Effect of spatial sampling from European flux towers for estimating carbon and water fluxes with artificial neural networks

Empirical modeling approaches are frequently used to upscale local eddy covariance observations of carbon, water, and energy fluxes to regional and global scales. The predictive capacity of such models largely depends on the data used for parameterization and identification of input-output relationships, while prediction for conditions outside the training domain is generally uncertain. In this work, artificial neural networks (ANNs) were used for the prediction of gross primary production (GPP) and latent heat flux (LE) on local and European scales with the aim to assess the portion of uncertainties in extrapolation due to sample selection. ANNs were found to be a useful tool for GPP and LE prediction, in particular for extrapolation in time (mean absolute error MAE for GPP between 0.53 and 1.56 gC m−2 d−1). Extrapolation in space in similar climatic and vegetation conditions also gave good results (GPP MAE 0.7–1.41 gC m−2 d−1), while extrapolation in areas with different seasonal cycles and controlling factors (e.g., the tropical regions) showed noticeably higher errors (GPP MAE 0.8–2.09 gC m−2 d−1). The distribution and the number of sites used for ANN training had a remarkable effect on prediction uncertainty in both, regional GPP and LE budgets and their interannual variability. Results obtained show that for ANN upscaling for continents with relatively small networks of sites, the error due to the sampling can be large and needs to be considered and quantified. The analysis of the spatial variability of the uncertainty helped to identify the meteorological drivers driving the uncertainty.

[1]  Michele Scardi,et al.  Developing an empirical model of phytoplankton primary production: a neural network case study , 1999 .

[2]  Chandra Giri,et al.  A comparative analysis of the Global Land Cover 2000 and MODIS land cover data sets , 2005 .

[3]  Heekuck Oh,et al.  Neural Networks for Pattern Recognition , 1993, Adv. Comput..

[4]  Dario Papale,et al.  Database Maintenance, Data Sharing Policy, Collaboration , 2012 .

[5]  A-Xing Zhu,et al.  Developing a continental-scale measure of gross primary production by combining MODIS and AmeriFlux data through Support Vector Machine approach , 2007 .

[6]  Andrew E. Suyker,et al.  Data-driven diagnostics of terrestrial carbon dynamics over North America , 2014 .

[7]  P. Cox,et al.  Observing terrestrial ecosystems and the carbon cycle from space , 2015, Global change biology.

[8]  J. Nash,et al.  River flow forecasting through conceptual models part I — A discussion of principles☆ , 1970 .

[9]  S. Seneviratne,et al.  Recent decline in the global land evapotranspiration trend due to limited moisture supply , 2010, Nature.

[10]  W. Oechel,et al.  A continuous measure of gross primary production for the conterminous United States derived from MODIS and AmeriFlux data , 2010, Remote Sensing of Environment.

[11]  S. Running,et al.  Global Terrestrial Gross and Net Primary Productivity from the Earth Observing System , 2000 .

[12]  Damiano Gianelle,et al.  Seasonal variation of photosynthetic model parameters and leaf area index from global Fluxnet eddy covariance data , 2011 .

[13]  I. C. Prentice,et al.  A dynamic global vegetation model for studies of the coupled atmosphere‐biosphere system , 2005 .

[14]  I. C. Prentice,et al.  Evaluation of ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic global vegetation model , 2003 .

[15]  B. Poulter,et al.  Evaluating the potential of large-scale simulations to predict carbon fluxes of terrestrial ecosystems over a European Eddy Covariance network , 2013 .

[16]  Dario Papale,et al.  Assessing and improving the representativeness of monitoring networks: The European flux tower network example , 2011 .

[17]  Damiano Gianelle,et al.  Vegetation-specific model parameters are not required for estimating gross primary production , 2014 .

[18]  R. Valentini,et al.  A new assessment of European forests carbon exchanges by eddy fluxes and artificial neural network spatialization , 2003 .

[19]  T. Vesala,et al.  Towards a standardized processing of Net Ecosystem Exchange measured with eddy covariance technique: algorithms and uncertainty estimation , 2006 .

[20]  T. Vesala,et al.  Reduction of ecosystem productivity and respiration during the European summer 2003 climate anomaly: a joint flux tower, remote sensing and modelling analysis , 2007 .

[21]  Markus Reichstein,et al.  Analyzing the causes and spatial pattern of the European 2003 carbon flux anomaly using seven models , 2007 .

[22]  Dario Papale,et al.  A full greenhouse gases budget of Africa: synthesis, uncertainties, and vulnerabilities , 2014 .

[23]  Nuno Carvalhais,et al.  Deciphering the components of regional net ecosystem fluxes following a bottom-up approach for the Iberian Peninsula , 2010 .

[24]  T. Vesala,et al.  On the separation of net ecosystem exchange into assimilation and ecosystem respiration: review and improved algorithm , 2005 .

[25]  F. Woodward,et al.  Terrestrial Gross Carbon Dioxide Uptake: Global Distribution and Covariation with Climate , 2010, Science.

[26]  Jerry Y. Pan,et al.  Intercomparison of MODIS albedo retrievals and in situ measurements across the global FLUXNET network , 2012 .

[27]  P. Ciais,et al.  Spatiotemporal patterns of terrestrial gross primary production: A review , 2015 .

[28]  Markus Reichstein,et al.  Mean annual GPP of Europe derived from its water balance , 2007 .

[29]  S. Wofsy,et al.  Modeling gross primary production of temperate deciduous broadleaf forest using satellite images and climate data , 2004 .

[30]  A. Arneth,et al.  Global patterns of land-atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations , 2011 .

[31]  Christopher M. Bishop,et al.  Neural networks for pattern recognition , 1995 .

[32]  J. Monteith.  SOLAR RADIATION AND PRODUCTIVITY IN TROPICAL ECOSYSTEMS , 1972 .

[33]  A. Bondeau,et al.  Towards global empirical upscaling of FLUXNET eddy covariance observations: validation of a model tree ensemble approach using a biosphere model , 2009 .

[34]  Markus Reichstein,et al.  Temporal and among‐site variability of inherent water use efficiency at the ecosystem level , 2009 .

[35]  I. C. Prentice,et al.  BIOME3: An equilibrium terrestrial biosphere model based on ecophysiological constraints, resource availability, and competition among plant functional types , 1996 .

[36]  Markus Reichstein,et al.  Uncertainties of modeling gross primary productivity over Europe: A systematic study on the effects of using different drivers and terrestrial biosphere models , 2007 .

[37]  Dario Papale,et al.  Eddy Covariance: A Practical Guide to Measurement and Data Analysis , 2012 .

[38]  Hank A. Margolis,et al.  Attributing uncertainties in simulated biospheric carbon fluxes to different error sources , 2011 .

[39]  D. Baldocchi.  ‘Breathing’ of the terrestrial biosphere: lessons learned from a global network of carbon dioxide flux measurement systems , 2008 .

[40]  Eric R. Ziegel,et al.  The Elements of Statistical Learning , 2003, Technometrics.

[41]  Piero Toscano,et al.  Remote sensing of annual terrestrial gross primary productivity from MODIS: an assessment using the FLUXNET La Thuile data set , 2013 .