Importance of the memory effect for assessing interannual variation in net ecosystem exchange

Liu, Weihua; He, Honglin; Wu, Xiaojing; Ren, Xiaoli; Zhang, Li; shi, Liang; Feng, Lili; Wang, Yangang; Lv, Yan

doi:10.1016/j.agrformet.2023.109691

Cited by 5 publications

(8 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the study by Liu et al. (2023), they found the optimal memory effect lengths diverged across PFTs. From their study, 6 months are proper for most PFTs.…”

Section: Discussionmentioning

confidence: 97%

“…The recent rapid development of deep learning (DL) technology has shed new light on Earth system modeling (Irrgang et al, 2021). In particular, its capacity for mining historical time-series information from multi-source ecosystem observations offers a great potential for improving terrestrial carbon flux estimation (Besnard et al, 2019;Liu et al, 2022Liu et al, , 2023, while for TBMs, it is more challenging to incorporate environmental memory information into carbon flux modeling. The Long Short-Term Memory model (LSTM) is a dynamic statistical method that has demonstrated excellent performance on sequence data, such as crop field classification (Rußwurm & Körner, 2018).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Exploring the Potential of Long Short‐Term Memory Networks for Predicting Net CO₂ Exchange Across Various Ecosystems With Multi‐Source Data

Huang,

He,

Liu

et al. 2024

JGR Atmospheres

View full text Add to dashboard Cite

Upscaling flux tower measurements based on machine learning (ML) algorithms is an essential approach for large‐scale net ecosystem CO2 exchange (NEE) estimation, but existing ML upscaling methods face some challenges, particularly in capturing NEE interannual variations (IAVs) that may relate to lagged effects. With the capacity to characterize temporal memory effects, the Long Short‐Term Memory (LSTM) networks are expected to help solve this problem. Here we explored the potential of LSTM for predicting NEE across various ecosystems using flux tower data over 82 sites in North America. The LSTM model with differentiated plant function types (PFTs) demonstrates the capability to explain 79.19% (R2 = 0.79) of the monthly variations in NEE within the testing set, with RMSE and Mean Absolute Error values of 0.89 and 0.57 g C m−2 d−1 respectively (r = 0.89, p < 0.001). Moreover, the LSTM model performed robustly in predicting cross‐site variability, with 67.19% of the sites that can be predicted by both LSTM models with and without distinguished PFTs showing improved predictive ability. Most importantly, the IAV of predicted NEE highly correlated with that in flux observations (r = 0.81, p < 0.001), clearly outperforming that by the random forest model (r = −0.21, p = 0.011). Among all nine PFTs, solar‐induced chlorophyll fluorescence, downward shortwave radiation, and leaf area index are the most important variables for explaining NEE variations, collectively accounting for approximately 54.01% in total. This study highlights the great potential of LSTM for improving carbon flux upscaling with multi‐source remote sensing data.

show abstract

“…In the study by Liu et al. (2023), they found the optimal memory effect lengths diverged across PFTs. From their study, 6 months are proper for most PFTs.…”

Section: Discussionmentioning

confidence: 97%

Section: Introductionmentioning

confidence: 99%

Exploring the Potential of Long Short‐Term Memory Networks for Predicting Net CO₂ Exchange Across Various Ecosystems With Multi‐Source Data

Huang,

He,

Liu

et al. 2024

JGR Atmospheres

View full text Add to dashboard Cite

show abstract

“…The memory effect is important factor in controlling the IAV of NEE (Bloom et al, 2020), but has not been considered in FLUXCOM. Recent researches demonstrated the potential of considering memory effects in improving terrestrial carbon flux simulations (Besnard et al, 2019;Liu et al, 2023).…”

Section: Advantages Of Lstm Over Rf In Predicting Nee and Its Iavmentioning

confidence: 99%

“…Recent rapid development of deep learning (DL) technology has shed new light on Earth system modeling (Irrgang et al, 2021). In particular, its capacity for mining historical time-series information from multi-source ecosystem observations offers a great potential for improving terrestrial carbon flux estimation (Besnard et al, 2019;Liu et al, 2023), which incorporates environmental memory into flux modeling while difficult to implement in state-of-the-art process models. The Long Short-Term Memory model (LSTM) is a dynamic statistical method that has demonstrated excellent performance on sequence data, such as crop field classification (Rußwurm & Körner, 2018).…”

