Comprehensive Evaluation of Machine Learning Techniques for Estimating the Responses of Carbon Fluxes to Climatic Forces in Different Terrestrial Ecosystems

Accurately estimating the carbon budgets in terrestrial ecosystems ranging from flux towers to regional or global scales is particularly crucial for diagnosing past and future climate change. This research investigated the feasibility of two comparatively advanced machine learning approaches, namely adaptive neuro-fuzzy inference system (ANFIS) and extreme learning machine (ELM), for reproducing terrestrial carbon fluxes in five different types of ecosystems. Traditional artificial neural network (ANN) and support vector machine (SVM) models were also utilized as reliable benchmarks to measure the generalization ability of these models according to the following statistical metrics: coefficient of determination (R2), index of agreement (IA), root mean square error (RMSE), and mean absolute error (MAE). In addition, we attempted to explore the responses of all methods to their corresponding intrinsic parameters in terms of the generalization performance. It was found that both the newly proposed ELM and ANFIS models achieved highly satisfactory estimates and were comparable to the ANN and SVM models. The modeling ability of each approach depended upon their respective internal parameters. For example, the SVM model with the radial basis kernel function produced the most accurate estimates and performed substantially better than the SVM models with the polynomial and sigmoid functions. Furthermore, a remarkable difference was found in the estimated accuracy among different carbon fluxes. Specifically, in the forest ecosystem (CA-Obs site), the optimal ANN model obtained slightly higher performance for gross primary productivity, with R2 = 0.9622, IA = 0.9836, RMSE = 0.6548 g C m−2 day−1, and MAE = 0.4220 g C m−2 day−1, compared with, respectively, 0.9554, 0.9845, 0.4280 g C m−2 day−1, and 0.2944 g C m−2 day−1 for ecosystem respiration and 0.8292, 0.9306, 0.6165 g C m−2 day−1, and 0.4407 g C m−2 day−1 for net ecosystem exchange. According to the findings in this study, we concluded that the proposed ELM and ANFIS models can be effectively employed for estimating terrestrial carbon fluxes.

[1]  T. Black,et al.  Impact of Nitrogen Fertilization on Forest Carbon Sequestration and Water Loss in a Chronosequence of Three Douglas-Fir Stands in the Pacific Northwest , 2015 .

[2]  Paresh Chandra Deka,et al.  Support vector machine applications in the field of hydrology: A review , 2014, Appl. Soft Comput..

[3]  Holger R. Maier,et al.  State of the Art of Artificial Neural Networks in Geotechnical Engineering , 2008 .

[4]  Atul K. Jain,et al.  Toward “optimal” integration of terrestrial biosphere models , 2015 .

[5]  R. Deo,et al.  Stream-flow forecasting using extreme learning machines: a case study in a semi-arid region in Iraq , 2016 .

[6]  Sungwon Kim,et al.  Daily water level forecasting using wavelet decomposition and artificial intelligence techniques , 2015 .

[7]  Dianhui Wang,et al.  Extreme learning machines: a survey , 2011, Int. J. Mach. Learn. Cybern..

[8]  Andrew E. Suyker,et al.  Annual carbon dioxide exchange in irrigated and rainfed maize-based agroecosystems , 2005 .

[9]  Michio Sugeno,et al.  Fuzzy identification of systems and its applications to modeling and control , 1985, IEEE Transactions on Systems, Man, and Cybernetics.

[10]  Nuno Carvalhais,et al.  Effect of spatial sampling from European flux towers for estimating carbon and water fluxes with artificial neural networks , 2015 .

[11]  Shifei Ding,et al.  Extreme learning machine and its applications , 2013, Neural Computing and Applications.

[12]  W. Oechel,et al.  Testing the applicability of neural networks as a gap-filling method using CH 4 flux data from high latitude wetlands , 2013 .

