We investigate the potential for aircraft-based top-down emission rate retrieval over- and under-estimation using a regional chemical transport model, the Global Environmental Multiscale-Modeling Air-Quality and CHemistry (GEM-MACH). In our investigations we consider the application of the mass-balance approach in the Top-down Emission Rate Retrieval Algorithm (TERRA). Aircraft-based mass-balance retrieval methodologies such as TERRA require relatively constant meteorological conditions and source emission rates to reliably estimate emission rates from aircraft observations. Avoiding cases where meteorology and emission rates change significantly is one means of reducing emissions retrieval uncertainty, and quantitative metrics that may be used for retrieval accuracy estimation are therefore desirable. Using these metrics has the potential to greatly improve emission rate retrieval accuracy. Here, we investigate the impact of meteorological variability on mass-balance emission rate retrieval accuracy by using model-simulated fields as a proxy for real-world chemical and meteorological fields, in which virtual aircraft sampling of the GEM-MACH output was used for top-down mass balance estimates. We also explore the impact of upwind emissions from nearby sources on the accuracy of the retrieved emission rates. This approach allows the state of the atmosphere used for top-down estimates to be characterized in time and 3D space; the input meteorology and emissions are “known”, and thus potential means for improving emission rate retrievals and determining the factors affecting retrieval accuracy may be investigated. We found that emissions retrieval accuracy is correlated with three key quantitative criteria, evaluated a priori from forecasts and/or from observations during the sampling period: (1) changes to the atmospheric stability (described as the change in gradient Richardson number), (2) variations in the direction of transport, as a result of plume vertical motion and in the presence of vertical wind shear, and (3) the combined effect of the upwind-to-downwind concentration ratio and the upwind-to-downwind concentration standard deviations. We show here that cases where these criteria indicate high temporal variability and/or high upwind emissions can result in “storage-and-release” events within the sampled region (control volume), which decrease emission rate retrieval accuracy. Storage-and-release events may contribute the bulk of mass-balance emission rate retrieval under- and over-estimates, ranging in the tests carried out here from
Aircraft-based measurements of air pollutants provide an effective top-down approach for estimating total integrated emissions of sources ranging from individual stacks to large industrial complexes and cities. These top-down estimates are commonly employed for the evaluation of bottom-up emissions reported to inventories and can provide insights into air quality and health impacts of anthropogenic emissions. Top-down estimates can also be used to provide highly resolved emissions data for input into air-quality models. Several studies have utilized aircraft-based top-down emission rate retrievals in the past
The potential for measurement error to affect retrieval accuracy has been discussed in the literature.
These and other similar studies, employed either mass transfer (e.g., the single-screen flight approach where the horizontal mass flux through a downwind vertical plane is equated to the upwind source emission rates) or a mass-balance approach (box flights) to infer point and area source emission rates from aircraft-measured data. Aircraft-based estimates are generally (and where possible) compared to bottom-up inventory reported emission data
Under time-invariant and spatially uniform conditions, a mass-balance approach based on Gauss's divergence theorem equates the rate of emissions from sources within a control volume (box flight) to the net mass flux through the boundaries (box walls) by assuming the net rate of accumulation of mass within the volume to be negligible. When conditions (e.g., meteorology, emissions) deviate from steady-state and/or localized inhomogeneity is developed in tracer concentration and meteorological fields, the rate of mass accumulation/release within the control volume may become significant enough to affect the accuracy of the estimates based on the mass balance approach. Herein, we refer to these circumstances as “storage-and-release” events. To our knowledge, all mass-balance techniques to date assume steady-state conditions. Here we investigate the uncertainty associated with that assumption and we introduce the term storage-release (used interchangeably with storage-and-release throughout the paper) events for transient non-steady-state conditions. This term is chosen to emphasize the cyclical nature of the events, during which material within the control volume accumulated and is then subsequently released.
