Capturing the Influence of ENSO on Land Surface Variables for Tropical South America

Abstract

<p>The response of tropical vegetation to El Ni&#241;o Southern Oscillation (ENSO) is considered a main driver of global annual atmospheric CO2 concentrations at inter-annual time scales. ENSO warm and cold phases, El Ni&#241;o and La Ni&#241;a respectively, cause contrasting climatic conditions along tropical South America. While some regions experience wetter conditions during El Ni&#241;o, such as&#160; the Pacific coast, others regions such as the Amazon are exposed to warmer and drier climates. Besides this spatial variation, the biospheric response also differs between ENSO type and intensity, overruling of local conditions and ecosystems types. Due to this complexity, there is a lack of understanding on what ecosystems and regions are systematically affected by ENSO and how biospheric variables respond. Here, we analysed the Northern region of tropical South America covering tropical savannas, forests, and mountainous ecosystems in several countries. To do this, we assessed different land surface (e.g. GPP, NDVI,&#160; FPAR, LST) and climate data streams compiled in the regional Earth System Data Lab (ESDL, https://www.earthsystemdatalab.net/) at 1 km and 10 km pixel size from 2001 to 2015. We applied Isomap, a non-linear dimensionality reduction method in the time domain for high dimensional dynamical systems. Our analysis was constrained to the fourth order continental basins and dominant land cover types. Land use change pixels were disregarded. Further, a comparison of ENSO indexes was conducted among basins. We found that isomap components&#160; are able to capture the biosphere variability related to ENSO in basins that have been historically affected such as Magdalena-Cauca valleys and the Caribbean region. Implementation of non-linear methods increases our understanding of ENSO impacts spatially in regions where events intensity and frequency is increasing, and effective ecosystems management is urgent.</p>

Guido Kraemer
Guido Kraemer
Postdoctoral associate / Earth System Data Science

My research interests include the interactions between society and biosphere. I have been working on the extraction of the global dynamics of ecosystems and society. I have an interest in using machine learning and multivariate statistics to understand the behavior of complex systems.

Miguel D. Mahecha
Miguel D. Mahecha
Professor for Earth System Data Science

Professor