Annual sums of carbon dioxide exchange over a heterogeneous urban landscape through machine learning based gap-filling

Menzer, Olaf; Meiring, Wendy; Kyriakidis, Phaedon C.; McFadden, J. P.

doi:10.1016/j.atmosenv.2014.11.006

Cited by 23 publications

(18 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It was found that the test set R 2 obtained by the ANN method was 0.827. The test set R 2 obtained by Menzer et al (2015) using the ANN method was 0.740, which was 8% lower than this study. The R 2 value obtained by the GP method was 0.844, which was 6% higher than that of Menzer et al (2015); R 2 obtained by the RBF method was 0.790, which was 39% higher than that of Järvi et al (2012) and 3% higher than that of Menzer et al (2015).…”

Section: Discussioncontrasting

confidence: 75%

“…The test set R 2 obtained by Menzer et al (2015) using the ANN method was 0.740, which was 8% lower than this study. The R 2 value obtained by the GP method was 0.844, which was 6% higher than that of Menzer et al (2015); R 2 obtained by the RBF method was 0.790, which was 39% higher than that of Järvi et al (2012) and 3% higher than that of Menzer et al (2015). The RMSE of the test set obtained by the five methods ranged from 2.04 to 2.50 mmol m −2 s −1 , which was smaller than that of Järvi et al (2012) and Menzer et al (2015).…”

Section: Discussioncontrasting

confidence: 75%

See 1 more Smart Citation

Uncertainty in gap filling and estimating the annual sum of carbon dioxide exchange for the desert Tugai forest, Ebinur Lake Basin, Northwest China

Teng

Wang

et al. 2020

PeerJ

View full text Add to dashboard Cite

In most eddy covariance (EC) studies, carbon flux measurements have a high defect rate for a variety of reasons. Obtaining the annual sum of carbon dioxide exchange requires imputation of data gaps with high precision and accuracy. This study used five methods to fill the gaps in carbon flux data and estimate the total annual carbon dioxide exchange of the Tugai forest in the arid desert ecosystem of Ebinur Lake Basin, Northwest China. The Monte Carlo method was used to estimate the random error and bias caused by gap filling. The results revealed that (1) there was a seasonal difference in the friction velocity threshold of nighttime flux, with values in the growing season and non-growing season of 0.12 and 0.10 m/s, respectively; (2) the five gap-filling methods explained 77–84% of the data variability in the fluxes, and the random errors estimated by these methods were characterized by non-normality and leptokurtic heavy tail features, following the Laplacian (or double-exponential) distribution; (3) estimates of the annual sum of carbon dioxide exchange using the five methods at the study site in 2015 ranged from −178.25 to −155.21 g C m−2 year−1, indicating that the Tugai forest in the Ebinur Lake Basin is a net carbon sink. The standard deviation of the total annual carbon dioxide exchange sums estimated by the five different methods ranged from 3.15 to 19.08 g C m−2 year−1, with bias errors ranging from −13.69 to 14.05 g C m−2 year−1. This study provides a theoretical basis for the carbon dioxide exchange and carbon source/sink assessment of the Tugai forest in an arid desert ecosystem. In order to explore the functioning of the Tugai forest at this site, a greater understanding of the underlying ecological mechanisms is necessary.

show abstract

Section: Discussioncontrasting

confidence: 75%

Section: Discussioncontrasting

confidence: 75%

Uncertainty in gap filling and estimating the annual sum of carbon dioxide exchange for the desert Tugai forest, Ebinur Lake Basin, Northwest China

Teng

Wang

et al. 2020

PeerJ

View full text Add to dashboard Cite

show abstract

“…As an outcome of the Hestia effort, a large multifaceted effort, the Indianapolis Flux Experiment (INFLUX), emerged (Whetstone, 2018;Davis et al, 2017). INFLUX aims to advance quantification and associated uncertainties of urban CO 2 and CH 4 emissions by integrating a high-resolution bottom-up emission data product, such as Hestia, with atmospheric concentration measurements (Turnbull et al, 2015;Miles et al, 2017;Richardson et al, 2017), flux measurements (Cambaliza et al, , 2015Heimburger et al, 2017), and atmospheric inverse modeling. In addition to its use as a key constraint in the INFLUX atmospheric inverse estimation (Lauvaux et al, 2016), Hestia has been informed by atmospheric observations making it useable as a standalone high-resolution flux estimate offering a detailed spacetime understanding of urban emissions.…”

