Desktop version

Home arrow Engineering

  • Increase font
  • Decrease font


<<   CONTENTS   >>

High Spatial Density Greenhouse Gas Observing Networks—The INFLUX Example

The Indianapolis Flux Experiment (INFLUX) began in 2010 to develop and demonstrate urban-scale or whole-city GHG flux measurement capabilities, and particularly to compare results from top-down and bottom-up methods. Indianapolis was chosen in part because of its moderate size (~ 1.8 million inhabitants in the city and surrounding area), relative isolation horn other strong emissions sources, and location in a landscape of flat terrain where NWP models are anticipated to have optimal performance. Critically. Indianapolis was the development site for an urban-scale, high spatiotemporal resolution, bottom-up emissions model or data product, known as Hestia (Gurney, 2017).

The initial INFLUX project concept to compare three independent and complementary whole-city emissions quantification methods: Two based on top-down atmospheric observing methods and the bottom-up Hestia emissions data product. Two top-down atmospheric flux measurement approaches were employed: The aircraft mass balance measurement and a two-station, surface-based, mole fraction observing network for the gases CO,, CH4, CO, and water vapor. The initial surface observing network had two tower-based nodes located to the southwest and the northeast of the city, respectively, along the prevailing wind direction.

The objectives of INFLUX were broadened shortly after the project began to investigate the capability of a spatially dense, surface observing network coupled with Bayesian inference analysis to identify CO, source locations and estimate flux magnitude at those locations (Davis, 2017). The second INFLUX phase aims to develop, assess, and minimize uncertainties of methods for quantifying greenhouse gas emissions at the urban scale, using the Indianapolis urban area (Figure 14) as a testbed. Target performance criteria for the expanded INFLUX goals were demonstration of emission source location identification within 1 to 2 km1, relative flux magnitude uncertainty of ~ 10%, all at a weekly or sub-weekly temporal scale. Similar expectations were put in place for whole-city flux determinations. The observing network density within the city was increased from 2 to 12 nodes. High-accuracy measurement of СО,, CH,. and CO mole fraction from continuous atmospheric sampling made at heights from 39 m to 136 m above ground level use existing communication towers fitted with a multi-line sampling manifold to sample the atmosphere at several elevations. Cavity ring-down spectrometers (CRDS) are used at all twelve sites. Each network node has at least one reference gas mixture contained in high pressure aluminum cylinders that is traceable to WMO standards (Zhao, 2006) and to the SI (Rhoderick, 2016). These measures are taken to ensure data quality and consistency within the observing network, based on a reference framework closely tied to global mole fraction references disseminated by the WMO Central Calibration Laboratory (Richardson, 2017). To assess CRDS instrument performance periodically and conserve valuable gas mixtures, gas mixtures in cylinders of known (calibration cylinders) or of unknown (target cylinders) mole fraction are used by a network nodes' control hardware to periodically present to the CRDS instrument a gas of stable mole fraction. These procedures are used to quantify instrument drift and detect failure conditions. Network-wide round robin tests performed every 1 to 2 years detected possible drift in mole fraction of both target and reference gases.

Flask sampling systems deployed at six network nodes drew samples fr om the same manifold as the node’s CRDS analyzer. Flasks, containers made of specialty glass so as to not contaminate gas samples and having a volume of approximately one liter, were filled approximately monthly and sent to NOAA- GMD for analysis. Laboratory gas analysis for approximately 40 trace gas species was made by NOAA- GMD on flask samples (Sweeney, 2015; Turnbull, 2012). Analyses also included the greenhouse gases and particularly the carbon 14 radio-isotope in the form of 14CO,. Radio-carbon data assist in estimating usage of fossil fuels in the city. Data from flask analyses provides a means to directly compare results obtained from the node’s CRDS spectrometer. These comparisons are a means of periodic verification of analyzer performance for CO„ CH4, and CO mole fraction. Comparison levels were less than or equal to 0.18 prnol/rnol CO,, 1.0 nmol/mol and 6 urnol/rnol for CH4 and CO, respectively (Turnbull. 2015).

To better characterize atmospheric dynamics and energy transfer between the surface and the atmosphere, additional measurement capabilities were added. Doppler LIDAR measurements located to the Northwest of the city center have provided three-dimensional wind vector and turbulence data and PBL depth information continuously for the last several years. Four network towers were fitted with eddy covariance measurements in order to investigate surface-atmosphere energy exchange behavior across the typology of the city. These towers sampled energy flux properties and behaviors of land surface for urban typologies ranging from rural to the heavily urban center of Indianapolis (Sarmiento, 2017). This rather diverse set of measurements is indicative of the many processes operating in urban environments

Vertical and horizontal aircraft transects; Baltunore-Washington D.C. regions and coded by observed CO, mole fraction. Image and data courtesy of O. Salmon and P. Shepson

Figure 13. Vertical and horizontal aircraft transects; Baltunore-Washington D.C. regions and coded by observed CO, mole fraction. Image and data courtesy of O. Salmon and P. Shepson.

Locations and designations of the INFLUX observing network

Figure 14. Locations and designations of the INFLUX observing network. Each communications tower-based, network observing node measures CO,, CO, and/or CH4 continuously, ~ 5 second cycle time, depending on the type of CRDS analyzer installed. Six nodes have flask sampling capability and several have eddy flux measurement capability. LIDAR, both Doppler and particulate, systems monitor PBL dynamics to the Northwest of the city, http://sites.psu.edu/mflux/site-information/.

found in U.S. cities of moderate size. This information allowed assessment of various processes impacting atmospheric transport within and around the city as a means of better understanding how those processes interact and impact GHG emissions estimation.

