scholarly journals On the parallelization of atmospheric inversions of CO<sub>2</sub> surface fluxes within a variational framework

2013 ◽  
Vol 6 (3) ◽  
pp. 783-790 ◽  
Author(s):  
F. Chevallier
Keyword(s):  
Transport Model ◽  
Spatial Scales ◽  
Computational Cost ◽  
Surface Fluxes ◽  
Atmospheric Measurements ◽  
Transport Modelling ◽  
Time Space ◽  
Variational Minimization ◽  
Analytical Approaches ◽  
Atmospheric Inversions

Abstract. The variational formulation of Bayes' theorem allows inferring CO2 sources and sinks from atmospheric concentrations at much higher time–space resolution than the ensemble or analytical approaches. However, it usually exhibits limited scalable parallelism. This limitation hinders global atmospheric inversions operated on decadal time scales and regional ones with kilometric spatial scales because of the computational cost of the underlying transport model that has to be run at each iteration of the variational minimization. Here, we introduce a physical parallelization (PP) of variational atmospheric inversions. In the PP, the inversion still manages a single physically and statistically consistent window, but the transport model is run in parallel overlapping sub-segments in order to massively reduce the computation wall-clock time of the inversion. For global inversions, a simplification of transport modelling is described to connect the output of all segments. We demonstrate the performance of the approach on a global inversion for CO2 with a 32 yr inversion window (1979–2010) with atmospheric measurements from 81 sites of the NOAA global cooperative air sampling network. In this case, we show that the duration of the inversion is reduced by a seven-fold factor (from months to days), while still processing the three decades consistently and with improved numerical stability.

scholarly journals On the parallelization of atmospheric inversions of CO<sub>2</sub> surface fluxes within a variational framework

2013 ◽  
Vol 6 (1) ◽  
pp. 37-57 ◽  
Author(s):  
F. Chevallier
Keyword(s):  
Transport Model ◽  
Spatial Scales ◽  
Computational Cost ◽  
Surface Fluxes ◽  
Atmospheric Measurements ◽  
Transport Modelling ◽  
Sources And Sinks ◽  
Variational Framework ◽  
Variational Minimization ◽  
Atmospheric Inversions

Abstract. The variational formulation of Bayes' theorem allows inferring CO2 sources and sinks from atmospheric concentrations at much higher space-time resolution than the ensemble approach or the analytical one. However, it usually exhibits limited scalable parallelism. This limitation hinders global atmospheric inversions operated on decadal time scales and regional ones with kilometric spatial scales, because of the computational cost of the underlying transport model that has to be run at each iteration of the variational minimization. Here, we introduce a Physical Parallelisation (PP) of variational atmospheric inversions. In the PP, the inversion still manages a single physically and statistically consistent window, but the transport model is run in parallel overlapping sub-segments in order to massively reduce the computation wall clock time of the inversion. For global inversions, a simplification of transport modelling is described to connect the output of all segments. We demonstrate the performance of the approach on a global inversion for CO2 with a 32-yr inversion window (1979–2010) with atmospheric measurements from 81 sites of the NOAA global cooperative air sampling network. In this case, we show that the duration of the inversion is reduced by a seven-fold factor (from months to days) while still processing the three decades consistently and with improved numerical stability.

scholarly journals Diagnostic methods for atmospheric inversions of long-lived greenhouse gases

2016 ◽  
Author(s):  
Anna M. Michalak ◽  
Nina A. Randazzo ◽  
Frédéric Chevallier
Keyword(s):  
Inverse Problems ◽  
Greenhouse Gases ◽  
Atmospheric Chemistry ◽  
Transport Model ◽  
Ghg Emissions ◽  
Surface Fluxes ◽  
Diagnostic Methods ◽  
Clear Understanding ◽  
Atmospheric Observations ◽  
Atmospheric Inversions

