@article{fdi:010091171,
  title = {{P}re-processing satellite rainfall products improves hydrological simulations with machine learning},
  author = {{B}oulmaiz, {T}. and {H}afsi, {R}. and {G}uermoui, {M}. and {B}outaghane, {H}. and {A}bida, {H}. and {S}aber, {M}. and {K}antoush, {S}. {A}. and {F}erkous, {K}. and {T}ramblay, {Y}ves},
  editor = {},
  language = {{ENG}},
  abstract = {{A} new pre-processing methodology for gridded {S}atellite {P}recipitation {P}roducts ({SPP}s) is developed to improve the performance of {M}achine {L}earning ({ML}) algorithms for runoff prediction. {T}he developed approach was applied to capture the rainfall patterns, and to select relevant input data. {T}his approach was tested using the {F}eed{F}orward {N}eural {N}etwork ({FFNN}) and the {E}xtreme {L}earning {M}achine ({ELM}) given their flexibility and ability in hydrological modelling. {T}he methodology was tested in a semiarid transboundary watershed located in {N}orth {A}frica ({A}lgeria, {T}unisia) with the {I}ntegrated {M}ultisatellit{E} {R}etrievals for {G}lobal {P}recipitation {M}easurement ({GPMIMERG}) and the {P}recipitation {E}stimation from {R}emotely {S}ensed {I}nformation using {A}rtificial {N}eural {N}etworks ({PERSIANN}) products. {T}he results demonstrate the effectiveness of the proposed approach using all employed {SPP}s. {I}n terms of {N}ash-{S}utcliffe efficiency, the suggested pre-processing technique improved the prediction ability of {FFNN} by 13%, and of {ELM} by 15%, which highlights how pre-processing techniques significantly enhance {ML} models with {SPP} data.},
  keywords = {machine learning ; {IMERG} ; {PERSIANN} ; pre-processing ; rainfall ; runoff ; satellite precipitation products ; {ALGERIE} ; {TUNISIE}},
  booktitle = {},
  journal = {{H}ydrological {S}ciences {J}ournal},
  volume = {69},
  numero = {10},
  pages = {1356--1370},
  ISSN = {0262-6667},
  year = {2024},
  DOI = {10.1080/02626667.2024.2378108},
  URL = {https://www.documentation.ird.fr/hor/fdi:010091171},
}