1. Introduction
Greenhouse gas (GHG) emissions from urban areas represent about 70% of the total anthropogenic emissions from fossil fuel consumption around the globe (IEA 2008). Current methodologies used to quantify these emissions rely primarily on energy use consumption and many indirect tracers of the socio-economic activity at national, regional, and city levels (Olivier et al. 1999; Gurney et al. 2009, 2012). The accuracy of these estimates highly depends on the level of available information (Marland et al. 1999). Because of the amount of information that has to be collected, the availability of these estimates is often delayed by several years and the absolute estimates rarely include an assessment of their related uncertainties. In this context, verification methods are needed to evaluate these estimates and reach sufficient levels of accuracy for future policies and regulations (Nisbet and Weiss 2010).
One of the proposed methodologies to evaluate and complement the bottom-up estimates uses atmospheric observations in an inverse framework to provide independent verification and to potentially increase the level of information in time and space (Enting 1998). The atmospheric inversion method has been used at global and continental scales, with a certain degree of success (e.g., Bousquet et al. 2000), to estimate the exchanges of GHGs between the atmosphere and land and oceans. Recently, the method has been successfully used to estimate regional fluxes at finer scales (Peters et al. 2007; Lauvaux et al. 2009; Schuh et al. 2010) and was compared to existing agricultural inventories (Schuh et al. 2013). This study demonstrated the utility of the method at regional scales, even though inversions over limited domains suffer from the lack of information on the boundary inflow (Göckede et al. 2010).
Recently, several attempts were made to evaluate urban-scale emissions of greenhouse gases using atmospheric observations with various inverse techniques. On the basis of tower measurements, Lowry et al. (2001) used a mass-balance approach to quantify London, England, CH4 emissions and their trend over several years. Turnbull et al. (2011) estimated the emissions from Sacramento, California, using 14CO2 aircraft flask samples and CO measurements collected over several days, whereas Mays et al. (2009) used CO2 mixing ratios from aircraft flights over Indianapolis, Indiana. Using a multiple box model, Strong et al. (2011) estimated the emissions from Salt Lake City, Utah, and estimated the contributions from biogenic and anthropogenic sources. These studies provided urban-scale estimates aggregated over various sectors of activity, but the complexity of the local atmospheric dynamics (which influences the modeling performance and the interpretation of the atmospheric measurements) combined with the sparsity of the observations limits the potential of these methods.
Atmospheric modeling, a critical piece of the inverse system, is often challenging over urban areas because of the heterogeneity of the surface properties (Masson et al. 2002) and the simulation of finescale structures such as plumes from point sources (Hanna et al. 2007). The accuracy of the atmospheric transport models can lead to large uncertainties unless the full complexity of the local dynamics is correctly simulated (Lac et al. 2012). In addition, the distribution of emitting sources in space is well defined, which requires better simulation of advection and vertical mixing in comparison with regional-scale inversions. Finally, inverse estimates are also limited by the lack of information on the emission uncertainties and their structures and the colocalization problem in urban inversion of emissions when large sources are located in the vicinity of the measurement site (Bocquet 2005). Inverse solutions attribute the signals to being near the measurement locations where the sensitivity to the emissions is at a maximum and can lead to unrealistic solutions without the use of correlations in the prior errors.
In this study, we present the first attempt to characterize daily CO2 emissions from a city, using a state-of-the-art mesoscale model, the Weather Research and Forecasting model (WRF) in four-dimensional data assimilation (FDDA) mode, nudged to meteorological observations over the region. Relative to previous studies, the modeling system is able to reproduce more accurately the dynamics at very fine scales (Deng et al. 2009; Rogers et al. 2013). The optimization system used here is an adjoint-free inverse approach over a small-sized city, Davos (Switzerland). Starting from an annual inventory (Walz et al. 2008), the system produces daily inverse estimates of the city CO2 emissions. Two observation sites have been deployed from 23 December 2011 to 3 March 2012 measuring the boundary inflow and the urban plume in the valley. During the campaign, we investigated specifically the impact of the World Economic Forum 2012 (WEF-2012) meeting and a severe cold wave affecting a large part of Europe, potentially impacting the city emissions. In addition, CO/CO2 hourly ratios were used to detect changes in activity sectors, and meteorological measurements (e.g., friction velocity and temperature) were used to evaluate the modeled vertical mixing in the valley. We finally discuss the observed daily changes in city emissions and their potential causes.