Abstract. The ability to predict the trajectory of climate change requires a clear understanding of the emissions and uptake (a.k.a. surface fluxes) of long-lived greenhouse gases (GHGs). Furthermore, the development of climate policies is driving a need to constrain the budgets of anthropogenic GHG emissions. Inverse problems that couple atmospheric observations of GHG concentrations with an atmospheric chemistry and transport model have increasingly been used to gain insights into surface fluxes. Given the inherent technical challenges associated with their solution, it is imperative that objective approaches exist for the evaluation of such inverse problems. Because direct observation of fluxes at compatible spatiotemporal scales is rarely possible, diagnostics tools must rely on indirect measures. Here we review diagnostics that have been implemented in recent studies, and discuss their use in informing adjustments to model setup. We group the diagnostics along a continuum starting with those that are most closely related to the scientific question being targeted, and ending with those most closely tied to the statistical and computational setup of the inversion. We thus begin with diagnostics based on assessments against independent information (e.g., unused atmospheric observations, large-scale scientific constraints), followed by statistical diagnostics of inversion results, diagnostics based on sensitivity tests and analyses of robustness (e.g., tests focusing on the chemistry and transport model, the atmospheric observations, or the statistical and computational framework), and close with the use of synthetic data experiments (a.k.a. observing system simulation experiments (OSSEs)). We find that existing diagnostics provide a crucial toolbox for evaluating and improving flux estimates, but, not surprisingly, cannot overcome the fundamental challenges associated with limited atmospheric observations or the lack of direct flux measurements at compatible scales. As atmospheric inversions are increasingly expected to contribute to national reporting of GHG emissions, the need for developing and implementing robust and transparent evaluation approaches will only grow.

scholarly journals Objective evaluation of surface- and satellite-driven carbon dioxide atmospheric inversions

2019 ◽  
Vol 19 (22) ◽  
pp. 14233-14251 ◽  
Author(s):  
Frédéric Chevallier ◽  
Marine Remaud ◽  
Christopher W. O'Dell ◽  
David Baker ◽  
Philippe Peylin ◽  
...  
Keyword(s):  
Carbon Dioxide ◽  
Large Scale ◽  
Transport Model ◽  
Spatial Scales ◽  
Objective Evaluation ◽  
Annual Growth Rate ◽  
Surface Network ◽  
Carbon Dioxide Co2 ◽  
Aircraft Data ◽  
Atmospheric Inversions

Abstract. We study an ensemble of six multi-year global Bayesian carbon dioxide (CO2) atmospheric inversions that vary in terms of assimilated observations (either column retrievals from one of two satellites or surface air sample measurements) and transport model. The time series of inferred annual fluxes are first compared with each other at various spatial scales. We then objectively evaluate the small inversion ensemble based on a large dataset of accurate aircraft measurements in the free troposphere over the globe, which are independent of all assimilated data. The measured variables are connected with the inferred fluxes through mass-conserving transport in the global atmosphere and are part of the inversion results. Large-scale annual fluxes estimated from the bias-corrected land retrievals of the second Orbiting Carbon Observatory (OCO-2) differ greatly from the prior fluxes, but are similar to the fluxes estimated from the surface network within the uncertainty of these surface-based estimates. The OCO-2-based and surface-based inversions have similar performance when projected in the space of the aircraft data, but the relative strengths and weaknesses of the two flux estimates vary within the northern and tropical parts of the continents. The verification data also suggest that the more complex and more recent transport model does not improve the inversion skill. In contrast, the inversion using bias-corrected retrievals from the Greenhouse Gases Observing Satellite (GOSAT) or, to a larger extent, a non-Bayesian inversion that simply adjusts a recent bottom-up flux estimate with the annual growth rate diagnosed from marine surface measurements both estimate much different fluxes and fit the aircraft data less. Our study highlights a way to rate global atmospheric inversions. Without any general claim regarding the usefulness of all OCO-2 retrieval datasets vs. all GOSAT retrieval datasets, it still suggests that some satellite retrievals can now provide inversion results that are, despite their uncertainty, comparable with respect to credibility to traditional inversions using the accurate but sparse surface network and that are therefore complementary for studies of the global carbon budget.

scholarly journals Objectified quantification of uncertainties in Bayesian atmospheric inversions

2015 ◽  
Vol 8 (5) ◽  
pp. 1525-1546 ◽  
Author(s):  
A. Berchet ◽  
I. Pison ◽  
F. Chevallier ◽  
P. Bousquet ◽  
J.-L. Bonne ◽  
...  
Keyword(s):  
Expert Knowledge ◽  
Spatial Scales ◽  
Surface Fluxes ◽  
Large Set ◽  
Error Statistics ◽  
Observation Operator ◽  
Probability Of Occurrence ◽  
Error Distributions ◽  
The Impact ◽  
Atmospheric Inversions

