Fusion of Multiple Models for Improving Gross Primary Production Estimation With Eddy Covariance Data Based on Machine Learning

Terrestrial gross primary production (GPP) represents the magnitude of CO2 uptake through vegetation photosynthesis, and is a key variable for carbon cycles between the biosphere and atmosphere. Light use efficiency (LUE) models have been widely used to estimate GPP for its physiological mechanisms and availability of data acquisition and implementation, yet each individual GPP model has exhibited large uncertainties due to input errors and model structure, and further studies of systematic validation, comparison, and fusion of those models with eddy covariance (EC) site data across diverse ecosystem types are still needed in order to further improve GPP estimation. We here compared and fused five GPP models (VPM, EC‐LUE, GOL‐PEM, CHJ, and C‐Fix) across eight ecosystems based on FLUXNET2015 data set using the ensemble methods of Bayesian Model Averaging (BMA), Support Vector Machine (SVM), and Random Forest (RF) separately. Our results showed that for individual models, EC‐LUE gave a better performance to capture interannual variability of GPP than other models, followed by VPM and GLO‐PEM, while CHJ and C‐Fix were more limited in their estimation performance. We found RF and SVM were superior to BMA on merging individual models at various plant functional types (PFTs) and at the scale of individual sites. On the basis of individual models, the fusion methods of BMA, SVM, and RF were examined by a five‐fold cross validation for each ecosystem type, and each method successfully improved the average accuracy of estimation by 8%, 18%, and 19%, respectively.

• Publication year

  2023

• Venue

  Journal of Geophysical Research: Biogeosciences

• Publication date

  2023-03-01

• Fields of study

  Not labeled

• Identifiers
  DOI 10.1029/2022JG007122
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Urban Land Use and Land Cover Change Analysis Using Random Forest Classification of Landsat Time Series

  2022cited by this paper
• Evolution of light use efficiency models: Improvement, uncertainties, and implications

  2022cited by this paper
• Quantitative assessment of the importance of bio-physical drivers of land cover change based on a random forest method

  2021cited by this paper
• How Well Do We Understand the Land‐Ocean‐Atmosphere Carbon Cycle?

  2021cited by this paper
• Probabilistic forecasting of crop yields via quantile random forest and Epanechnikov Kernel function

  2020cited by this paper
• The FLUXNET2015 dataset and the ONEFlux processing pipeline for eddy covariance data

  2020influential reference
• Estimating Crop Primary Productivity with Sentinel-2 and Landsat 8 using Machine Learning Methods Trained with Radiative Transfer Simulations

  2019cited by this paper
• The physics and ecology of mining carbon dioxide from the atmosphere by ecosystems

  2019cited by this paper
• Improving global terrestrial evapotranspiration estimation using support vector machine by integrating three process-based algorithms.

  2017influential reference
• A global moderate resolution dataset of gross primary production of vegetation for 2000–2016

  2017influential reference
• Temporal consistency between gross primary production and solar-induced chlorophyll fluorescence in the ten most populous megacity areas over years

  2017cited by this paper
• Dominant role of plant physiology in trend and variability of gross primary productivity in North America

  2017influential reference
• The response of ecosystem carbon fluxes to LAI and environmental drivers in a maize crop grown in two contrasting seasons

  2016cited by this paper
• Consistency Between Sun-Induced Chlorophyll Fluorescence and Gross Primary Production of Vegetation in North America

  2016cited by this paper
• A global study of GPP focusing on light-use efficiency in a random forest regression model

  2016cited by this paper
• Comparison of Four Machine Learning Methods for Generating the GLASS Fractional Vegetation Cover Product from MODIS Data

  2016influential reference
• Seasonal controls of canopy chlorophyll content on forest carbon uptake: Implications for GPP modeling

  2015cited by this paper
• Comparison of four light use efficiency models for estimating terrestrial gross primary production

  2015influential reference
• Using Bayesian model averaging to estimate terrestrial evapotranspiration in China

  2015cited by this paper
• NEP of a Swiss subalpine forest is significantly driven not only by current but also by previous year's weather

  2014cited by this paper
• Comparison of Gross Primary Productivity Derived from GIMMS NDVI3g, GIMMS, and MODIS in Southeast Asia

  2014cited by this paper
• Bayesian multimodel estimation of global terrestrial latent heat flux from eddy covariance, meteorological, and satellite observations

  2014cited by this paper
• Global comparison of light use efficiency models for simulating terrestrial vegetation gross primary production based on the LaThuile database

  2014influential reference
• Greenhouse gas budget (CO2, CH4 and N2O) of intensively managed grassland following restoration

  2014cited by this paper
• Distribution of non-structural nitrogen and carbohydrate compounds in mature oak trees in a temperate forest at four key phenological stages

  2013cited by this paper
• Spatial and temporal validation of the MODIS LAI and FPAR products across a boreal forest wildfire chronosequence