[13]  Markus Reichstein,et al.  Effects of climate extremes on the terrestrial carbon cycle: concepts, processes and potential future impacts , 2015, Global change biology.

[14]  Luis Alonso,et al.  Machine learning regression algorithms for biophysical parameter retrieval: Opportunities for Sentinel-2 and -3 , 2012 .

[15]  Holger R. Maier,et al.  Neural network based modelling of environmental variables: A systematic approach , 2001 .

[16]  Gulay Tezel,et al.  Monthly evaporation forecasting using artificial neural networks and support vector machines , 2016, Theoretical and Applied Climatology.

[17]  Yiqi Luo,et al.  Main and interactive effects of warming, clipping, and doubled precipitation on soil CO2 efflux in a grassland ecosystem , 2006 .

[18]  Ozgur Kisi,et al.  Streamflow Forecasting and Estimation Using Least Square Support Vector Regression and Adaptive Neuro-Fuzzy Embedded Fuzzy c-means Clustering , 2015, Water Resources Management.

[19]  Markus Reichstein,et al.  Biosphere-atmosphere exchange of CO2 in relation to climate: a cross-biome analysis across multiple time scales , 2009 .

[20]  Mimi Haryani Hassim,et al.  Artificial neural networks: applications in chemical engineering , 2013 .

[21]  Atul K. Jain,et al.  Role of CO2, climate and land use in regulating the seasonal amplitude increase of carbon fluxes in terrestrial ecosystems: a multimodel analysis , 2016, Biogeosciences.

[22]  Zaher Mundher Yaseen,et al.  Novel approach for streamflow forecasting using a hybrid ANFIS-FFA model , 2017 .

[23]  Joao P. S. Catalao,et al.  Short-term wind power forecasting using adaptive neuro-fuzzy inference system combined with evolutionary particle swarm optimization, wavelet transform and mutual information , 2015 .

[24]  Christian W. Dawson,et al.  Hydrological modelling using artificial neural networks , 2001 .

[25]  Brian L. McGlynn,et al.  Land-atmosphere carbon and water flux relationships to vapor pressure deficit, soil moisture, and stream flow , 2015 .

[26]  Michael Bahn,et al.  Seasonal and inter-annual variability of the net ecosystem CO2 exchange of a temperate mountain grassland: effects of climate and management. , 2008, Journal of geophysical research. Atmospheres : JGR.

[27]  G. Lewicki,et al.  Approximation by Superpositions of a Sigmoidal Function , 2003 .

[28]  Hubert H. G. Savenije,et al.  Hydrological model coupling with ANNs , 2006 .

[29]  A. Vahedi,et al.  Artificial neural network application in comparison with modeling allometric equations for predicting above-ground biomass in the Hyrcanian mixed-beech forests of Iran , 2016 .

[30]  Julian D Olden,et al.  Machine Learning Methods Without Tears: A Primer for Ecologists , 2008, The Quarterly Review of Biology.

[31]  Bernhard Schölkopf,et al.  A tutorial on support vector regression , 2004, Stat. Comput..

[32]  T. A. Black,et al.  Factors controlling the interannual variability in the carbon balance of a southern boreal black spruce forest , 2008 .

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

[34]  Jungho Im,et al.  ISPRS Journal of Photogrammetry and Remote Sensing , 2022 .

[35]  C. Peng,et al.  Applying an artificial neural network to simulate and predict Chinese fir (Cunninghamia lanceolata) plantation carbon flux in subtropical China , 2014 .

[36]  Philippe Ciais,et al.  Evaluation of continental carbon cycle simulations with North American flux tower observations , 2013 .

[37]  Ozgur Kisi,et al.  Evaluation of data driven models for river suspended sediment concentration modeling , 2016 .

[38]  Shaoqiang Wang,et al.  Impact of meteorological anomalies in the 2003 summer on Gross Primary Productivity in East Asia , 2009 .

[39]  Vladimir Vapnik,et al.  An overview of statistical learning theory , 1999, IEEE Trans. Neural Networks.