Section: Discussionmentioning

confidence: 99%

“…Through a combination of atmospheric measurements, atmospheric transport modeling, and datadriven "bottom-up" estimation, the scientific community is exploring different methodologies, applications, and uncertainty estimation of these approaches (Hutyra et al, 2014). Atmospheric monitoring includes ground-based CO 2 concentration measurements (McKain et al, 2012;Djuricin et al, 2010;Miles et al, 2017;Turnbull et al, 2015;Verhulst et al, 2017), ground-based eddy flux (i.e., emissions of CO 2 into the atmosphere and/or CO 2 being removed from the atmospheric by vegetation) measurements (Christen, 2014;Grimmond et al, 2002;Menzer et al, 2015;Velasco and Roth, 2010;Velasco et al, 2005), aircraft-based flux measurements (Mays et al, 2009;Cambaliza et al, , 2015, and whole column abundances from both ground-and spacebased remote-sensing platforms (Wunch et al, 2009;Kort et al, 2012;Wong et al, 2015;Schwandner et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

The Hestia fossil fuel CO2 emissions data product for the Los Angeles megacity (Hestia-LA)

Gurney

Patarasuk²,

Liang

et al. 2019

Earth Syst. Sci. Data

View full text Add to dashboard Cite

Abstract. High-resolution bottom-up estimation provides a detailed guide for city greenhouse gas mitigation options, offering details that can increase the economic efficiency of emissions reduction options and synergize with other urban policy priorities at the human scale. As a critical constraint to urban atmospheric CO2 inversion studies, bottom-up spatiotemporally explicit emissions data products are also necessary to construct comprehensive urban CO2 emission information systems useful for trend detection and emissions verification. The “Hestia Project” is an effort to provide bottom-up granular fossil fuel (FFCO2) emissions for the urban domain with building/street and hourly space–time resolution. Here, we report on the latest urban area for which a Hestia estimate has been completed – the Los Angeles megacity, encompassing five counties: Los Angeles County, Orange County, Riverside County, San Bernardino County and Ventura County. We provide a complete description of the methods used to build the Hestia FFCO2 emissions data product for the years 2010–2015. We find that the LA Basin emits 48.06 (±5.3) MtC yr−1, dominated by the on-road sector. Because of the uneven spatial distribution of emissions, 10 % of the largest-emitting grid cells account for 93.6 %, 73.4 %, 66.2 %, and 45.3 % of the industrial, commercial, on-road, and residential sector emissions, respectively. Hestia FFCO2 emissions are 10.7 % larger than the inventory estimate generated by the local metropolitan planning agency, a difference that is driven by the industrial and electricity production sectors. The detail of the Hestia-LA FFCO2 emissions data product offers the potential for highly targeted, efficient urban greenhouse gas emissions mitigation policy. The Hestia-LA v2.5 emissions data product can be downloaded from the National Institute of Standards and Technology repository (https://doi.org/10.18434/T4/1502503, Gurney et al., 2019).

show abstract

“…Artificial neural networks (ANNs) form a category of non-parametric models that have frequently been used to fill gaps in EC CO 2 flux time series. Mostly, multilayer perceptrons (MLP) were chosen (Papale and Valentini, 2003;Moffat et al, 2007;Moffat, 2012;Järvi et al, 2012;Pypker et al, 2013;Menzer et al, 2015) while other authors utilized radial basis function (RBF) networks (Schmidt et al, 2008;Kordowski and Kuttler, 2010;Menzer et al, 2015). For CH 4 fluxes, MLP models are described by Dengel et al (2013), Deshmukh et al (2014), Knox et al (2015) and Goodrich et al (2015) as well as a special kind of RBF network, a generalized regression neural network (GRNN), by Zhu et al (2013).…”

mentioning

confidence: 99%

Comparison of eddy covariance CO2 and CH4 fluxes from mined and recently rewetted sections in a NW German cutover bog