Numerical dispersion modelling has been proven useful in assessing the uncertainties in top-down retrievals in the past
During an aircraft campaign conducted in August and September of 2013 as part of the Joint Canada Alberta implementation plan for Oil Sands Monitoring (JOSM), the National Research Council of Canada Flight Research Laboratory (NRC-FRL) Convair 580 research aircraft was used to make air-quality measurements over the Athabasca oil sands region (Alberta, Canada). Air-quality forecasts created using ECCC's air-quality model GEM-MACH, provided advice on flight planning for the JOSM 2013 campaign
ECCC's air-quality model, GEM-MACH, is a fully coupled online air-quality and chemical transport model. The GEM-MACH modelling system resides within the Global Environmental Multiscale (GEM) numerical weather prediction model
For this work, a series of retrospective high-resolution nested air-quality simulations were made for the period of the JOSM summer 2013 aircraft intensive campaign
Flux box data, including meteorological fields and Air-quality models usually require the use of some form of “mass conservation” correction for their advection algorithms Regional air-quality models provide instantaneous output at every grid cell on a fixed (2 min) time step – however, if these are interpolated to a finer time resolution (e.g., along an aircraft flight track), errors in interpolation may affect the use of the model output in retrieval algorithms. Values of model variables at 3D grid points, in the vertical columns making up the emissions “box”, were therefore used in the analysis which follows to avoid these (temporal and spatial) interpolation errors in the use of model output as a proxy for observations. In aircraft-based sampling (both in real-world aircraft measurements and in extraction of emissions retrieval algorithm inputs from 4D air-quality model-predicted output), temporal lag between data points collected along flight tracks may represent an additional source of error. This may introduce additional errors if the “static atmosphere” assumption is not met. Here, instantaneous model output at each time step (2 min) was used for mass-balance estimates, to avoid the potential for interpolation from the model resolution (2.5 km) and time step to influence the retrievals generated here.
For this purpose, GEM-MACH model four-dimensional data (3D space, 1D time) were extracted at every 2 min model chemistry time step for the flux boxes over individual oil sands facilities for our nine case studies. Aircraft-based mass-balance retrievals were simulated by employing the divergence theorem in analyzing the model extracted data. The estimated emission rate time series and flight-time (1.5–2.5 h) averages were then compared to model input emissions (MIEs) to evaluate top-down mass balance retrievals.
During an actual aircraft campaign, measurements are made along a flight trajectory on the lateral walls of a flux box (control volume); measured data on a single 2D screen (closed surface) comprised of box lateral walls are all that is available for emission rate retrievals based on mass flux and mass balance calculations. Here we use instantaneous model values along lateral box walls as our 2D screen (“TR” approach, Sect.
The temporal evolution of the system can be studied by increasing the number of flights around the same facility. Here, this approach is explored with GEM-MACH simulated fields by analyzing multiple 2D screens at consecutive model time steps (3D fields) and generating mass balance estimates (“TR
A truly comprehensive analysis of the tracer mass budget within the flux box is only possible with 3D-volumetric data over the entire extent of the box and over time (4D fields). However, collection of 4D (volumetric time series) data is not practically feasible during aircraft measurements as the aircraft cannot sample the entire volume of the box even with increased flights during a time window of only a few (e.g., 2–3) hours. One advantage of the use of a model such as GEM-MACH as a proxy for observations is that 4D forecast fields can be generated, to provide complementary information for mass balance retrieval investigation. Here we use GEM-MACH model 4D chemical and meteorological fields to conduct a comprehensive analysis of the transport of the emitted mass from sources within the control volume and to further investigate uncertainties associated with storage-and-release events in top-down mass balance retrievals (“DR” approach, Sect.
Gauss's divergence theorem states that the total integrated divergence of a vector field (e.g., fluid flow)
We start with a comprehensive analysis of GEM-MACH 4D data for direct retrieval of model input emissions, where we explore all the processes (including storage-and-release) contributing to change in tracer mass within the flux box and over time (Sect.