[40]  John A. Gamon,et al.  Application of the photosynthetic light-use efficiency model in a northern Great Plains grassland , 2015 .

[41]  Henry W. Loescher,et al.  Uncertainties in, and interpretation of, carbon flux estimates using the eddy covariance technique , 2006 .

[42]  Matthew B Jones,et al.  Ecoinformatics: supporting ecology as a data-intensive science. , 2012, Trends in ecology & evolution.

[43]  Fatih Evrendilek.  Assessing neural networks with wavelet denoising and regression models in predicting diel dynamics of eddy covariance-measured latent and sensible heat fluxes and evapotranspiration , 2012, Neural Computing and Applications.

[44]  George Sugihara,et al.  Equation-free mechanistic ecosystem forecasting using empirical dynamic modeling , 2015, Proceedings of the National Academy of Sciences.

[45]  W. Oechel,et al.  Variability in exchange of CO2 across 12 northern peatland and tundra sites , 2009 .

[46]  P. Stassen.  Carbon cycle: Global warming then and now , 2016 .

[47]  Saad Mekhilef,et al.  Adaptive neuro-fuzzy approach for solar radiation prediction in Nigeria , 2015 .

[48]  Daniel M. Ricciuto,et al.  Predicting long‐term carbon sequestration in response to CO2 enrichment: How and why do current ecosystem models differ? , 2015 .

[49]  Atul K. Jain,et al.  Impact of large‐scale climate extremes on biospheric carbon fluxes: An intercomparison based on MsTMIP data , 2013 .

[50]  S. Seneviratne,et al.  Hydrological and biogeochemical constraints on terrestrial carbon cycle feedbacks , 2017 .

[51]  A. Rogers,et al.  The response of photosynthesis and stomatal conductance to rising [CO2]: mechanisms and environmental interactions. , 2007, Plant, cell & environment.

[52]  N. K. Goel,et al.  Improving real time flood forecasting using fuzzy inference system , 2014 .

[53]  Ozgur Kisi,et al.  Evaluation of several soft computing methods in monthly evapotranspiration modelling , 2018 .

[54]  S. Seneviratne,et al.  Energy balance closure of eddy-covariance data: a multisite analysis for European FLUXNET stations. , 2010 .

[55]  S. Wofsy,et al.  Carbon neutral or a sink? Uncertainty caused by gap-filling long-term flux measurements for an old-growth boreal black spruce forest , 2017 .

[56]  Chih-Jen Lin,et al.  LIBSVM: A library for support vector machines , 2011, TIST.

[57]  Young-Chan Lee,et al.  Bankruptcy prediction using support vector machine with optimal choice of kernel function parameters , 2005, Expert Syst. Appl..

[58]  W. Bauerle,et al.  Carbon and water flux responses to physiology by environment interactions: a sensitivity analysis of variation in climate on photosynthetic and stomatal parameters , 2014, Climate Dynamics.

[59]  Badih Ghattas,et al.  A review of supervised machine learning algorithms and their applications to ecological data , 2012 .

[60]  Chee Kheong Siew,et al.  Extreme learning machine: Theory and applications , 2006, Neurocomputing.

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

[62]  José M. Paruelo,et al.  Land cover and precipitation controls over long‐term trends in carbon gains in the grassland biome of South America , 2015 .

[63]  Hossein Tabari,et al.  Applicability of support vector machines and adaptive neurofuzzy inference system for modeling potato crop evapotranspiration , 2012, Irrigation Science.

[64]  D. Baldocchi,et al.  Measuring fluxes of trace gases and energy between ecosystems and the atmosphere – the state and future of the eddy covariance method , 2014, Global change biology.

[65]  Corinna Cortes,et al.  Support-Vector Networks , 1995, Machine Learning.

