INVESTIGADORES
DIAZ Leandro Baltasar
artículos
Título:
Evapotranspiration trends and variability in southeastern South America: the roles of land‐cover change and precipitation variability
Autor/es:
RUSCICA, ROMINA C.; SÖRENSSON, ANNA A.; DIAZ, LEANDRO B.; VERA, CAROLINA; CASTRO, ALINE; PAPASTEFANOU, PHILLIP; RAMMIG, ANJA; REZENDE, LUIZ F.C.; SAKSCHEWSKI, BORIS; THONICKE, KIRSTEN; VIOVY, NICOLAS; RANDOW, CELSO
Revista:
INTERNATIONAL JOURNAL OF CLIMATOLOGY
Editorial:
JOHN WILEY & SONS LTD
Referencias:
Año: 2021 vol. 42 p. 2019 - 2038
ISSN:
0899-8418
Resumen:
Southeastern South America is subject to considerable precipitation variability on seasonal to decadal timescales and has undergone very heavy land-cover changes since the middle of the past century. The influence of local land-cover change and precipitation as drivers of regional evapotranspiration long-term trends and variability remains largely unknown in the region. Here, ensembles of stand-alone Dynamic Global Vegetation Models with different atmospheric forcings are used to disentangle the influence of those two drivers on austral summer evapotranspiration from 1950 to 2010. This paper examines the influence of both the ENSO and the dipole-like first-mode of southeastern South American precipitation variability (EOF1) on regional evapotranspiration. We found that in the lower La Plata Basin, evapotranspiration was driven by precipitation variability and showed a positive summer trend. Moreover, the region showed marked seasonal anomalies during El Niño and La Niña summers but mainly during EOF1 phases. On the contrary, in the upper La Plata Basin, land-cover changes forced the negative summer evapotranspiration trend and particularly reduced the summer anomalies of the late 1990s, a period of ENSO and EOF1-positive phases. In the South Atlantic Convergence Zone region, the high evapotranspiration uncertainty across ensemble members impeded finding robust results, which highlights the importance of using multiple DGVMs and atmospheric forcings instead of relying on single model/forcing results.