Abstract. Classical Bayesian atmospheric inversions process atmospheric observations and prior emissions, the two being connected by an observation operator picturing mainly the atmospheric transport. These inversions rely on prescribed errors in the observations, the prior emissions and the observation operator. When data pieces are sparse, inversion results are very sensitive to the prescribed error distributions, which are not accurately known. The classical Bayesian framework experiences difficulties in quantifying the impact of mis-specified error distributions on the optimized fluxes. In order to cope with this issue, we rely on recent research results to enhance the classical Bayesian inversion framework through a marginalization on a large set of plausible errors that can be prescribed in the system. The marginalization consists in computing inversions for all possible error distributions weighted by the probability of occurrence of the error distributions. The posterior distribution of the fluxes calculated by the marginalization is not explicitly describable. As a consequence, we carry out a Monte Carlo sampling based on an approximation of the probability of occurrence of the error distributions. This approximation is deduced from the well-tested method of the maximum likelihood estimation. Thus, the marginalized inversion relies on an automatic objectified diagnosis of the error statistics, without any prior knowledge about the matrices. It robustly accounts for the uncertainties on the error distributions, contrary to what is classically done with frozen expert-knowledge error statistics. Some expert knowledge is still used in the method for the choice of an emission aggregation pattern and of a sampling protocol in order to reduce the computation cost. The relevance and the robustness of the method is tested on a case study: the inversion of methane surface fluxes at the mesoscale with virtual observations on a realistic network in Eurasia. Observing system simulation experiments are carried out with different transport patterns, flux distributions and total prior amounts of emitted methane. The method proves to consistently reproduce the known "truth" in most cases, with satisfactory tolerance intervals. Additionally, the method explicitly provides influence scores and posterior correlation matrices. An in-depth interpretation of the inversion results is then possible. The more objective quantification of the influence of the observations on the fluxes proposed here allows us to evaluate the impact of the observation network on the characterization of the surface fluxes. The explicit correlations between emission aggregates reveal the mis-separated regions, hence the typical temporal and spatial scales the inversion can analyse. These scales are consistent with the chosen aggregation patterns.

scholarly journals Simulating CH<sub>4</sub> and CO<sub>2</sub> over South and East Asia using the zoomed chemistry transport model LMDzINCA

2017 ◽  
Author(s):  
Xin Lin ◽  
Philippe Ciais ◽  
Philippe Bousquet ◽  
Michel Ramonet ◽  
Yi Yin ◽  
...  
Keyword(s):  
East Asia ◽  
Transport Model ◽  
Model Performance ◽  
Horizontal Resolution ◽  
Surface Fluxes ◽  
Vertical Resolution ◽  
Monitoring Networks ◽  
Atmospheric Measurements ◽  
Monsoon Period ◽  
Chemistry Transport Model

Abstract. The increasing availability of atmospheric measurements of greenhouse gases (GHGs) from surface stations can improve the retrieval of their fluxes at higher spatial and temporal resolutions by inversions, provided that chemistry transport models are able to properly represent the variability of concentrations observed at different stations. South and East Asia (SEA) is a region with large and very uncertain emissions of carbon dioxide (CO2) and methane (CH4), the most potent anthropogenic GHGs. Monitoring networks have expanded greatly during the past decade in this region, which should contribute to reducing uncertainties in estimates of regional GHG budgets. In this study, we simulate concentrations of CH4 and CO2 using a zoomed version of the global chemistry transport model LMDzINCA during the period 2006–2013. The zoomed version has a fine horizontal resolution of ~ 0.66° in longitude and ~0.51° in latitude over SEA and a coarser resolution elsewhere. The concentrations of CH4 and CO2 simulated from the zoomed model (abbreviated as ‘ZASIA’) are compared to those from the same model but with a uniform regular grid of 2.50° in longitude and 1.27° in latitude (abbreviated as ‘REG’), both having the same vertical 19 sigma pressure levels and prescribed with the same biogenic and anthropogenic fluxes. Model performance is evaluated for annual gradients between sites, seasonal, synoptic and diurnal variations, against a new dataset including 30 surface stations over SEA and adjacent regions. Our results show that, when prescribed with identical surface fluxes, compared to REG, the ZASIA version moderately improves the representation of CH4 mean annual gradients between stations as well as the seasonal and synoptic variations of this trace gas within the zoomed region. This moderate improvement probably results from reduction of representation errors and a better description of the CH4 concentration gradients related to the skewed spatial distribution of surface CH4 emissions, suggesting that the zoom transport model will be better suited for inversions of CH4 fluxes in SEA. With the relatively coarse vertical resolution and low-frequency (monthly) prescribed fluxes, the model generally does not capture the diurnal cycle of CH4 at most stations even with its zoomed configuration, emphasizing the need to increase the vertical resolution, and to improve parameterizations of turbulent diffusion in the planetary boundary layer and deep convection during the monsoon period. The model performance for CH4 is better than that for CO2 at any temporal scale, likely due to inaccuracies in the CO2 fluxes prescribed in this study.

