@article{fdi:010090220,
  title = {{T}he {AMG} model coupled with {R}ock-{E}val ({R}) analysis accurately predicts cropland soil organic carbon dynamics in the {T}uojiang {R}iver {B}asin, {S}outhwest {C}hina},
  author = {{W}ang, {Q}. and {B}arre, {P}. and {B}audin, {F}. and {C}livot, {H}. and {F}erchaud, {F}. and {L}i, {Y}. and {G}ao, {X}. {S}. and {L}e {N}oe, {J}ulia},
  editor = {},
  language = {{ENG}},
  abstract = {{A}ccurate soil organic carbon models are key to understand the mechanisms governing carbon sequestration in soil and to help develop targeted management strategies to carbon budget. {T}he accuracy and reliability of soil organic carbon ({SOC}) models remains strongly limited by incorrect initialization of the conceptual kinetic pools and lack of stringent model evaluation using time-series datasets. {N}otably, due to legacy effects of management and land use change, the traditional spin-up approach for initial allocation of {SOC} among kinetic pools can bring substantial uncertainties in predicting the evolution of {SOC} stocks. {T}he {AMG} model can fulfill these conditions as it is a parsimonious yet accurate {SOC} model using widely-available input data. {I}n this study, we first evaluated the performance of {AMG}v2 before and after optimizing the potential mineralization rate (k0) of {SOC} stock following a leave-one-site-out cross-validation based on 24 long-term field experiments ({LTE}s) in the {S}outhwest of {C}hina. {T}hen, we used {R}ock-{E}val & {REG}; thermal analysis results as input variables in the {PARTYSOC} machine learning model to estimate the initial stable {SOC} fraction ({CS}/{C}0) for the 14 {LTE}s where soil samples were available. {T}he results showed that initializing the {CS}/{C}0 ratio using {PARTYSOC} combined with the optimized k0 further improved the accuracy of model simulations ({R}2 = 0.87, {RMSE} = 0.25, d = 0.90). {C}ombining average measured {CS}/{C}0 and k0 optimization across all 24 {LTE}s also improved the model predictive capability by 25% compared to using default parameterization, thus suggesting promising avenue for upscaling model applications at the regional level where only a few measurement data on {SOC} stability can be available. {I}n conclusion, the new version of the {AMG} model developed in the {T}uojiang {R}iver {B}asin context exhibits excellent performance. {T}his result paves the way for further calibration and validation of the {AMG} model in a wider set of contexts, with the potential to significantly improve confidence in {SOC} predictions in croplands over regional scales.},
  keywords = {{C}alibration ; {L}ong-term experiment sites ; {PARTYSOC} method ; {SOC} stable ; pool initialization ; {P}otential mineralization rate ; {CHINE}},
  booktitle = {},
  journal = {{J}ournal of {E}nvironmental {M}anagement},
  volume = {345},
  numero = {},
  pages = {118850 [11 ]},
  ISSN = {0301-4797},
  year = {2023},
  DOI = {10.1016/j.jenvman.2023.118850},
  URL = {https://www.documentation.ird.fr/hor/fdi:010090220},
}