[66]  T. A. Black,et al.  Evaluating the agreement between measurements and models of net ecosystem exchange at different times and timescales using wavelet coherence: an example using data from the North American Carbon Program Site-Level Interim Synthesis , 2013 .

[67]  A. Arneth,et al.  Terrestrial biogeochemical feedbacks in the climate system , 2010 .

[68]  Atul K. Jain,et al.  North American Carbon Program (NACP) regional interim synthesis: Terrestrial biospheric model intercomparison , 2012 .

[69]  Maurizio Mencuccini,et al.  After more than a decade of soil moisture deficit, tropical rainforest trees maintain photosynthetic capacity, despite increased leaf respiration , 2015, Global change biology.

[70]  N. Hanan,et al.  The sensitivity of carbon exchanges in Great Plains grasslands to precipitation variability , 2016 .

[71]  Daniel E. Schindler,et al.  Prediction, precaution, and policy under global change , 2015, Science.

[72]  Yongsheng Ding,et al.  Forecasting financial condition of Chinese listed companies based on support vector machine , 2008, Expert Syst. Appl..

[73]  Mehmet Şahin,et al.  Application of extreme learning machine for estimating solar radiation from satellite data , 2014 .

[74]  Garth van der Kamp,et al.  Interannual variation of evapotranspiration from forest and grassland ecosystems in western canada in relation to drought , 2010 .

[75]  M. Heimann,et al.  Comprehensive comparison of gap-filling techniques for eddy covariance net carbon fluxes , 2007 .

[76]  F. Evrendilek.  Quantifying biosphere-atmosphere exchange of CO2 using eddy covariance, wavelet denoising, neural networks, and multiple regression models , 2013 .

[77]  J. Tenhunen,et al.  Supplement understanding of the relative importance of biophysical factors in determination of photosynthetic capacity and photosynthetic productivity in rice ecosystems , 2017 .

[78]  M. Aubinet,et al.  Carbon sequestration by a crop over a 4-year sugar beet/winter wheat/seed potato/winter wheat rotation cycle , 2009 .

[79]  P. Coulibaly,et al.  Two decades of anarchy? Emerging themes and outstanding challenges for neural network river forecasting , 2012 .

[80]  F. Woodward,et al.  Global response of terrestrial ecosystem structure and function to CO2 and climate change: results from six dynamic global vegetation models , 2001 .

[81]  C. W. Tong,et al.  Retraction Note to: Application of extreme learning machine for estimation of wind speed distribution , 2018, Climate Dynamics.

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

[83]  Dara Entekhabi,et al.  Regionally strong feedbacks between the atmosphere and terrestrial biosphere. , 2017, Nature geoscience.

[84]  T. Griffisa,et al.  Ecophysiological controls on the carbon balances of three southern boreal forests , 2003 .

[85]  Yiqi Luo.  Terrestrial Carbon-Cycle Feedback to Climate Warming , 2007 .

[86]  Mehdi Vafakhah,et al.  A Wavelet-ANFIS Hybrid Model for Groundwater Level Forecasting for Different Prediction Periods , 2013, Water Resources Management.

[87]  K. S. Yap,et al.  Extreme Learning Machines: A new approach for prediction of reference evapotranspiration , 2015 .

[88]  Shahaboddin Shamshirband,et al.  Predicting the wind power density based upon extreme learning machine , 2015 .

[89]  Douglas A. Johnson,et al.  Grazing effects on carbon fluxes in a Northern China grassland , 2015 .

[90]  M. Aubinet,et al.  Carbon budget measurement over 12 years at a crop production site in the silty-loam region in Belgium , 2017 .

[91]  T. Meyers,et al.  An assessment of storage terms in the surface energy balance of maize and soybean , 2004 .

[92]  S. Seneviratne,et al.  Drought and ecosystem carbon cycling , 2011 .

[93]  Ken-ichi Funahashi,et al.  On the approximate realization of continuous mappings by neural networks , 1989, Neural Networks.