2. Methods
a. The Davos measurement campaign: Network design and instrumentation
Davos is located in the Swiss Alps at an altitude of 1560 m, with a population of about 11 000 permanent residents. Despite the large surface of the county (about 250 km2), the city itself covers only 6 km2 at the bottom of a broad valley. The economic activity in the area is dominated by tourism because of a large ski resort and mountain agriculture. Every year, the city of Davos hosts the WEF meeting, which brings to the town a few thousand participants, along with several hundred security personnel. The WEF-2012 meeting took place from 25 to 29 January, with 2600 participants. Two atmospheric GHG mixing ratio measurement sites were deployed in mid-December 2011 approximately 3.7 km apart [using Picarro, Inc., model G2401 sensors (Rella et al. 2012)], one on the valley floor in the urban area and one to the west in the surrounding mountains (Fig. 1). The mountainside site provides background inflow atmospheric mole fractions while the downtown site measures the urban environment. Each site measures CO2, CH4, CO, and H2O, the latter being used to correct for the dilution effect due to water vapor interfering with other greenhouse gas spectroscopic measurements (Chen et al. 2010; Rella et al. 2012). The mountainside site was installed with assistance from the Institute of Snow and Avalanche Research (SLF) at the Weissfluhjoch test site at an altitude of 2540 m, just 150 m below the Weissfluhjoch mountain peak. The inlet height was approximately 5 m AGL at both sites, and measurements of all species were made every 10 s, averaged hourly. Both instruments were calibrated using a single National Oceanic and Atmospheric Administration/ Earth System Research Laboratory (NOAA/ESRL) traceable tank (CO2 = 394.36 ppm; CH4 = 1846.3 ppb) at the beginning of the data collection period (mid-January 2012) and at the conclusion of data collection (3 March 2012). At both sites, the pre- and postcalibrations differed by less than 0.1 ppm CO2 and less than 1ppb CH4; thus, a single offset correction was applied to all measurements. The calibration tank was not characterized for CO and, thus, no correction was applied, although the pre- and postmeasurements differed by less than 2 ppb CO. In addition to mole fraction measurements at the downtown site, a three-dimensional sonic anemometer (Campbell Scientific, Inc., CSAT3) was installed at 5 m AGL along with an inlet to a CO2 flux analyzer (Picarro, Inc., model G2301). The flux analyzer collected 10-Hz measurements of CO2; the x, y, and z components of the wind speed; and sonic virtual temperature. These data were used to measure the friction velocity and buoyancy flux and to evaluate the modeled vertical mixing during the campaign (see section 4a).
WRF-FDDA simulation domain topography in the WRF-FDDA modeling system at the resolution of 1.3 km. The mountain site (red dot) is used to measure the background mixing ratios for CO2 and CH4 and the downtown site (blue dot) measures the GHG mixing ratios in the valley. The valley is about 4 km wide with an elevation difference of about 1 km.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
WRF-FDDA simulation domain topography in the WRF-FDDA modeling system at the resolution of 1.3 km. The mountain site (red dot) is used to measure the background mixing ratios for CO2 and CH4 and the downtown site (blue dot) measures the GHG mixing ratios in the valley. The valley is about 4 km wide with an elevation difference of about 1 km.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
WRF-FDDA simulation domain topography in the WRF-FDDA modeling system at the resolution of 1.3 km. The mountain site (red dot) is used to measure the background mixing ratios for CO2 and CH4 and the downtown site (blue dot) measures the GHG mixing ratios in the valley. The valley is about 4 km wide with an elevation difference of about 1 km.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
b. Inverse modeling system
1) Atmospheric FDDA modeling system
The core of our real-time modeling system that Deng et al. (2012a) used in this research is the WRF model coupled with chemistry (WRF-Chem; Grell et al. 2005) modified for passive tracers as in Lauvaux et al. (2012). The WRF configuration for the model physics used here was based on previous numerical modeling studies (e.g., Gaudet et al. 2009; Rogers et al. 2013; Deng et al. 2012b) using 1) the single-moment 3-class simple ice scheme for microphysical processes, 2) the Kain–Fritsch scheme for cumulus parameterization on the 36- and 12-km grids, 3) the Rapid Radiative Transfer Model for longwave atmospheric radiation and the Dudhia scheme for shortwave atmospheric radiation, 4) the turbulent kinetic energy (TKE)–predicting Mellor–Yamada–Jancić (MYJ) level-2.5 turbulent closure scheme for the turbulence parameterization in the planetary boundary layer (PBL), and 5) the five-layer thermal diffusion scheme for representation of the interaction between the land surface and the atmospheric surface layer (Skamarock et al. 2008).