Holl¹,

Pfeiffer²,

Kutzbach³

2019

Preprint

View full text Add to dashboard Cite

Abstract. With respect to their role in the global carbon cycle, natural peatlands are characterized by their ability to sequester atmospheric carbon. This trait is strongly connected to the water regime of these ecosystems. Large parts of the soil profile in natural peatlands are water-saturated, leading to anoxic conditions and to a diminished decomposition of plant litter. In functioning peatlands, the rate of carbon fixation by plant photosynthesis is larger than the decomposition rate of dead organic material. Over time, the amount of carbon that remains in the soil and is not converted back to carbon dioxide grows. Land use of peatlands often goes along with water level manipulations and thereby with alterations of carbon flux dynamics. In this study, carbon dioxide (CO2) and methane (CH4) flux measurements from a bog site in NW Germany that has been heavily degraded by peat mining are presented. Two contrasting types of management have been implemented at the site: (1) drainage during ongoing peat-harvesting on one half of the central bog area and (2) rewetting on the other half that had been taken out of use shortly before measurements commenced. The presented two-year data set was collected with an eddy covariance (EC) system set up on a central railroad dam that divides the two halves of the (former) peat harvesting area. We used footprint analysis to split the obtained CO2 and CH4 flux time series into data characterizing the gas exchange dynamics of both contrasting land use types individually. The time series gaps resulting from data division were filled using the response of artificial neural networks (ANNs) to environmental variables, footprint variability and fuzzy transformations of seasonal and diurnal cyclicity. We used the gap-filled gas flux time series from two consecutive years to evaluate the impact of rewetting on the annual vertical carbon balances of the cutover bog. Rewetting had a considerable effect on the annual carbon fluxes and led to increased CH4 and decreased CO2 release. The larger relative difference between cumulative CO2 fluxes from the rewetted (22&#8201;&#177;&#8201;7&#8201;mol&#8201;m&#8722;2&#8201;a&#8722;1) and drained (13&#8201;&#177;&#8201;6&#8201;mol&#8201;m&#8722;2&#8201;a&#8722;1) section occurred in the second observed year when rewetting apparently reduced CO2 emissions by 40&#8201;%. The absolute difference in annual CH4 flux sums was more similar between both years while the relative difference of CH4 release between the rewetted (0.83&#8201;&#177;&#8201;0.15&#8201;mol&#8201;m&#8722;2&#8201;a&#8722;1) and drained (0.45&#8201;&#177;&#8201;0.11&#8201;mol&#8201;m&#8722;2&#8201;a&#8722;1) section was larger in the first observed year indicating a maximum increase of annual CH4 release of 84&#8201;% caused by rewetting; at this particular site during the study period.

show abstract

Annual sums of carbon dioxide exchange over a heterogeneous urban landscape through machine learning based gap-filling

Cited by 23 publications

References 57 publications

Uncertainty in gap filling and estimating the annual sum of carbon dioxide exchange for the desert Tugai forest, Ebinur Lake Basin, Northwest China

Uncertainty in gap filling and estimating the annual sum of carbon dioxide exchange for the desert Tugai forest, Ebinur Lake Basin, Northwest China

The Hestia fossil fuel CO<sub>2</sub> emissions data product for the Los Angeles megacity (Hestia-LA)

Comparison of eddy covariance CO<sub>2</sub> and CH<sub>4</sub> fluxes from mined and recently rewetted sections in a NW German cutover bog

Contact Info

Product

Resources

About

Annual sums of carbon dioxide exchange over a heterogeneous urban landscape through machine learning based gap-filling

Cited by 23 publications

References 57 publications

Uncertainty in gap filling and estimating the annual sum of carbon dioxide exchange for the desert Tugai forest, Ebinur Lake Basin, Northwest China

Uncertainty in gap filling and estimating the annual sum of carbon dioxide exchange for the desert Tugai forest, Ebinur Lake Basin, Northwest China

The Hestia fossil fuel CO&lt;sub&gt;2&lt;/sub&gt; emissions data product for the Los Angeles megacity (Hestia-LA)

Comparison of eddy covariance CO&lt;sub&gt;2&lt;/sub&gt; and CH&lt;sub&gt;4&lt;/sub&gt; fluxes from mined and recently rewetted sections in a NW German cutover bog

Contact Info

Product

Resources

About

The Hestia fossil fuel CO<sub>2</sub> emissions data product for the Los Angeles megacity (Hestia-LA)

Comparison of eddy covariance CO<sub>2</sub> and CH<sub>4</sub> fluxes from mined and recently rewetted sections in a NW German cutover bog