These various information sources provided an opportunity for extensive observation-to-model comparison, e.g., the WRF model, to assess simulation errors in meteorological variables such as latent heat and sensible heat11 fluxes, air temperature near the surface and in the PBL, wind speed, direction, turbulence, and PBL height (Deng, 2017). Although results such as these have yet to be used to modify NWP models and additional information is needed, they represent progress toward a better understanding of PBL dynamics to improve NWP capabilities, which in turn is anticipated to have a positive effect

Sensible heat m a thennodynamic system is an exchange that results in a temperature change m the system.

on reduction of uncertainties in top-down flux estimation. Extensive descriptions of the methods used and some results derived from INFLUX are published in a journal special issue (Elementa Special Feature, 2016).

Frequently updated information on emission rates and their change will considerably enhance evaluation of the effectiveness of carbon management approaches in cities. Recent results from four of the independent CO, emissions methods of the INFLUX project were analyzed in order to address two key questions: (1) what the magnitude of the whole-city emission is and (2) what the uncertainty of these quantification results is. As one of the few urban areas where multiple emission assessment methods have been implemented, Indianapolis pror ides a unique opportunity to compare different methods directly. To address these questions, a comparison of the whole-city CO, emission rate was made, based upon different approaches: A science-driven high-resolution urban inventory-based emission data product (Gumey, 2011), an atmospheric transport model inversion based on in situ tower observations, and two different mass balance flux estimates from aircraft observations. To evaluate differences between the methods, discrete flask-based measurements of 14CO, were used to determine fossil fuel CO, separately from biogenic CO, flux contributions. To minimize complications associated with biospheric CO, fluxes in this first attempt to compare differences between methods, only winter-time emissions were considered (Turnbull, 2019).

CO, emission rate values reported initially for the methods used in this comparison ranged from 14,600 to 22,400 mol/s with a 1 sigma variation of 21% and a maximum difference 42%. Although an improvement over previous uncertainty estimates for urban emissions of 50-100% from other studies (Gately, 2017; Pecala, 2010), these initial INFLUX flux estimate uncertainties are insufficient for detection of emissions trends on the order of 10% per decade. Re-analysis of the initial estimate involved the following adjustments:

1. Adjustment for fossil fuel CO, emissions alone relative to total CO, emissions.

The Hestia data product compiles data for anthropogenic CO, emissions derived from combustion of fossil fuels that include bioethanol. The atmospheric inversion and aircraft mass balance methods both estimate the net total urban enhancement in CO„ which includes the influence of both anthropogenic and biogenic CO, fluxes.

2. Adjustment of geographic region sampled by the various methods.

The geographic area for which emissions are sampled differ among the different methods. For example, the Hestia data product incorporates information from the 8 surrounding counties and Marion County that contains the geopolitical boundaries of Indianapolis. The aircraft mass balance method samples fluxes for more confined areas that differ from one flight to another.

3. Adjustment for the time of day and time period of an estimated flux.

Hestia is the mean flux over all hours of the day. The inversion analysis only uses observational data from the 11 am to 4 pm local time period. The nine aircraft mass balance flights used in the analysis were only conducted on weekdays in November and December 2014, in the daytime. Adjustments to Hestia and the inversion method were made to coincide as much as practicable with the mass balance observation times.

4. Accounting for CO, background or atmospheric concentration choice in the incoming atmosphere. In principle, the background CO, signal would be the mole fraction that would have been observed by at tower or aircraft location with no urban emissions present. Each method treats estimation of this background concentration differently. Consistent methods are needed that properly account for differences in choice of backgr ound for different methods.

Results were compared in winter, only to avoid the substantial biogenic CO, flux contributions both coming into the Indianapolis domain from the large agricultural regions outside it duiing the growing season and those occurring within the city from urban vegetation. Hestia did not contain a biogenic component, which is typical of emissions models in their current state, although efforts are underway to include urban domain vegetation estimates (Sargent, 2018). As shown in Figure 15, by using this

Whole-city fossil fuel CO, emissions rate estimates for Indianapolis in winter agree within 7%

Figure 15. Whole-city fossil fuel CO, emissions rate estimates for Indianapolis in winter agree within 7%. This is achieved by accounting for differences in spatial and temporal coverage, as well as the trace gas species measured. Image courtesy of

J. Turnbull.

reconciliation procedure to account for differences in spatial and temporal coverage, agreement among the wintertime whole-city fossil fuel CO, emission estimates falls within 7% (Turnbull, 2019). Error bars on each method indicate uncertainty estimates for each.

Relative to previous urban-scale comparisons, this result is a significant advance. This study represents the first comprehensive, multiple-method assessment of urban fossil fuel CO, emissions and demonstrates that agreement across these can be useful to those tasked with carbon management of whole city emissions.

The complementary application of multiple scientifically driven, evidence-based emissions quantification methods enables and improves confidence levels and demonstrates the strength of the joint implementation of rigorous inventory and atmospheric emissions monitoring approaches. However, this advance has been made with significant restrictions from the point of view of general application, e.g., exclusion of biogenic emissions, data from the midday only, and through the use of highly accurate measurements, costly equipment and research-grade analysis methods. INFLUX represents the application of considerable research resources, not likely to be available in most urban regions on a routine basis. Continued advances in measurement capabilities are needed in order to move research and implementation of results toward broad application. For example, replication of convergence among methods as demonstrated by the initial success of the INFLUX research should be demonstrated over all seasons, times of day, in different urban regions, and at considerably reduced resource expenditure to make them practically applicable. Doing so requires continued efforts by both the scientific and the stakeholder communities and requires considerably expanded engagement and effort.

 
<<   CONTENTS   >>

Related topics