The WRF modeling system used in this study has FDDA capabilities to allow the meteorological observations to be continuously assimilated into the model. The FDDA technique used in this study was originally developed for the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5; Stauffer and Seaman 1994) and was recently implemented into WRF (Deng et al. 2009). The present system was used in several studies (e.g., Deng et al. 2004; Deng and Stauffer 2006; Reen et al. 2006). Nudging of the wind field is applied through all model layers, but nudging of the mass fields (temperature and moisture) is only allowed above the model-simulated PBL so that the PBL structure produced by the model is dominated by the model physics. In this specific real-time application, the World Meteorological Organization (WMO) observations were assimilated into the WRF-Chem system to produce a dynamic analysis, blending the model simulations and the observations to produce the most accurate meteorological conditions possible to simulate the atmospheric CO2 concentrations in space and time throughout the Davos region.
The WRF model grid configuration used for this demonstration is composed of four grids: 36, 12, 4, and 1.33 km (see Fig. 2 for the 4- and 1.33-km grids), all of which are cocentered at Davos. The 36-km grid, with a mesh of 110 × 110 grid points, contains the entire continental Europe and parts of the Atlantic Ocean. The 12-km grid, with a mesh of 151 × 151 grid points, contains Switzerland, France, Italy, Poland, and Germany. The 4-km grid, with a mesh of 175 × 175 grid points, contains western France, southern Germany, northern Italy, western Austria, and all of Switzerland. The 1.33-km grid, with a mesh of 202 × 202, covers portions of northern Italy, southern Germany, western Austria, and western Switzerland, with the grid centered at Davos. Fifty vertical terrain-following layers are used, with the center point of the lowest model layer located ~12 m AGL. The thickness of the layers increases gradually with height, with 27 layers below 850 hPa (~1550 m AGL).
Surface meteorological observation distribution on the 4- and 1.33-km grids, valid at 0000 UTC 29 Jan 2012, including surface meteorological stations (open circles) and rawinsondes (filled circles) from the WMO database used in the WRF-FDDA modeling system. Davos is located at the center of the domains (gray star).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Surface meteorological observation distribution on the 4- and 1.33-km grids, valid at 0000 UTC 29 Jan 2012, including surface meteorological stations (open circles) and rawinsondes (filled circles) from the WMO database used in the WRF-FDDA modeling system. Davos is located at the center of the domains (gray star).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Surface meteorological observation distribution on the 4- and 1.33-km grids, valid at 0000 UTC 29 Jan 2012, including surface meteorological stations (open circles) and rawinsondes (filled circles) from the WMO database used in the WRF-FDDA modeling system. Davos is located at the center of the domains (gray star).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
The FDDA parameters used in this application can be found in Deng et al. (2012a). For this application, 3D analysis nudging and surface analysis nudging were applied on both the 36- and 12-km grids, with reduced nudging strength on the 12-km grid, and observation nudging was applied on all grids with the same nudging strength. No mass fields (temperature and moisture) observations are assimilated within the WRF-predicted PBL. The meteorological observations assimilated into the WRF system are based on the WMO observations distributed by the National Weather Service and include both 12-hourly upper-air rawinsondes and hourly surface observations. Figure 2 shows the WMO surface observation distributions at 0000 UTC 29 January 2012 on the 4- and 1.33-km grids, indicating a significant amount of observations over the region. The gridded meteorological data used to initialize the WRF-Chem real-time system were the National Centers for Environmental Prediction Global Forecast System analyses (i.e., 0-h forecast) available every 6 h in real time.
