Partitioning net carbon dioxide fluxes into photosynthesis and respiration using neural networks

Tramontana, Gianluca; Migliavacca, Mirco; Jung, Martin; Reichstein, Markus; Keenan, Trevor F.; Camps‐Valls, Gustau; Ogée, Jérôme; Verrelst, Jochem; Papale, Dario

doi:10.1111/gcb.15203

Cited by 67 publications

(69 citation statements)

References 88 publications

(96 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Artificial Neural Networks have been applied for gap-filling and partitioning EC fluxes in the past (Papale & Valentini, 2003;Oikawa et al, 2017;Tramontana et al, 2020). Specifically, for CO 2 fluxes, ANNs have shown to perform well when used to gap-fill missing data (Moffat et al, 2007) and partitioning net CO 2 fluxes into the component fluxes of gross primary production (GPP) and ecosystem respiration (R eco ) (Desai et al, 2008;Oikawa et al, 2017;Tramontana et al, 2020). Following a similar approach to partitioning CO 2 data, we assumed that nighttime ET data is dominated by E at these flooded sites:…”

Section: Artificial Neural Network Partitioning Routinementioning

confidence: 99%

“…A newer approach used to partition net ecosystem carbon fluxes into the individual components of gross primary production and ecosystem respiration uses Artificial Neural Networks (ANN) (Papale & Valentini, 2003;Desai et al, 2008;Tramontana et al, 2020).…”

mentioning

confidence: 99%

“…Although the use of ANNs could also be directed at T/ET partitioning, the application of this technique has not been done yet and needs further exploration. Since machine learning methods can resolve complex, nonlinear relationships between environmental drivers and flux variables (Tramontana et al, 2020), ANNs are a promising approach to partition T/ET in ecosystems where existing ET partitioning methods face limitations, such as wetlands and river deltas.…”

mentioning

confidence: 99%

See 2 more Smart Citations

A novel approach to partitioning evapotranspiration into evaporation and transpiration in flooded ecosystems

Eichelmann

Mantoani

Chamberlain

et al. 2021

Preprint

View full text Add to dashboard Cite

Reliable partitioning of micrometeorologically measured evapotranspiration (ET) into evaporation (E) and transpiration (T) would greatly enhance our understanding of the water cycle and its response to climate change. While some methods on ET partitioning have been developed, their underlying assumptions make them difficult to apply more generally, especially in sites with large contributions of E. Here, we report a novel ET partitioning method using Artificial Neural Networks (ANN) in combination with a range of environmental input variables to predict daytime E from nighttime ET measurements. The study uses eddy covariance data from four restored wetlands in the Sacramento-San Joaquin Delta, California, USA, as well as leaf-level T data for validation. The four wetlands vary in structure from some with large areas of open water and little vegetation to very densely vegetated wetlands, representing a range of ET conditions. The ANNs were built with increasing complexity by adding the input variable that resulted in the next highest average value of model testing R2 across all sites. The order of variable inclusion (and importance) was: vapor pressure deficit (VPD) > gap-filled sensible heat flux (H_gf) > air temperature (Tair) > friction velocity (u∗) > other variables. Overall, 36 ANNs were analyzed. The model using VPD, H_gf, Tair, and u∗ (F11), showed an average testing R2 value across all sites of 0.853. In comparison with the model that included all 10 variables (F36), F11 generally performed better during validation with independent data. In comparison to other methods described in the literature, the ANN method generated more consistent T/ET partitioning results especially for more complex sites with large E contributions. Our method improves the understanding of T/ET partitioning. While it may be particularly suited to flooded ecosystems, it can also improve T/ET partitioning in other systems, increasing our knowledge of the global water cycle.

show abstract

Section: Artificial Neural Network Partitioning Routinementioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

A novel approach to partitioning evapotranspiration into evaporation and transpiration in flooded ecosystems

Eichelmann

Mantoani

Chamberlain

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Similarly to other numerical NEE partitioning methods such as that by Lasslop et al [4], although very useful and easy to implement, it does not consider multiple co-acting factors that modulate GPP and R eco . Tramontana et al [1] develop an approach to account for the said factors modulating these two carbon fluxes by implementing a hybrid data-driven method, NN C-part , based on feedforward neural networks [5], that can use a comprehensive dataset of soil and micro-meteorological variables. The expert knowledge is incorporated in the algorithm by introducing a photosynthesis response based on the radiation-use efficiency concept.…”

Section: A Related Workmentioning

confidence: 99%

“…The task of net ecosystem exchange (NEE) partitioning is highly relevant in environmental science, as it deepens the understanding of the underlying mechanisms constraining the ecosystem function, in the global warming context for instance [1]. NEE is a measure of the net exchange of carbon This work is funded by the ERC Synergy Grant 2019: Understanding and Modelling of the Earth System with Machine Learning (USMILE) and by the Carl Zeiss Foundation within the scope of the program line "Breakthroughs: Exploring Intelligent Systems" for "Digitization -explore the basics, use applications".…”

Section: Introductionmentioning

confidence: 99%

A Data-Driven Approach to Partitioning Net Ecosystem Exchange Using a Deep State Space Model

et al. 2021

Self Cite

View full text Add to dashboard Cite

Describing ecosystem carbon fluxes is essential for deepening the understanding of the Earth system. However, partitioning net ecosystem exchange (NEE), i.e. the sum of ecosystem respiration (R eco ) and gross primary production (GPP), into these summands is ill-posed since there can be infinitely many mathematically-valid solutions. We propose a novel data-driven approach to NEE partitioning using a deep state space model which combines the interpretability and uncertainty analysis of state space models with the ability of recurrent neural networks to learn the complex functions governing the data. We validate our proposed approach on the FLUXNET dataset. We suggest using both the past and the future of R eco 's predictors for training along with the nighttime NEE (NEE night ) to learn a dynamical model of R eco . We evaluate our nighttime R eco forecasts by comparing them to the ground truth NEE night and obtain the best accuracy with respect to other partitioning methods. The learned nighttime R eco model is then used to forecast the daytime R eco conditioning on the future observations of different predictors, i.e., global radiation, air temperature, precipitation, vapor pressure deficit, and daytime NEE (NEE day ). Subtracted from the NEE day , these estimates yield the GPP, finalizing the partitioning. Our purely data-driven daytime R eco forecasts are in line with the recent empirical partitioning studies reporting lower daytime R eco than the Reichstein method, which can be attributed to the Kok effect, i.e., the plant respiration being higher at night. We conclude that our approach is a good alternative for data-driven NEE partitioning and complements other partitioning methods. INDEX TERMSDeep state space models, net ecosystem exchange, NEE partitioning, time series forecasting

show abstract

A multi-perspective input selection strategy for daily net ecosystem exchange predictions based on machine learning methods

Ekmekçioğlu

Başakın²,

Altınbaş³

et al. 2022

Theor Appl Climatol

View full text Add to dashboard Cite

Partitioning net carbon dioxide fluxes into photosynthesis and respiration using neural networks

Abstract: This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Cited by 67 publications

References 88 publications

A novel approach to partitioning evapotranspiration into evaporation and transpiration in flooded ecosystems

A novel approach to partitioning evapotranspiration into evaporation and transpiration in flooded ecosystems

A Data-Driven Approach to Partitioning Net Ecosystem Exchange Using a Deep State Space Model

A multi-perspective input selection strategy for daily net ecosystem exchange predictions based on machine learning methods

Contact Info

Product

Resources

About