Here we apply the divergence theorem to the GEM-MACH model four-dimensional (volumetric time series) output data to estimate model input emissions. Data from volumes corresponding to the nine JOSM 2013 cases were extracted from the GEM-MACH model output at every 2 min model time step (no interpolation), creating nine sets of 4D metadata with variables (including wind speed and direction, air density, temperature, species mixing ratio, ground surface deposition). Model-extracted 4D flux box metadata, spanning model vertical levels from the surface up to
The total mass of compound
Aircraft-based retrieval methods such as TERRA rely on measured data along the lateral walls of the box. In observation-based retrievals, the horizontal flux
For this portion of our analysis, in which we limit the inputs to mass-balance estimations to the information available on the box lateral walls in analogy to aircraft observations, instantaneous 2D screens of relevant fields (e.g.,
Furthermore, due to lack of measurements at the box top, the vertical flux
Estimations of emission rates were made for each model time step (using instantaneous 2D screens) during the time periods of the nine studied JOSM cases by using GEM-MACH output to approximate the input data for TRs as described above. Note that the TR method, as described in this section, makes use of GEM-MACH 2D data at each time
Utilizing time-consecutive instantaneous screens around an emission source, the temporal evolution of the system and its impact on mass-balance estimates can be studied in more detail, as discussed at the beginning of this section (Sect.
Considering the potentially prohibitive operational cost of conducting repeat flights and the fact that few (e.g., two to three) time-consecutive measurements may not provide enough information for the estimate of temporal trends in within-box tracer mass budget, the following alternative is proposed. As we discuss later in Sect.
We note that the correlation between
Using the consecutive screens at every model time step (extracted from GEM-MACH model output fields), estimations of
Table
Summary of the three retrieval methods: direct retrieval (DR), TERRA retrieval (TR) approximation and revised TERRA retrieval (TR
Figure
Emission rates determined by the three methods (DR, TR and TR
Contribution (%) of different terms in Eq. (
DR estimations were made by analyzing model 4D data for flux boxes approximating the nine cases. The numerical integration expressions for calculating each term, using model data, are described in Table
DR method estimates were compared to the MIEs, as shown in Figs.
Time series of the estimates by DR, TR and TR
For the majority of our studied cases, our results indicate that the storage term (if not accounted for) contributes the bulk of emission rate over-/under-estimates for the variables investigated here, and hence methods to predict and/or reduce its influence are desirable for aircraft retrieval algorithms. We note that the impact of storage on emissions retrieval estimates will depend on the relative magnitude of the estimated emissions themselves in comparison to other sources of uncertainty. That is, an emissions estimate double that of previous estimates, which has a
Storage-and-release events occur when mass is accumulated within the volume of the box and released at a later time through box lateral walls; see the detailed discussions in Sect.
By limiting the analysis to the extracted model data along box lateral walls (Fig.
Comparing the performance of the two methods (DR and TR) against MIEs reveals the significance of the storage term (
By combining the information from Figs.
This three-panel graph demonstrates the correlation between estimates of
This schematic demonstrates (qualitatively) the correlation between change in atmospheric stability represented by the gradient Richardson number (
Here we show that change in tracer mass within the flux box can also be observed on box lateral walls over time (using time-consecutive screens), as described in Sect.
The main processes contributing to the change in
Aircraft-based emissions retrieval methods such as TERRA, utilize the mass-balance approach with the underlying assumption of mean steady-state conditions over the region of study (
Atmospheric dynamic stability can be described in terms of the gradient Richardson number (
Correlation between the NRMS error and the change in atmospheric stability represented by the absolute change in the gradient Richardson number
For our analysis, 4D fields of
The average value
The direction of the wind experienced by the plume may change when the plume remains at one altitude, or may change as the plume rises or falls in the atmosphere (e.g., via the well-known “Ekman spiral” of wind direction changes with increasing height,
Cases 6 and 9 (both release events) along with rejected case 5 (storage event) were the three cases most impacted by variations in wind direction at plume-center height. Flight time average
Correlation between the NRMS error and the change in wind direction at plume height (
In applying the mass-balance technique, the mass flux associated with upwind emissions (and regional background levels) entering the box volume (inflow) is subtracted from the downwind outgoing flux containing the emissions within the box (outflow) to estimate the net emission rate from sources within the box volume. The mass inflow associated with regional background levels (e.g., for
For an isolated source with relatively low upwind emissions (which in turn depends on the mean wind direction and the location of the nearby sources)
