bei Google ScholarTitelangaben
Sippel, Sebastian ; Lange, Holger ; Mahecha, Miguel D. ; Hauhs, Michael ; Bodesheim, Paul ; Kaminski, Thomas ; Gans, Fabian ; Rosso, Osvaldo A.:
Diagnosing the dynamics of observed and simulated ecosystem gross primary productivity with time causal information theory quantifiers.
In: PLoS One.
Bd. 11
(2016)
Heft 10
.
- e0164960.
ISSN 1932-6203
DOI: https://doi.org/10.1371/journal.pone.0164960
Abstract
Model-data comparison in the environmental sciences requires metrics that quantify time series dynamics and structure, and are robust to noise in observational data. In this context, we demonstrate the usefulness of time causal Information Theory quantifiers as robust and efficient data-analytical and model benchmarking tools. As anexample of their application, we investigate the dynamics of Gross Primary Productivity (GPP) observations on one hand, and simulations using an ensemble of climate and carbon cycle models on the other, at site, continental and global scales and for different temporal resolutions.GPP is a key variable in terrestrial ecosystems, quantifying the gross carbon uptake of plants. Changes in the short- and long-term dynamics of GPP can be induced by landuse change, extreme events, and changing climate. However, the dynamics, patterns and magnitudes of GPP time series, both observed and simulated, on various temporal and spatial scales remain often uncertain.In this context, we show that Information Theory (or complexity) quantifiers serve as useful diagnostics for evaluating model structure and dynamics. At continental scale, we compare GPP time series simulated with two models (JSBACH and LPJmL) andobtained from an observations-based product (MTE) and demonstrate that modelevaluation based on information theory quantifiers yields qualitatively different resultscompared to traditional model evaluation metrics. This indicates that a good model performance in terms of absolute or relative error does not necessarily imply that it also captures the dynamical structure of the observations well, as documented by these complexity quantifiers. We show that Information Theory quantifiers are able to diagnose model structure based only on dynamical aspects: The measures are largely insensitive to climatic scenarios, land use and atmospheric gas concentrations used to drive them, but clearly separate the structure of 13 different land models taken from the CMIP5 archive and the observational product.In conclusion, quantifiers of this kind provide data-analytical tools that distinguish different types of natural processes and are thus highly suitable for environmentalscience applications such as model structural diagnostics.