  2013cited by this paper
• A data-driven analysis of energy balance closure across FLUXNET research sites: The role of landscape-scale heterogeneity

  2013cited by this paper
• Evaluation of continental carbon cycle simulations with North American flux tower observations

  2013influential reference
• Temporal and spatial variations of soil CO 2 , CH 4 and N 2 O fluxes at three differently managed grasslands

  2013cited by this paper
• Soil surface CO2 efflux measurements in Norway spruce forests: Comparison between four different sites across Europe — from boreal to alpine forest

  2013cited by this paper
• A long-term Global LAnd Surface Satellite (GLASS) data-set for environmental studies

  2013influential reference
• Estimation of gross primary production over the terrestrial ecosystems in China

  2013influential reference
• Use of change-point detection for friction–velocity threshold evaluation in eddy-covariance studies

  2013cited by this paper
• Eddy Covariance: A Practical Guide to Measurement and Data Analysis

  2012cited by this paper
• Water use of forests in the Netherlands

  2012cited by this paper
• Terrestrial biosphere model performance for inter‐annual variability of land‐atmosphere CO2 exchange

  2012cited by this paper
• Estimation of clear‐sky land surface longwave radiation from MODIS data products by merging multiple models

  2012influential reference
• Increasing net CO2 uptake by a Danish beech forest during the period from 1996 to 2009

  2011cited by this paper
• Climatic controls and ecosystem responses drive the inter-annual variability of the net ecosystem exchange of an alpine meadow

  2011cited by this paper
• SPECIAL—Savanna Patterns of Energy and Carbon Integrated across the Landscape

  2011cited by this paper
• Carbon, nitrogen and Greenhouse gases budgets over a four years crop rotation in northern France

  2011cited by this paper
• Bayesian estimation of local signal and noise in multimodel simulations of climate change

  2010cited by this paper
• Comparison of multiple models for estimating gross primary production using MODIS and eddy covariance data in Harvard Forest

  2010cited by this paper
• Land use regulates carbon budgets in eastern Germany: From NEE to NBP

  2010cited by this paper
• Separation of net ecosystem exchange into assimilation and respiration using a light response curve approach: critical issues and global evaluation

  2010influential reference
• Gross primary production estimation from MODIS data with vegetation index and photosynthetically active radiation in maize

  2010cited by this paper
• Global estimates of evapotranspiration and gross primary production based on MODIS and global meteorology data

  2010influential reference
• Climate control of terrestrial carbon exchange across biomes and continents

  2010influential reference
• Interannual CO2 exchange of a sparse Mediterranean shrubland on a carbonaceous substrate

  2009cited by this paper
• A new mass conservation approach to the study of CO2 advection in an alpine forest

  2009cited by this paper
• Modeling carbon dynamics in two adjacent spruce forests with different soil conditions in Russia

  2008cited by this paper
• The contribution of advective fluxes to net ecosystem exchange in a high-elevation, subalpine forest.

  2008cited by this paper
• The energy balance closure problem: an overview.

  2008cited by this paper
• A new model of gross primary productivity for North American ecosystems based solely on the enhanced vegetation index and land surface temperature from MODIS

  2008cited by this paper
• Seasonal and inter-annual variability of the net ecosystem CO2 exchange of a temperate mountain grassland: effects of climate and management.

  2008cited by this paper
• Impact of severe dry season on net ecosystem exchange in the Neotropical rainforest of French Guiana

  2008cited by this paper
• Parametrization of two photosynthesis models at the canopy scale in a northern boreal Scots pine forest

  2007cited by this paper
• The growing season greenhouse gas balance of a continental tundra site in the Indigirka lowlands, NE Siberia

  2007cited by this paper
• Deriving a light use efficiency model from eddy covariance flux data for predicting daily gross primary production across biomes

  2007influential reference
• The impact of diffuse sunlight on canopy light‐use efficiency, gross photosynthetic product and net ecosystem exchange in three forest biomes

  2007cited by this paper
• Classification and Regression by randomForest

  2007cited by this paper
• Variability of annual CO 2 exchange from Dutch grasslands

  2007cited by this paper
• On the use of MODIS EVI to assess gross primary productivity of North American ecosystems

  2006cited by this paper
• Sensitivity of Moderate Resolution Imaging Spectroradiometer (MODIS) terrestrial primary production to the accuracy of meteorological reanalyses

  2006influential reference
• Annual net ecosystem carbon exchange by a sugar beet crop

  2006cited by this paper
• Multiobjective analysis of chaotic dynamic systems with sparse learning machines

  2006cited by this paper
• Conditions for sink-to-source transitions and runaway feedbacks from the land carbon cycle

  2006cited by this paper
• Using Bayesian Model Averaging to Calibrate Forecast Ensembles