2) A priori CO2 emissions
The inventory, used as an initial estimate of CO2 emissions in Davos, consists of one single number for the year 2005 from Walz et al. (2008). From this study, the direct emissions (as opposed to indirect emissions from manufacturing or activities emitting carbon out of the valley) represent 91% of the 106 kilotons of total CO2 (ktCO2) emitted per year, which correspond to 8–9 ktCO2 per capita annually. The house heating emissions alone represent 76% of the direct CO2 emissions, followed by traffic (17%) and other sources such as machines and electricity production for about 7%. House heating emissions originate from the combustion of fuel oil at 98.5%, whereas electricity, thermal heating, wood, geothermal, and natural gas represent only 1.5%. Similarly, human respiration from the city represents about 3% of the total direct emissions based on the population of about 12 000 people and using human respiration factor from West et al. (2009). Although the inventory provides annual emission estimates for each energy source, no spatial or seasonal references are available. To correct for the seasonal variability, the first month of the experiment was used as a correction period, with a large increase due to house heating in wintertime. The correction may also include potential changes in the emissions between 2005 and 2012. Results are presented in section 3b. For the spatial distribution, the inventories account for emissions at the county scale of Davos, with a total surface of about 254 km2. Only 6 km2 (or about 2%) of the total area corresponds to a low-density urban area. The land cover map from the U.S. Geological Survey (land cover map classification used in WRF) was used to distribute the county-level emissions in space, using pixels dominated by the low-density residential land cover type. At 1.33-km resolution, three pixels represented the source area in our prior flux estimate.
3) Inversion technique: direct interpolation (adjoint-free)
This method also assumes a perfect transport model and excludes the use of uncertainties for prior fluxes or transport. Whereas potential model errors may affect daily inverse estimates because of the absence of explicitly described model errors in the inversion, an optimal period of observation was determined on the basis of well-mixed criteria. We evaluated the atmospheric model simulations using meteorological measurements and selected the optimal time period to ensure the well-mixed conditions in the PBL. The results are presented in section 4a.
Similarly to the PBL budget approach (Turnbull et al. 2011), here the model–data residuals are used directly to estimate the daily emission corrections (multiplicative factors), which are applied to the initial emission estimate. The mismatch between the modeled mixing ratios and the observed mixing ratios was directly applied to the daily emissions. The use of the FDDA system allows us to simulate the full complexity of the atmospheric circulation and dynamics at the local scale, which is more robust than simplified PBL budget approaches using idealized box models or the differential of mixing and concentrations. The three scenarios were performed daily, all on the basis of the initial inventory estimate, to evaluate the transport linearity assumption. The first scenario used the initial estimate from Walz et al. (2008), a second scenario applied a +20% increase relative to the initial estimate, and a third scenario applied a +40% increase. The correlation between the increase in the emissions and in the modeled mole fractions (i.e., the coefficient α) is within 2% of the linear solution, confirming the robustness of the linear assumption. Therefore, a linear regression technique was used to interpolate the mixing ratio mismatch to the emissions, using observations during well-mixed conditions only.
3. Results
a. Observed CO2 and CH4 mole fraction time series
The observed CO2 atmospheric mixing ratios are presented in Fig. 3 at an hourly time scale. The downtown site observations show a large diurnal cycle, with high peaks during nighttime and lower values during daytime. The large amplitude of the diurnal cycle (up to 200 ppm) is due to the changes in stability conditions in the valley of Davos, from stable to unstable, despite the low temperature in winter and potential changes in emissions to a smaller extent. The CO2 emitted in Davos accumulates in the valley, with peaks of up to 250 ppm in comparison with the background mixing ratios. Mixing ratios observed at the mountain site are systematically lower than at the downtown site, supporting the assumption of background air at the mountain site and the absence of strong local pollution sources for CO2. The low variability at the mountain site shows the absence of other sources in the area, despite three minor events that correspond to elevated mixing ratios at the downtown site. These three events may correspond to synoptic events transporting air masses with elevated CO2 from Davos, from the surrounding valleys, or from other parts of Europe. The site-to-site mixing ratio differences (Fig. 4) show no correlation with temperature and are strongly influenced by the local atmospheric dynamics. The importance of the PBL dynamics (vertical mixing and PBL depth) in a valley of limited dimensions requires the use of the high-resolution modeling system to extract emission signals in the observed atmospheric mixing ratios. The results of the inverse system are presented in section 3b.
