@article{fdi:010076139,
  title = {{H}igh resolution mapping of inundation area in the {A}mazon basin from a combination of {L}-band passive microwave, optical and radar datasets},
  author = {{P}arrens, {M}. and {A}l {B}itar, {A}. and {F}rappart, {F}. and {P}aiva, {R}. and {W}ongchuig, {S}. and {P}apa, {F}abrice and {Y}amasaki, {D}. and {K}err, {Y}.},
  editor = {},
  language = {{ENG}},
  abstract = {{I}n this paper, we present a methodology to map inland water in tropical areas under dense vegetation at high spatial and temporal resolution using multi-source remote sensing data. {A} new inundation product ({SWAF}-{HR}) is presented. {I}t is characterized by a high spatial resolution (30', 1 km) and high temporal resolution (3 days). {T}he {SWAF}-{HR} product is estimated over the {A}mazon basin for the 2010-2016 period. {I}t is based on a downscaling procedure and the synergistic use of: (1) water surface fraction at coarse spatial resolution from an {L}-band passive microwave sensor ({S}oil {M}oisture and {O}cean {S}alinity - {SMOS}), (2) {G}lobal {S}urface {W}ater {O}ccurrence from {L}andsat ({GSWO}) and (3) the {D}igital {E}levation {M}odel ({DEM}) {M}ulti-{E}rror-{R}emoved-{I}mproved-{T}errain ({MERIT}) based on the {S}huttle {R}adar {T}opography {M}ission ({SRTM}). {T}hanks to the high capability of {L}-band microwave emission to reveal surface water under all-weather conditions and beneath the vegetation, the inundated area extent estimated by the {SWAP}-{HR} product is always larger than {GSWO} estimates obtained by the optical sensor ({L}andsat). {SWAF}-{HR} data is compared to {ESA} {CCI} and {IGBP} land covers, two {SAR} images and flooded areas over the {P}urus basin computed by the {MGB}-{IPH} model simulation. {T}he results show the coherence of spatial and temporal dynamics of the {SWAF}-{HR} data. {W}e show that the flooded area of the {B}ranco {R}iver floodplain in {R}oraima ({B}razil) varies from 0.2 x 10(4) to 2.7 x 10(4) km(2) whereas the extent of the {B}olivian floodplain ({L}lanos de {M}oxos) inundation ranges between 0.8 x 10(4) and 8.1 x 10(4) km(2) during 2010-2016. {T}he flooded area in the {B}ranco floodplain gradually decreased from 2010 to 2015 but in 2016, the flooded area has increased during the rainy season. {D}uring 2010-2016, the minimum of the inundated surface extent was reached during 2015-2016 reflecting to a drought event related to {ENSO}. {T}he most important uncertainties of the {DEM} are located over tropical areas but this information is essential in the downscaling procedure. {T}herefore, we investigate the impact of the choice of the {DEM} for the downscaling procedure. {I}t is found that the choice of the {DEM} introduces 5% of error in the instantaneous water surface extent estimate but can reach up to 10% in the flood probability estimations over seven years. {T}his new {SWAF}-{HR} product will be helpful for the understanding of the water, carbon and biogeochemical cycles of the {A}mazon.},
  keywords = {{A}mazon ; {I}nundation ; {W}ater surface extent ; {SMOS} ; {SWOT} ; {H}ydrology ; {MERIT} ; {DEM} ; {GSWO} ; {BRESIL} ; {GUYANA} ; {AMAZONE} {BASSIN}},
  booktitle = {},
  journal = {{I}nternational {J}ournal of {A}pplied {E}arth {O}bservation and {G}eoinformation},
  volume = {81},
  numero = {},
  pages = {58--71},
  ISSN = {0303-2434},
  year = {2019},
  DOI = {10.1016/j.jag.2019.04.011},
  URL = {https://www.documentation.ird.fr/hor/fdi:010076139},
}