@article{fdi:010079311,
  title = {{A}irborne {L}idar sampling pivotal for accurate regional {AGB} predictions from multispectral images in forest-savanna landscapes},
  author = {{S}agang, {L}. {T}. and {P}loton, {P}ierre and {S}onke, {B}. and {P}oilve, {H}. and {C}outeron, {P}ierre and {B}arbier, {N}icolas},
  editor = {},
  language = {{ENG}},
  abstract = {{P}recise accounting of carbon stocks and fluxes in tropical vegetation using remote sensing approaches remains a challenging exercise, as both signal saturation and ground sampling limitations contribute to inaccurate extrapolations. {A}irborne {L}i{DAR} {S}canning ({ALS}) data can be used as an intermediate level to radically increase sampling and enhance model calibration. {H}ere we tested the potential of using {ALS} data for upscaling vegetation aboveground biomass ({AGB}) from field plots to a forest-savanna transitional landscape in the {G}uineo-{C}ongolian region in {C}ameroon, using either a design-based approach or a model-based approach leveraging multispectral satellite imagery. {T}wo sets of reference data were used: (1) {AGB} values collected from 62 0.16-ha plots distributed both in forests and savannas; and (2) an {AGB} map generated form {ALS} data. {I}n the model-based approach, we trained {R}andom {F}orest models using predictors from recent sensors of varying spectral and spatial resolutions ({S}pot 6/7, {L}andsat 8, and {S}entinel 2), along with biophysical predictors derived after pre-processing into the {O}verland processing chain, following a forward variable selection procedure with a spatial 4-folds cross validation. {T}he models calibrated with field plots lead to a systematic overestimation in {AGB} density estimates and a root mean squared prediction error ({RMSPE}) of up to 65 {M}g.ha(-1) (90%), whereas calibration with {ALS} lead to low bias and a drop of similar to 30% in {RMSPE} (down to 43 {M}g.ha(-1), 58%) with little effect of the satellite sensor used. {D}ecomposing bias along the {AGB} density range, we show that multispectral images can (in some specific cases) be used for unbiased prediction at landscape scale on the basis of {ALS}-calibrated statistical models. {H}owever, our results also confirm that, whatever the spectral indices used and attention paid to sensor quality and pre-processing, the signal is not sufficient to warrant accurate pixelwise predictions, because of large relative {RMSPE}, especially above (200-250 t/ha). {T}he design-based approach, for which average {AGB} density values were attributed to mapped land cover classes, proved to be a simple and reliable alternative (for landscape to region level estimations), when trained with dense {ALS} samples.},
  keywords = {forest-savanna mosaics ; {AGB} ; {A}irborne {L}i{DAR} ; satellite ; upscaling ; model-based ; design-based ; bias ; {CAMEROUN} ; {ZONE} {GUINEENNE} ; {ZONE} {EQUATORIALE} ; {SANAGA} {COURS} {D}'{EAU} {REGION}},
  booktitle = {},
  journal = {{R}emote {S}ensing},
  volume = {12},
  numero = {10},
  pages = {art. 1637 [20 ]},
  year = {2020},
  DOI = {10.3390/rs12101637},
  URL = {https://www.documentation.ird.fr/hor/fdi:010079311},
}