Observed CO2 hourly mixing ratios (ppm) at the downtown site (blue) filtered with a minimum value of 0.3 m s−1 of friction velocity for well-mixed conditions and at the mountain site (green).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed CO2 hourly mixing ratios (ppm) at the downtown site (blue) filtered with a minimum value of 0.3 m s−1 of friction velocity for well-mixed conditions and at the mountain site (green).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed CO2 hourly mixing ratios (ppm) at the downtown site (blue) filtered with a minimum value of 0.3 m s−1 of friction velocity for well-mixed conditions and at the mountain site (green).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and modeled (red) intersite mole fraction differences (ppm) between the downtown site and the mountain site filtered using the friction velocity criteria (daytime values only). Modeled mole fractions were simulated at 1.33-km resolution by the WRF-FDDA modeling system coupled to the constant prior emissions used in the inverse study.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and modeled (red) intersite mole fraction differences (ppm) between the downtown site and the mountain site filtered using the friction velocity criteria (daytime values only). Modeled mole fractions were simulated at 1.33-km resolution by the WRF-FDDA modeling system coupled to the constant prior emissions used in the inverse study.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and modeled (red) intersite mole fraction differences (ppm) between the downtown site and the mountain site filtered using the friction velocity criteria (daytime values only). Modeled mole fractions were simulated at 1.33-km resolution by the WRF-FDDA modeling system coupled to the constant prior emissions used in the inverse study.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
The atmospheric CH4 mixing ratios were not used in the inverse system but are presented in Fig. 5 as a distinct GHG emitted from different local and distant sources. In comparison with the observed CO2 mixing ratios, the background mixing ratios are highly variable at the mountain site. The minimum values at the downtown site remain systematically higher than the mountain site, but sources out of the valley of Davos contribute to the observed variability as shown by the large variability at the mountain site. Potential sources of methane in the area are primarily farming but may also include waste water treatment plants, landfills, or during combustion of fossil fuels. No clear correlation was found between the wind direction and the peaks, indicating several nearby sources in different directions. Similarly to CO2, the diurnal amplitude at the downtown site is large (about 200 ppb), and daytime periods correspond clearly to low mixing ratios. Stable atmospheric conditions combined with urban emissions of CH4 within Davos from farming as well as waste water and traffic are responsible for high mixing ratios during nighttime at the downtown site (up to 2200 ppb).
As in Fig. 3, but for CH4.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
As in Fig. 3, but for CH4.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
As in Fig. 3, but for CH4.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
b. Impact of temperature and the WEF-2012 meeting on CO2 emissions
First, we present the intersite differences for the observed and the modeled mole fractions in Fig. 4 using the friction velocity criteria. The day-to-day variability is about 15 ppm, with maximum observed differences of about 100 ppm. The interpretation of modeled versus observed intersite mole fraction differences is limited by the large variability, with the exception of the month of February, which shows a clear underestimation of the simulated atmospheric mole fractions. The daily calculated emissions derived from inverse analysis are presented in Fig. 6 with the heating degree-day (HDD), which represents the temperature difference between the outside and the indoor air (15.5°C). The instruments measuring the atmospheric mixing ratios of GHGs were deployed from 23 December 2011 to 3 March 2012. During the first half of the deployment, the outdoor temperature was close to the climatological average for the area (about −6°C). During the second half, a cold wave affected a large part of western Europe and, more severely, central and eastern Europe, as well as North Africa and western Asia, from 27 January until 17 February. The intensity of the 2012 cold wave in February was exceptional, especially in France, Switzerland, and Germany, being one of the 10 most intense events in Zurich since 1864. In Davos, the minimum temperature reached −24.5°C on 4 February (from the WMO station in Davos, WMO identifier 6784). Finally, during the last days of the campaign, the temperature increased with a corresponding HDD of less than 15°C (starting from 20 February).
