE), and used to map AGBt from UAV-lidar data collected over 1,854 ha spanning the three major vegetation formations (forest, savanna, and grassland) in Cerrado. The model using vegetation height and cover was the most effective, with an overall model adj-R-2 of 0.79 and a leave-one-out cross-validated RMSE of 19.11 Mg/ha (33.40%). The uncertainty and errors of our estimations were assessed for each vegetation formation separately, resulting in RMSEs of 27.08 Mg/ha (25.99%) for forests, 17.76 Mg/ha (43.96%) for savannas, and 7.72 Mg/ha (44.92%) for grasslands. These results prove the feasibility and potential of the UAV-lidar technology in Cerrado but also emphasize the need for further developing the estimation of biomass in grasslands, of high importance in the characterization of the global carbon balance and for supporting integrated fire management activities in tropical savanna ecosystems. Our results serve as a benchmark for future studies aiming to generate accurate biomass maps and provide baseline data for efficient management of fire and predicted climate change impacts on tropical savanna ecosystems.
【 授权许可】
Free
【 预 览 】
Files | Size | Format | View |
---|---|---|---|
10_1016_j_foreco_2021_119155.pdf | 13527KB | download |