  2005cited by this paper
• Modeling and measuring the nocturnal drainage flow in a high‐elevation, subalpine forest with complex terrain

  2005cited by this paper
• Net ecosystem production in a temperate pine plantation in southeastern Canada

  2005cited by this paper
• On the separation of net ecosystem exchange into assimilation and ecosystem respiration: review and improved algorithm

  2005influential reference
• Modelling carbon budget of Mediterranean forests using ground and remote sensing measurements

  2005cited by this paper
• Satellite-based modeling of gross primary production in a seasonally moist tropical evergreen forest

  2005influential reference
• Improvements of the MODIS terrestrial gross and net primary production global data set

  2005influential reference
• Carbon and water fluxes over a temperate Eucalyptus forest and a tropical wet/dry savanna in Australia: measurements and comparison with MODIS remote sensing estimates

  2005cited by this paper
• Dynamics of evapotranspiration in semiarid grassland and shrubland ecosystems during the summer monsoon season, central New Mexico

  2004cited by this paper
• A Continuous Satellite-Derived Measure of Global Terrestrial Primary Production

  2004influential reference
• Photosynthetic characteristics of rice leaves grown under red light with or without supplemental blue light.

  2004cited by this paper
• The growth respiration component in eddy CO2 flux from a Quercus ilex mediterranean forest

  2004cited by this paper
• Hourly and daytime evapotranspiration from grassland using radiometric surface temperatures

  2004cited by this paper
• Forest and agricultural land‐use‐dependent CO2 exchange in Thuringia, Germany

  2004cited by this paper
• Seasonal changes in photosynthesis, respiration and NEE of a mixed temperate forest

  2004cited by this paper
• Remotely Sensed Interannual Variations and Trends in Terrestrial Net Primary Productivity 1981–2000

  2004influential reference
• The growth respiration component in eddy CO 2 flux from a Quercus ilex mediterranean forest

  2004cited by this paper
• Inter-annual variability in the leaf area index of a boreal aspen-hazelnut forest in relation to net ecosystem production

  2004cited by this paper
• A nonparametric method for separating photosynthesis and respiration components in CO2 flux measurements

  2004cited by this paper
• Satellite-based modeling of gross primary production in an evergreen needleleaf forest

  2004cited by this paper
• Modeling gross primary production of temperate deciduous broadleaf forest using satellite images and climate data

  2004influential reference
• Canopy Architecture and Turbulence Structure in a Coniferous Forest

  2003cited by this paper
• Climate-Driven Increases in Global Terrestrial Net Primary Production from 1982 to 1999

  2003cited by this paper
• Advantages of diffuse radiation for terrestrial ecosystem productivity

  2002cited by this paper
• Designing a spectral index to estimate vegetation water content from remote sensing data: Part 1 - Theoretical approach

  2002cited by this paper
• Estimation of carbon mass fluxes over Europe using the C-Fix model and Euroflux data

  2002cited by this paper
• Large carbon uptake by an unmanaged 250-year-old deciduous forest in Central Germany

  2002cited by this paper
• Characterization of forest types in Northeastern China, using multi-temporal SPOT-4 VEGETATION sensor data

  2002cited by this paper
• Long term carbon dioxide exchange above a mixed forest in the Belgian Ardennes

  2001cited by this paper
• Random Forests

  2001cited by this paper
• Support Vector Machines: A Nonlinear Modelling and Control Perspective

  2001cited by this paper
• Global Terrestrial Gross and Net Primary Productivity from the Earth Observing System

  2000influential reference
• Interannual variability of global terrestrial primary production: Results of a model driven with satellite observations

  2000cited by this paper

• MCI GPP: ensembling a global model- and climate-independent gross primary productivity for 2001–2023

  2025cites this paper
• Modeling Forest Growth Under Current and Future Climate

  2025cites this paper
• A 2001–2022 global gross primary productivity dataset using an ensemble model based on the random forest method

  2024cites this paper
• Gross primary productivity estimation through remote sensing and machine learning techniques in the high Andean Region of Ecuador

  2024cites this paper
• Solar-Induced Chlorophyll Fluorescence-Based GPP Estimation and Analysis of Influencing Factors for Xinjiang Vegetation

  2024cites this paper
• Estimating Forest Gross Primary Production Using Machine Learning, Light Use Efficiency Model, and Global Eddy Covariance Data

  2024cites this paper
• Predicting Gross Primary Productivity of the Forest Ecosystems using Machine Learning Techniques: A Review of Existing Approaches

  2024cites this paper
• Prediction of Gross Primary Productivity using Hybrid LSTM-GRU Neural Networks and ERA5-Reanalysis Data

  2023cites this paper
• Assessment of Six Machine Learning Methods for Predicting Gross Primary Productivity in Grassland

  2023cites this paper
• SOMALIA: Drought Response

  2022cites this paper
• an 1 ensemble model based on random forest 2

  year unknowncites this paper