scholarly journals Objective evaluation of surface- and satellite-driven CO<sub>2</sub> atmospheric inversions

2019 ◽  
Author(s):  
Frédéric Chevallier ◽  
Marine Remaud ◽  
Christopher W. O'Dell ◽  
David Baker ◽  
Philippe Peylin ◽  
...  
Keyword(s):  
Large Scale ◽  
Transport Model ◽  
Spatial Scales ◽  
Objective Evaluation ◽  
Annual Growth Rate ◽  
Marine Surface ◽  
Surface Measurements ◽  
Surface Network ◽  
Aircraft Data ◽  
Atmospheric Inversions

Abstract. We study an ensemble of six multi-year global Bayesian CO2 atmospheric inversions that vary in terms of assimilated observations (either column retrievals from one of two satellites or surface air sample measurements) and transport model. The time series of inferred annual fluxes are first compared with each other at various spatial scales. We then objectively evaluate the small inversion ensemble based on a large dataset of accurate aircraft measurements in the free troposphere over the globe, that are independent from all assimilated data. The measured variables are connected with the inferred fluxes through mass-conserving transport in the global atmosphere and are part of the inversion results. Large-scale annual fluxes estimated from the bias-corrected land retrievals of the second Orbiting Carbon Observatory (OCO-2) differ from the prior fluxes much, but are similar to the fluxes estimated from the surface network within the uncertainty of these surface-based estimates. The OCO-2- and surface-based inversions have similar performance when projected in the space of the aircraft data, but relative strengths and weaknesses of the two flux estimates vary within the Northern and Tropical parts of the continents. The verification data also suggests that the more complex and more recent transport model does not improve the inversion skill. In contrast, the inversion using bias-corrected retrievals from the Greenhouse Gases Observing Satellite (GOSAT) or, to a larger extent, a non-Bayesian inversion that simply adjusts a recent bottom-up flux estimate with the annual growth rate diagnosed from marine surface measurements, estimate much different fluxes and fit the aircraft data less. Our study highlights a way to rate global atmospheric inversions. It suggests that some satellite retrievals can now provide inversion results that are, despite their uncertainty, comparable in credibility to traditional inversions using the accurate but sparse surface network and that are therefore complementary for studies of the global carbon budget.

scholarly journals Constraining tropospheric mixing timescales using airborne observations and numerical models

2003 ◽  
Vol 3 (2) ◽  
pp. 1213-1245 ◽  
Author(s):  
P. Good ◽  
C. Giannakopoulos ◽  
F. M. O’Connor ◽  
S. R. Arnold ◽  
M. de Reus ◽  
...  
Keyword(s):  
Spatial Scale ◽  
Time Scales ◽  
Transport Model ◽  
Numerical Models ◽  
Spatial Scales ◽  
Lagrangian Model ◽  
Small Scale ◽  
Tracer Transport ◽  
Transport Modelling ◽  
Trajectory Length