Daily CO2 emissions from the inverse system smoothed by applying a 3-day running mean from 23 December to 29 February (red) shown as a deviation from the constant prior emissions (percent) and daily HDDs using a reference temperature of 15.5°C (black line). The WEF-2012 meeting period is indicated by the thin black horizontal line (25–29 Jan 2012).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Daily CO2 emissions from the inverse system smoothed by applying a 3-day running mean from 23 December to 29 February (red) shown as a deviation from the constant prior emissions (percent) and daily HDDs using a reference temperature of 15.5°C (black line). The WEF-2012 meeting period is indicated by the thin black horizontal line (25–29 Jan 2012).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Daily CO2 emissions from the inverse system smoothed by applying a 3-day running mean from 23 December to 29 February (red) shown as a deviation from the constant prior emissions (percent) and daily HDDs using a reference temperature of 15.5°C (black line). The WEF-2012 meeting period is indicated by the thin black horizontal line (25–29 Jan 2012).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
The initial annual inventory estimate was calibrated during the first month of the campaign, before the WEF-2012 meeting. Daily emissions were 40% above the initial estimate on average over the month of January. This first estimate is used to provide a corrected value for the 2012 winter season. Over the two months of the campaign, the corresponding daily emissions from the atmospheric inverse system are highly correlated with the HDD (correlation coefficient R = 0.67), which is consistent with the contribution of house heating from the initial inventory [76% of the total direct emissions (Walz et al. 2008)]. The contribution from house heating and traffic are presented in the next section using the CO/CO2 concentration ratio (see section 3c). The maximum of daily emissions occurred during the maximum of intensity of the cold wave, during which the daily emissions more than doubled for a week. During the WEF-2012 meeting, before the start of the cold wave, the inverse daily emissions decreased by 35% relative to emissions in January, despite similar temperatures. The decrease in emissions during the WEF-2012 meeting is discussed in sections 4a and 4b. Finally, the last period of the deployment (starting on 20 February) showed a decrease in the daily emissions as the HDD decreased and reached its minimum over the entire campaign.
c. Composite diurnal cycle of CO/CO2 emission ratios
Both sites measured carbon monoxide (CO) continuously in addition to CO2 and CH4. The averaged diurnal cycles were computed at the hourly time scale for three periods: before, during, and after the WEF-2012 meeting. The diurnal cycles for CO2 and CO are shown in Fig. 7 for the mountaintop site and the downtown site. Whereas the downtown CO2 signal reveals a clear and pronounced diurnal cycle indicating the presence of local pollution sources, the mountaintop site shows almost no variability during the day, indicating that all the variability in CO2 observed at the mountaintop site is from more distant sources, along with variability in the background. Therefore, the enhancement in carbon dioxide observed in the urban site is well represented by simply subtracting the mountain site from the urban site.
Diurnal cycles for (a) CO and (b) CO2 from the mountaintop site (black) and the downtown site (red). The early morning rush hour correlates with a traffic peak for both gases (0700–0900 LT). The minimum occurs at midday for CO2 with low traffic and at night (0400 LT) for CO. Error bars correspond to the hourly RMSs over the period.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Diurnal cycles for (a) CO and (b) CO2 from the mountaintop site (black) and the downtown site (red). The early morning rush hour correlates with a traffic peak for both gases (0700–0900 LT). The minimum occurs at midday for CO2 with low traffic and at night (0400 LT) for CO. Error bars correspond to the hourly RMSs over the period.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Diurnal cycles for (a) CO and (b) CO2 from the mountaintop site (black) and the downtown site (red). The early morning rush hour correlates with a traffic peak for both gases (0700–0900 LT). The minimum occurs at midday for CO2 with low traffic and at night (0400 LT) for CO. Error bars correspond to the hourly RMSs over the period.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
The CO diurnal cycle at the downtown site (Fig. 7b) shows two distinct peaks (morning and afternoon) similar to the observed peaks in CO2. At the opposite, the mountain site cycle shows strong enhancement around midday, with a CO enhancement of about 80 ppb above the nocturnal baseline. The absence of concurrent enhancement of CO2 may correspond to a local source with a very high CO/CO2 ratio, potentially photochemical production of CO from volatile organic compounds in the atmosphere, combined with the valley circulation with upwelling of polluted air from Davos. Other sources of CO in the area may have contributed to the observed CO enhancement, for example, Arosa, located southwest of the mountain site in a large valley. This absence of CO2 enhancement indicates that the mountain site cannot be used as a background signal for the urban enhancement during the midday hours. For the analysis, the minimum CO value observed at the mountaintop site within a given 24-h period will be used as the baseline value for CO for the urban site.
