Upscaling Net Ecosystem Exchange Over Heterogeneous Landscapes With Machine Learning

Reitz, Oliver; Leuchner, Michael; Schmidt, Marius; Ketzler, Gunnar; Graf, Alexander

doi:10.1029/2020jg005814

Cited by 29 publications

(24 citation statements)

References 82 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Estimates of carbon fluxes are highly dependent on both spatial and temporal scales, with spatial prediction being usually more difficult (55). According to our estimations, the Inland Pampa croplands seem to be acting as a carbon sink as a whole, even though the spatial distribution of NEE and NBP at the county and pixel-site scales is highly heterogeneous (Figure 5, Figure 6).…”

Section: Discussionmentioning

confidence: 73%

Cropland Net Ecosystem Exchange Estimation for the Inland Pampas (Argentina) Using EVI, Land Cover Maps, and Eddy Covariance Fluxes

Marconato¹,

Fernández²,

Posse³

2022

Front. Soil Sci.

View full text Add to dashboard Cite

Estimations of Net Ecosystem Exchange (NEE) are crucial to assess the carbon sequestration/carbon source capacity of agricultural systems. Although several global models have been built to describe carbon flux patterns based on flux tower data, South American ecosystems (and croplands in particular) are underrepresented in the databases used to calibrate these models, leading to large uncertainties in regional and global NEE estimation. Despite the fact that almost half of the land surface is used worldwide for agricultural activities, these models still do not include variables related to cropland management. Using enhanced vegetation index (EVI) derived from MODIS imagery (250 m) and monthly CO2 exchange from a 9-year record of an eddy covariance (EC) flux tower in a crop field in the Inland Pampas region, we developed regression models to predict monthly NEE. We tested whether including a term for crop identity/land cover as a categorical variable (maize, soybean, wheat, and fallow) could improve model capability in capturing monthly NEE dynamics. NEE measured at the flux tower site was scaled to croplands across the Inland Pampa using crop-type maps, from which annual NEE maps were generated for the 2018–2019, 2019–2020, and 2020–2021 agricultural campaigns. The model based solely on EVI showed to be a good predictor of monthly NEE for the study region (r2 = 0.78), but model adjustment was improved by including a term for crop identity (r2 = 0.83). A second set of maps was generated taking into account carbon exports during harvest to estimate Net Biome Productivity (NBP) at the county level. Crops across the region as a whole acted as a carbon sink during the three studied campaigns, although with highly heterogeneous spatial and temporal patterns. Between 60% and 80% of the carbon sequestered was exported during harvest, a large decrease from the carbon sequestration capacity estimated using just NEE, which further decreased if fossil carbon emissions from agricultural supplies are taken into account. Estimates presented in this study are a first step towards upscaling carbon fluxes at the regional scale in a South American cropland area, and could help to improve regional to global estimations of carbon fluxes and refine national greenhouse gas (GHG) inventories.

show abstract

Section: Discussionmentioning

confidence: 73%

Cropland Net Ecosystem Exchange Estimation for the Inland Pampas (Argentina) Using EVI, Land Cover Maps, and Eddy Covariance Fluxes

Marconato¹,

Fernández²,

Posse³

2022

Front. Soil Sci.

View full text Add to dashboard Cite

show abstract

“…This situation may be caused by the irregular and noisy data of FK and HM (Malik, Tikhamarine, Al‐Ansari, et al, 2021), but the specific reasons need to be deeply researched to improve the simulation accuracy. Additionally, though the universal models can overcome the spatial heterogeneity to predict regional target variables (Reitz et al, 2021), its performance should be further verified on a global scale with greater spatial heterogeneity.…”

Section: Discussionmentioning

confidence: 99%

Simulation of site‐scale water fluxes in desert and natural oasis ecosystems of the arid region in Northwest China

