@article{fdi:010078029,
  title = {{R}econstructing climatic modes of variability from proxy records using {C}lim{I}nd{R}ec version 1.0},
  author = {{M}ichel, {S}. and {S}wingedouw, {D}. and {C}havent, {M}. and {O}rtega, {P}. and {M}ignot, {J}uliette and {K}hodri, {M}yriam},
  editor = {},
  language = {{ENG}},
  abstract = {{M}odes of climate variability strongly impact our climate and thus human society. {N}evertheless, the statistical properties of these modes remain poorly known due to the short time frame of instrumental measurements. {R}econstructing these modes further back in time using statistical learning methods applied to proxy records is useful for improving our understanding of their behaviour. {F}or doing so, several statistical methods exist, among which principal component regression is one of the most widely used in paleoclimatology. {H}ere, we provide the software {C}lim{I}nd{R}ec to the climate community; it is based on four regression methods (principal component regression, {PCR}; partial least squares, {PLS}; elastic net, {E}net; random forest, {RF}) and cross-validation ({CV}) algorithms, and enables the systematic reconstruction of a given climate index. {A} prerequisite is that there are proxy records in the database that overlap in time with its observed variations. {T}he relative efficiency of the methods can vary, according to the statistical properties of the mode and the proxy records used. {H}ere, we assess the sensitivity to the reconstruction technique. {C}lim{I}nd{R}ec is modular as it allows different inputs like the proxy database or the regression method. {A}s an example, it is here applied to the reconstruction of the {N}orth {A}tlantic {O}scillation by using the {PAGES} 2k database. {I}n order to identify the most reliable reconstruction among those given by the different methods, we use the modularity of {C}lim{I}nd{R}ec to investigate the sensitivity of the methodological setup to other properties such as the number and the nature of the proxy records used as predictors or the targeted reconstruction period. {W}e obtain the best reconstruction of the {N}orth {A}tlantic {O}scillation ({NAO}) using the random forest approach. {I}t shows significant correlation with former reconstructions, but exhibits higher validation scores.},
  keywords = {},
  booktitle = {},
  journal = {{G}eoscientific {M}odel {D}evelopment},
  volume = {13},
  numero = {2},
  pages = {841--858},
  ISSN = {1991-959{X}},
  year = {2020},
  DOI = {10.5194/gmd-13-841-2020},
  URL = {https://www.documentation.ird.fr/hor/fdi:010078029},
}