@article{fdi:010095422,
  title = {{CHIRTS} gridded air temperature downscaling integrating {MODIS} land surface temperature estimates in machine-learning models},
  author = {{U}scamayta-{F}errano, {E}. and {S}atg{\'e}, {F}r{\'e}d{\'e}ric and {P}illco-{Z}olá, {R}. and {R}oig, {H}. and {T}ola-{A}guilar, {D}. and {P}erez-{F}lores, {M}. and {B}ustillos, {L}. and {R}akotomandrindra, {F}. {P}. {M}. and {R}abefitia, {Z}. and {C}arri{\`e}re, {S}. {D}.},
  editor = {},
  language = {{ENG}},
  abstract = {{D}ue to its sensitivity to topographic and land use land cover features, air temperature (maximum, minimum, and mean-{T}x, {T}n, and {T}mean) is extremely variable in space and time. {T}he sparse and unevenly distributed meteorological stations observed across remote regions cannot monitor such variability. {F}reely available, gridded temperature datasets ({T}-datasets) are positioned as an opportunity to overcome this issue. {S}till, their coarse spatial resolution (i.e., >= 5 km) does not allow for the observation of air temperature variations on a fine spatial scale. {I}n this context, a set of variables that have a close relationship with daily air temperature ({MODIS} maximum, minimum, and mean {L}and {S}urface {T}emperature-{LST}x, {LST}n, and {LST}mean; {MODIS} {NDVI}; {SRTM} topographic features-elevation, slope, and aspect) are integrated in three regression machine-learning models ({R}andom {F}orest-{RF}, e{X}treme {G}radient {B}oosting-{XGB}, {M}ultiple {L}inear {R}egression-{MLR}) to propose a {T}-dataset estimates ({T}x, {T}n, and {T}mean) spatial resolution downscaling framework. {T}he approach consists of two main steps: firstly, the machine-learning models are trained at the native 5 km spatial resolution of the studied {T}-dataset (i.e., {CHIRTS}); secondly, the application of the trained machine-learning models at a 1 km spatial resolution to downscale {CHIRTS} from 5 km to 1 km. {T}he results show that the method not only improves the spatial resolution of the {CHIRTS} dataset, but also its accuracy, with higher improvements for {T}n than for {T}x and {T}mean. {A}mong the considered models, {RF} performs the best, with an {R}2, {RMSE}, and {MAE} improvement of 2.6% (0%), 47.1% (6.1%), and 55.3% (7%) for {T}n ({T}x). {T}hese results will support air temperature monitoring and related extreme events such as heat and cold waves, which are of prime importance in the actual climate change context.},
  keywords = {{CHIRTS} ; downscaling ; temperature ; machine-learning ; {M}adagascar ; {MADAGASCAR}},
  booktitle = {},
  journal = {{A}tmosphere},
  volume = {16},
  numero = {10},
  pages = {1188 [21 p.]},
  year = {2025},
  DOI = {10.3390/atmos16101188},
  URL = {https://www.documentation.ird.fr/hor/fdi:010095422},
}