@article{fdi:010080544,
  title = {{S}urface and sub-surface flow estimation at high temporal resolution using deep neural networks},
  author = {{A}bbas, {A}. and {B}aek, {S}. and {K}im, {M}. and {L}igaray, {M}. and {R}ibolzi, {O}livier and {S}ilvera, {N}orbert and {M}in, {J}. {H}. and {B}oithias, {L}aurie and {C}ho, {K}. {H}.},
  editor = {},
  language = {{ENG}},
  abstract = {{R}ecent intensification in climate change have resulted in the rise of hydrological extreme events. {T}his demands modeling of hydrological processes at high temporal resolution to better understand flow patterns in catchments. {T}o model surface and sub-surface flows in a catchment we utilized a physically based model called {H}ydrological {S}imulated {P}rogram-{FORTRAN} and two deep learning-based models. {O}ne deep learning model consisted of only one long short-term memory (simple {LSTM}), whereas the other model simulated processes in each hydrological response unit ({HRU}) by defining one separate {LSTM} for each {HRU} ({HRU}-based {LSTM}). {T}he models use environmental time-series data and two-dimensional spatial data to predict surface and sub-surface flows at 6-minute time step simultaneously. {W}e tested our models in a tropical humid headwater catchment in northern {L}ao {PDR} and compared their performances. {O}ur results showed that the simple {LSTM} model outperformed the other models on surface runoff prediction with the lowest {MSE} (7.4e - 5 m(3 )s(-1)), whereas {HRU}-based {LSTM} model better predicted patterns and slopes in sub-surface flow in comparison with the other models by having the smallest {MSE} value (3.2e - 4 m(3 )s(-1)). {T}his study demonstrated the performance of a deep learning model when simulating hydrological cycle with high temporal resolution.},
  keywords = {{D}eep learning model ; {L}ong short-term memory ({LSTM}) ; {S}ub-surface flow ; {S}urface runoff ; {H}ydrological {S}imulated {P}rogram-{FORTRAN} ; {LAOS}},
  booktitle = {},
  journal = {{J}ournal of {H}ydrology},
  volume = {590},
  numero = {},
  pages = {125370 [14 p.]},
  ISSN = {0022-1694},
  year = {2020},
  DOI = {10.1016/j.jhydrol.2020.125370},
  URL = {https://www.documentation.ird.fr/hor/fdi:010080544},
}