Abstract. A technique is demonstrated for estimating atmospheric mixing time-scales from in-situ data, using a Lagrangian model initialised from an Eulerian chemical transport model (CTM). This method is applied to airborne tropospheric CO observations taken during seven flights of the Mediterranean Intensive Oxidant Study (MINOS) campaign, of August 2001. The time-scales derived, correspond to mixing applied at the spatial scale of the CTM grid. Specifically, they are upper bound estimates of the mix-down lifetime that should be imposed for a Lagrangian model to reproduce the observed small-scale tracer structure. They are relevant to the family of hybrid Lagrangian-Eulerian models, which impose Eulerian grid mixing to an underlying Lagrangian model. The method uses the fact that in Lagrangian tracer transport modelling, the mixing spatial and temporal scales are decoupled: the spatial scale is determined by the resolution of the initial tracer field, and the time scale by the trajectory length. The chaotic nature of lower-atmospheric advection results in the continuous generation of smaller spatial scales, a process terminated in the real atmosphere by mixing. Thus, a mix-down lifetime can be estimated by varying trajectory length so that the model reproduces the observed amount of small-scale tracer structure. Selecting a trajectory length is equivalent to choosing a mixing timescale. For the cases studied, the results are very insensitive to CO photochemical change calculated along the trajectories. The method is most appropriate for relatively homogeneous regions, i.e. it is not too important to account for changes in aircraft altitude or the positioning of stratospheric intrusions, so that small scale structure is easily distinguished. The chosen flights showed a range of mix-down time upper limits: 1 and 3 days for 8 August and 3 August, due to recent convective and boundary layer mixing respectively, and 7–9 days for 16, 17, 22a, 22c and 24 August. For the flight of 3 August, the observed concentrations result from a complex set of transport histories, and the models are used to interpret the observed structure, while illustrating where more caution is required with this method of estimating mix-down lifetimes.

scholarly journals Diagnostic methods for atmospheric inversions of long-lived greenhouse gases

2017 ◽  
Vol 17 (12) ◽  
pp. 7405-7421 ◽  
Author(s):  
Anna M. Michalak ◽  
Nina A. Randazzo ◽  
Frédéric Chevallier
Keyword(s):  
Inverse Problems ◽  
Greenhouse Gases ◽  
Atmospheric Chemistry ◽  
Transport Model ◽  
Ghg Emissions ◽  
Surface Fluxes ◽  
Diagnostic Methods ◽  
Clear Understanding ◽  
Atmospheric Observations ◽  
Atmospheric Inversions

Abstract. The ability to predict the trajectory of climate change requires a clear understanding of the emissions and uptake (i.e., surface fluxes) of long-lived greenhouse gases (GHGs). Furthermore, the development of climate policies is driving a need to constrain the budgets of anthropogenic GHG emissions. Inverse problems that couple atmospheric observations of GHG concentrations with an atmospheric chemistry and transport model have increasingly been used to gain insights into surface fluxes. Given the inherent technical challenges associated with their solution, it is imperative that objective approaches exist for the evaluation of such inverse problems. Because direct observation of fluxes at compatible spatiotemporal scales is rarely possible, diagnostics tools must rely on indirect measures. Here we review diagnostics that have been implemented in recent studies and discuss their use in informing adjustments to model setup. We group the diagnostics along a continuum starting with those that are most closely related to the scientific question being targeted, and ending with those most closely tied to the statistical and computational setup of the inversion. We thus begin with diagnostics based on assessments against independent information (e.g., unused atmospheric observations, large-scale scientific constraints), followed by statistical diagnostics of inversion results, diagnostics based on sensitivity tests, and analyses of robustness (e.g., tests focusing on the chemistry and transport model, the atmospheric observations, or the statistical and computational framework), and close with the use of synthetic data experiments (i.e., observing system simulation experiments, OSSEs). We find that existing diagnostics provide a crucial toolbox for evaluating and improving flux estimates but, not surprisingly, cannot overcome the fundamental challenges associated with limited atmospheric observations or the lack of direct flux measurements at compatible scales. As atmospheric inversions are increasingly expected to contribute to national reporting of GHG emissions, the need for developing and implementing robust and transparent evaluation approaches will only grow.

Better constraining the CO2 plant uptake at global scale: joint assimilation of COS and CO2 atmospheric measurements into a transport model.