Section: Introductionmentioning

confidence: 99%

Exploring the Potential of Long Short-Term Memory Networks for Predicting Net CO2 Exchange Across Various Ecosystems With Multi-Source Data

Huang,

He,

Liu

et al. 2023

Preprint

View full text Add to dashboard Cite

Upscaling flux tower measurements based on machine learning (ML) algorithms is an essential approach for large-scale net ecosystem CO2 exchange (NEE) estimation, but existing ML upscaling methods face some challenges, particularly in capturing NEE interannual variations (IAVs) that may relate to lagged effects. With the capacity of characterizing temporal memory effects, the Long Short-Term Memory (LSTM) networks are expected to help solve this problem. Here we explored the potential of LSTM for predicting NEE across various ecosystems using flux tower data over 82 sites in North America. The LSTM model with differentiated plant function types (PFTs) demonstrates the capability to explain 79.19% (R2 = 0.79) of the monthly variations in NEE within the testing set, with RMSE and MAE values of 0.89 and 0.57 g C m-2 d-1 respectively (r = 0.89, p < 0.001). Moreover, the LSTM model performed robustly in predicting cross-site variability, with 67.19% of the sites that can be predicted by both LSTM models with and without distinguished PFTs showing improved predictive ability. Most importantly, the IAV of predicted NEE highly correlated with that in flux observations (r = 0.81, p < 0.001), clearly outperforming that by the random forest model (r = -0.21, p = 0.011). Among all nine PFTs, solar-induced chlorophyll fluorescence, downward shortwave radiation, and leaf area index are the most important variables for explaining NEE variations, collectively accounting for approximately 54.01% in total. This study highlights the great potential of LSTM for improving carbon flux upscaling with multi-source remote sensing data.

show abstract

“…Although this method has been refined over the years (Bodesheim et al, 2018;Jung et al, 2011Jung et al, , 2020, some major drawbacks remain. For example, these estimates under-estimate the magnitude of the inter-annual variability of the net ecosystem uptake of CO 2 (Jung et al, 2020;Liu et al, 2023), due to the tendency of these models to predict the mean flux. Moreover, as these methods are purely trained on historical data, without including processes understanding, their extrapolative capacity is unknown.…”

Section: Process-unaware Modellingmentioning

confidence: 99%

Monitoring and modelling European carbon fluxes using atmospheric observations

van der Woude

View full text Add to dashboard Cite

Importance of the memory effect for assessing interannual variation in net ecosystem exchange

Cited by 5 publications

References 59 publications

Exploring the Potential of Long Short‐Term Memory Networks for Predicting Net CO₂ Exchange Across Various Ecosystems With Multi‐Source Data

Exploring the Potential of Long Short‐Term Memory Networks for Predicting Net CO₂ Exchange Across Various Ecosystems With Multi‐Source Data

Exploring the Potential of Long Short-Term Memory Networks for Predicting Net CO2 Exchange Across Various Ecosystems With Multi-Source Data

Monitoring and modelling European carbon fluxes using atmospheric observations

Contact Info

Product

Resources

About

Importance of the memory effect for assessing interannual variation in net ecosystem exchange

Cited by 5 publications

References 59 publications

Exploring the Potential of Long Short‐Term Memory Networks for Predicting Net CO2 Exchange Across Various Ecosystems With Multi‐Source Data

Exploring the Potential of Long Short‐Term Memory Networks for Predicting Net CO2 Exchange Across Various Ecosystems With Multi‐Source Data

Exploring the Potential of Long Short-Term Memory Networks for Predicting Net CO2 Exchange Across Various Ecosystems With Multi-Source Data

Monitoring and modelling European carbon fluxes using atmospheric observations

Contact Info

Product

Resources

About

Exploring the Potential of Long Short‐Term Memory Networks for Predicting Net CO₂ Exchange Across Various Ecosystems With Multi‐Source Data

Exploring the Potential of Long Short‐Term Memory Networks for Predicting Net CO₂ Exchange Across Various Ecosystems With Multi‐Source Data