@article{fdi:010085351,
  title = {{M}apping gully erosion variability and susceptibility using remote sensing, multivariate statistical analysis, and machine learning in {S}outh {M}ato {G}rosso, {B}razil},
  author = {{B}ouramtane, {T}. and {H}ilal, {H}. and {R}ezende, {A}. {T}. and {B}ouramtane, {K}. and {B}arbi{\'e}ro, {L}aurent and {A}braham, {S}. and {V}alles, {V}. and {K}acimi, {I}. and {S}anhaji, {H}. and {T}orres-{R}ondon, {L}. and de {C}astro, {D}. {D}. and {S}antos, {J}. {D}. {V}. and {O}uardi, {J}. and {E}l {B}eqqali, {O}. and {K}assou, {N}. and {M}orarech, {M}.},
  editor = {},
  language = {{ENG}},
  abstract = {{I}n {B}razil, the development of gullies constitutes widespread land degradation, especially in the state of {S}outh {M}ato {G}rosso, where fighting against this degradation has become a priority for policy makers. {H}owever, the environmental and anthropogenic factors that promote gully development are multiple, interact, and present a complexity that can vary by locality, making their prediction difficult. {I}n this framework, a database was constructed for the {R}io {I}vinhema basin in the southern part of the state, including 400 georeferenced gullies and 13 geo-environmental descriptors. {M}ultivariate statistical analysis was performed using principal component analysis ({PCA}) to identify the processes controlling the variability in gully development. {S}usceptibility maps were created through four machine learning models: multivariate discriminant analysis ({MDA}), logistic regression ({LR}), classification and regression tree ({CART}), and random forest ({RF}). {T}he predictive performance of the models was analyzed by five evaluation indices: accuracy ({ACC}), sensitivity ({SST}), specificity ({SPF}), precision ({PRC}), and {R}eceiver {O}perating {C}haracteristic curve ({ROC} curve). {T}he results show the existence of two major processes controlling gully erosion. {T}he first is the surface runoff process, which is related to conditions of slightly higher relief and higher rainfall. {T}he second also reflects high surface runoff conditions, but rather related to high drainage density and downslope, close to the river network. {H}uman activity represented by peri-urban areas, construction of small earthen dams, and extensive rotational farming contribute significantly to gully formation. {T}he four machine learning models yielded fairly similar results and validated susceptibility maps ({ROC} curve > 0.8). {H}owever, we noted a better performance of the random forest ({RF}) model (86% and 89.8% for training and test, respectively, with an {ROC} curve value of 0.931). {T}he evaluation of the contribution of the parameters shows that susceptibility to gully erosion is not governed primarily by a single factor, but rather by the interconnection between different factors, mainly elevation, geology, precipitation, and land use.},
  keywords = {gully erosion ; natural hazard ; machine learning ; principal components ; analysis ; random forest ; {BRESIL} ; {MATO} {GROSSO}},
  booktitle = {},
  journal = {{G}eosciences},
  volume = {12},
  numero = {6},
  pages = {235 [25 ]},
  year = {2022},
  DOI = {10.3390/geosciences12060235},
  URL = {https://www.documentation.ird.fr/hor/fdi:010085351},
}