2020 ◽  
Author(s):  
Marine Remaud ◽  
Frédéric Chevallier ◽  
Philippe Peylin ◽  
Antoine Berchet ◽  
Fabienne Maignan
Keyword(s):  
Plant Uptake ◽  
Land Surface ◽  
Transport Model ◽  
Atmospheric Transport ◽  
Gross Primary Production ◽  
Surface Fluxes ◽  
Surface Model ◽  
Atmospheric Measurements ◽  
Inverse Systems ◽  
Atmospheric Transport Model

&lt;p&gt;Inverse systems that assimilate atmospheric carbon dioxide measurements (CO2) into a global atmospheric transport model, are commonly used together with anthropogenic emission inventories to infer net biospheric surface fluxes. However, when assimilating CO2 measurements only, the respiration fluxes cannot be disentangled from the gross primary production (GPP) fluxes, leaving few possibilities to interpret the inferred fluxes from a mechanistic point of view. Measurements of carbonyl sulfide (COS) may help to fill this gap: COS has similar diffusion pathway inside leaves&amp;#160;as CO2 but is not re-emitted into the atmosphere by the plant respiration. We explore here the benefit of assimilating both COS and CO2 measurements into the LMDz atmospheric transport model to constrain GPP and respiration fluxes separately. To this end, we develop an analytic inverse system based on the 14 Plant functional Type (PFTs) as defined in the ORCHIDEE land surface model. The vegetation uptake of COS is parameterized as a linear function of GPP and of the leaf relative uptake (LRU), which is the ratio of COS to CO2 deposition velocities in plants. A new parameterization of the atmosphere soil exchanges is also included. We use the system to optimize GPP and respiration fluxes separately at the seasonal scale over the globe. The results lead to a balanced COS global budget and a seasonality of the COS fluxes in better agreement with observations. We find a large sensitivity of the partition between the ocean emissions and the COS plant uptake to the LRU parameterizations.&lt;/p&gt;

scholarly journals Objectified quantification of uncertainties in Bayesian atmospheric inversions

2014 ◽  
Vol 7 (4) ◽  
pp. 4777-4827 ◽  
Author(s):  
A. Berchet ◽  
I. Pison ◽  
F. Chevallier ◽  
P. Bousquet ◽  
J.-L. Bonne ◽  
...  
Keyword(s):  
Expert Knowledge ◽  
Spatial Scales ◽  
Surface Fluxes ◽  
Error Statistics ◽  
Observation Operator ◽  
Aggregation Pattern ◽  
Error Distributions ◽  
The Impact ◽  
Meso Scale ◽  
Atmospheric Inversions

Abstract. Classical Bayesian atmospheric inversions process atmospheric observations and prior emissions, the two being connected by an observation operator picturing mainly the atmospheric transport. These inversions rely on prescribed errors in the observations, the prior emissions and the observation operator. At the meso-scale, inversion results are very sensitive to the prescribed error distributions, which are not accurately known. The classical Bayesian framework experiences difficulties in quantifying the impact of mis-specified error distributions on the optimized fluxes. In order to cope with this issue, we rely on recent research results and enhance the classical Bayesian inversion framework through a marginalization on all the plausible errors that can be prescribed in the system. The marginalization consists in computing inversions for all possible error distributions weighted by the probability of occurence of the error distributions. The posterior distribution of the fluxes calculated by the marginalization is complicated and not explicitly describable. We then carry out a Monte-Carlo sampling relying on an approximation of the probability of occurence of the error distributions. This approximation is deduced from the well-tested algorithm of the Maximum of Likelihood. Thus, the marginalized inversion relies on an automatic objectified diagnosis of the error statistics, without any prior knowledge about the matrices. It robustly includes the uncertainties on the error distributions, contrary to what is classically done with frozen expert-knowledge error statistics. Some expert knowledge is still used in the method for the choice of emission aggregation pattern and sampling protocol in order to reduce the computation costs of the method. The relevance and the robustness of the method is tested on a case study: the inversion of methane surface fluxes at the meso-scale with real observation sites in Eurasia. Observing System Simulation Experiments are carried out with different transport patterns, flux distributions and total prior amounts of emitted gas. The method proves to consistently reproduce the known "truth" in most cases, with satisfactory tolerance intervals. Additionnaly, the method explicitly provides influence scores and posterior correlation matrices. An in-depth interpretation of the inversion results is then possible. The more objective quantification of the influence of the observations on the fluxes proposed here allows us to evaluate the impact of the observation network on the characterization of the surface fluxes. The explicit correlations between emission regions reveal the mis-separated regions, hence the typical temporal and spatial scales the inversion can analyze. These scales proved to be consistent with the chosen aggregation patterns.

Sign in / Sign up

Export Citation Format

Share Document