Case 8 screens around Suncor:
Small
Correlations between the NRMS error and the forecast parameter
Our results indicate a strong positive correlation between
The three forecast parameters
Summary of correlations between NRMS error in TR estimates and the three forecast parameters for storage-release events, in terms of the Pearson correlation coefficient (
We have examined storage and release events in the specific context of “box” flights around an emitting facility, however note that other strategies have been put forward in the literature
Table
A summary of the meteorological conditions and source emission scenarios by the three oil sands facilities for the period of the nine JOSM 2013 box flight cases in our GEM-MACH model simulations. Flight date-time and duration is provided for each case, LT stands for local time (UT
We have carried out a series of retrospective air-quality simulations employing Environment and Climate Change Canada's (ECCC) air-quality model, Global Environmental Multiscale – Modelling Air-quality and Chemistry (GEM-MACH), for the period of the JOSM 2013 campaign over the Athabasca oil sands region, with the primary objective of evaluating aircraft-based mass balance emission rate retrievals such as applied in TERRA (i.e., the Top-down Emission Rate Retrieval Algorithm). We considered the simulations of nine JOSM 2013 emission estimation flight cases, over three oil sands facilities with high
Results were compared to the MIEs and the performance of the three methods was analyzed in terms of normalized (to MIE) root-mean-square (NRMS) error. The DR method estimates (
Storage-release events occur when temporal and spatial variations in meteorological conditions and/or source emission rate result in a temporary imbalance between the addition of mass to the volume of the box through source emissions and removal of mass by vertical and horizontal fluxes through box top and lateral walls and the deposition to ground surface. The transient storage of the emitted mass within box volume and its later release contribute to emission rate over- and under-estimations based on the mass-balance technique. We introduced a correction term,
We have introduced an approach for estimating the storage term as the TR
The investigation carried out here has resulted in the creation of quantitative measures for the extent to which storage and release events may impact aircraft emissions retrieval accuracy. Retrieval uncertainties, using regional air-quality model output, were shown to be of similar magnitude to previously published values, with the exception of cases where the underlying assumption of time-invariant meteorological conditions was not valid and/or where significant upwind emissions impacted the estimates. We have also devised a methodology to reduce the impact of this form of emissions retrieval error by estimating the storage rate (
GEM-MACH configuration details.
The net horizontal flux exiting through lateral walls of the box (
The normal wind (positive outwards) along the screen is calculated as
The vertical flux through the box top can be calculated as
The horizontal air flux through the screen (positive outwards) is calculated from extracted air density and wind fields along the screen at time
The rate of change in the air mass (accumulation) within the box due to changing air density (compressible fluid) can be estimated as
The vertical air flux leaving the box (positive upwards) is calculated similar to Eq. (
Discrete integral expressions of Eqs. (
Discrete integral expressions for Eqs. (
Discrete integral expressions for Eqs. (
The numerical integration expressions for calculating the terms in Eqs. (
The contribution of the storage rate term (
Note that this work made use of an air-quality model: no observational datasets were used in this work. The model results are available upon request to Paul Makar (paul.makar@ec.gc.ca). GEM-MACH, the atmospheric chemistry library for the GEM numerical atmospheric model (©2007–2013, Air Quality Research Division and National Prediction Operations Division, Environment and Climate Change Canada), is a free software which can be redistributed and/or modified under the terms of the GNU Lesser General Public License as published by the Free Software Foundation – either version 2.1 of the license or any later version. The specific GEM-MACH version used in this work may be obtained on request to paul.makar@ec.gc.ca. Much of the emissions data used in our model are available online: Executive Summary, Joint Oil Sands Monitoring Program Emissions Inventory report (
SF wrote the paper with supervision and input from MG and PAM. SF ran the GEM-MACH simulations and implemented the mass-balance retrieval algorithm into GEM-MACH. PAM coordinated the GEM-MACH work. AA setup the GEM-MACH code and provided simulation support. AD, JL, KH, and SML provided advice during analysis and contributed to paper revisions.
Some of the authors are members of the editorial board of
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was funded under the Oil Sands Monitoring (OSM) Program. It is independent of any position of the OSM Program.
This research has been supported by the Oil Sands Monitoring Program (grant no. A-PD-6; Integrated Atmospheric Deposition Monitoring), and the Natural Sciences and Engineering Research Council of Canada – NSERC (grant no. RGPIN 2015-04292).
This paper was edited by Joshua Fu and reviewed by Wayne Angevine and two anonymous referees.