Xie

Luo

Hellwich

et al. 2021

Hydrological Processes

View full text Add to dashboard Cite

The study of water fluxes is important to better understand hydrological cycles in arid regions. Data‐driven machine learning models have been recently applied to water flux simulation. Previous studies have built site‐scale simulation models of water fluxes for individual sites separately, requiring a large amount of data from each site and significant computation time. For arid areas, there is no consensus as to the optimal model and variable selection method to simulate water fluxes. Using data from seven flux observation sites in the arid region of Northwest China, this study compared the performance of random forest (RF), support vector machine (SVM), back propagation neural network (BPNN), and multiple linear regression (MLR) models in simulating water fluxes. Additionally, the study investigated inter‐annual and seasonal variation in water fluxes and the dominant drivers of this variation at different sites. A universal simulation model for water flux was built using the RF approach and key variables as determined by MLR, incorporating data from all sites. Model performance of the SVM algorithm (R2 = 0.25–0.90) was slightly worse than that of the RF algorithm (R2 = 0.41–0.91); the BPNN algorithm performed poorly in most cases (R2 = 0.15–0.88). Similarly, the MLR results were limited and unreliable (R2 = 0.00–0.66). Using the universal RF model, annual water fluxes were found to be much higher than the precipitation received at each site, and natural oases showed higher fluxes than desert ecosystems. Water fluxes were highest during the growing season (May–September) and lowest during the non‐growing season (October–April). Furthermore, the dominant drivers of water flux variation were various among different sites, but the normalized difference vegetation index (NDVI), soil moisture and soil temperature were important at most sites. This study provides useful insights for simulating water fluxes in desert and oasis ecosystems, understanding patterns of variation and the underlying mechanisms. Besides, these results can make a contribution as the decision‐making basis to the water management in desert and oasis ecosystems.

show abstract

“…Lack of EC replication and a persistent inability to close the surface energy budget have been noted [67]. While these limitations have been addressed to the extent possible for the US-Ha1 data [6], details of standardization of QA/QC thresholds [35,68], and upscaling using remote sensing data [57,61,69,70] are outside of the scope of this study. Due to the short-term EC results for two additional towers reported for the US-Ha1 research area, the Hemlock tower (11 years) and clear-cut tower (6 years) [6], these results were omitted in the 28-year long term analysis presented.…”

Section: Study Limitationsmentioning

confidence: 99%

Science to Commerce: A Commercial-Scale Protocol for Carbon Trading Applied to a 28-Year Record of Forest Carbon Monitoring at the Harvard Forest

Bautista

Marino

Munger

2021

Land

View full text Add to dashboard Cite

Forest carbon sequestration offset protocols have been employed for more than 20 years with limited success in slowing deforestation and increasing forest carbon trading volume. Direct measurement of forest carbon flux improves quantification for trading but has not been applied to forest carbon research projects with more than 600 site installations worldwide. In this study, we apply carbon accounting methods, scaling hours to decades to 28-years of scientific CO2 eddy covariance data for the Harvard Forest (US-Ha1), located in central Massachusetts, USA, establishing commercial carbon trading protocols and applications for similar sites. We illustrate and explain transactions of high-frequency direct measurement for CO2 net ecosystem exchange (NEE, gC m−2 year−1) that track and monetize ecosystem carbon dynamics in contrast to approaches that rely on forest mensuration and growth models. NEE, based on eddy covariance methodology, quantifies loss of CO2 by ecosystem respiration accounted for as an unavoidable debit to net carbon sequestration. Retrospective analysis of the US-Ha1 NEE times series including carbon pricing, interval analysis, and ton-year exit accounting and revenue scenarios inform entrepreneur, investor, and landowner forest carbon commercialization strategies. CO2 efflux accounts for ~45% of US-Ha1 NEE, or an error of ~466% if excluded; however, the decades-old coupled human and natural system remains a financially viable net carbon sink. We introduce isoflux NEE for t13C16O2 and t12C18O16O to directly partition and quantify daytime ecosystem respiration and photosynthesis, creating new soil carbon commerce applications and derivative products in contrast to undifferentiated bulk soil carbon pool approaches. Eddy covariance NEE methods harmonize and standardize carbon commerce across diverse forest applications including, a New England, USA regional eddy covariance network, the Paris Agreement, and related climate mitigation platforms.

show abstract

Upscaling Net Ecosystem Exchange Over Heterogeneous Landscapes With Machine Learning

Cited by 29 publications

References 82 publications

Cropland Net Ecosystem Exchange Estimation for the Inland Pampas (Argentina) Using EVI, Land Cover Maps, and Eddy Covariance Fluxes

Cropland Net Ecosystem Exchange Estimation for the Inland Pampas (Argentina) Using EVI, Land Cover Maps, and Eddy Covariance Fluxes

Simulation of site‐scale water fluxes in desert and natural oasis ecosystems of the arid region in Northwest China

Science to Commerce: A Commercial-Scale Protocol for Carbon Trading Applied to a 28-Year Record of Forest Carbon Monitoring at the Harvard Forest

Contact Info

Product

Resources

About