From Walz et al. (2008), local direct emissions sources to Davos include no significant power-generation facilities and limited industrial activity. On a yearly basis, about 17% of the direct CO2 emissions are from road transit and 76% are from fossil fuel–driven heating. Although the estimates are yearly, the emissions for both transit and heating will vary during the year because of changing weather and tourist populations. Further, these emissions will not be distributed uniformly over the day. Consumption of fossil fuels for heating will be driven primarily by ambient temperature and will be greatest during the late night and early morning hours. Transit emissions will be greatest during the daylight hours and will drop dramatically in the middle of the night when there is very little traffic. The heating sources also vary substantially in their CO/CO2 emission ratio, with estimates by the U.S. Environmental Protection Agency of 
Figure 8 shows the observed CO/CO2 ratio diurnal cycle composite for the three different periods. The ratio is much lower in the late night and early morning hours than it is during the daylight hours. The signal between 0000 and 0600 LT averages about 
Diurnal cycle composite of the CO/CO2 emission ratio (ppm/ppb) as a function of time of day (local time), for the time period (a) before, (b) during, and (c) after the WEF-2012. Values are lower during the WEF-2012 meeting relative to the first period whereas temperatures are similar, indicating a decrease in traffic. The last period [(c)] corresponds to the severe cold wave affecting Europe with an increase in the house heating contribution.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Diurnal cycle composite of the CO/CO2 emission ratio (ppm/ppb) as a function of time of day (local time), for the time period (a) before, (b) during, and (c) after the WEF-2012. Values are lower during the WEF-2012 meeting relative to the first period whereas temperatures are similar, indicating a decrease in traffic. The last period [(c)] corresponds to the severe cold wave affecting Europe with an increase in the house heating contribution.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Diurnal cycle composite of the CO/CO2 emission ratio (ppm/ppb) as a function of time of day (local time), for the time period (a) before, (b) during, and (c) after the WEF-2012. Values are lower during the WEF-2012 meeting relative to the first period whereas temperatures are similar, indicating a decrease in traffic. The last period [(c)] corresponds to the severe cold wave affecting Europe with an increase in the house heating contribution.
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
The daytime CO/CO2 ratio is lower after the WEF-2012 than it is before the meeting. We note that the temperatures were significantly colder during the last period (cold wave). The observed ratios are consistent with the idea that the emissions from heating were higher during that period. During the WEF-2012, the ratios are lower than before the WEF-2012, despite the fact that the temperatures were similar. This may indicate that the drop in emissions during the WEF-2012 is due to a reduction in traffic emissions, at least in part.
4. Discussion
a. Evaluation of systematic transport errors
This first attempt to monitor daily emissions over a small-sized city using a top-down approach captured two pieces of information: large day-to-day variability and weather-related events over several days. However, the region is challenging for current mesoscale models, mainly because of the steep topography and the stability conditions at high altitudes during winter in the Alps (snow on the ground, low temperatures, and local atmospheric dynamics). For these reasons, systematic errors may affect the atmospheric transport, especially the representation of the vertical mixing under stable or neutral conditions with low turbulent surface fluxes, driven primarily by temperature at the surface and wind shear in the valley. To evaluate the model errors, a flux site was deployed, including a 3D sonic anemometer, to measure the friction velocity and the buoyancy flux in the valley of Davos and to quantify the modeling performance during the campaign. Figure 9 shows the modeled and observed friction velocity at the downtown site over the campaign whereas Fig. 10 shows the buoyancy flux in watts per square meter. Considering the friction velocity, two periods reveal larger mismatch around 5 January and during the cold wave in early February. In both cases, the WRF model overestimated the vertical mixing, which suggests that the emissions could be overestimated during these periods. The inverse estimates may increase because of the lower modeled mixing ratios (dilution errors in the PBL). Overall, the friction velocity model–data mismatch is 0.07 ± 0.16 m s−1 between 1200 and 1700 LT over the campaign. The buoyancy flux is low over the period, from 10 to 40 W m−2 for most days during the deployment period. The mean horizontal wind speed is presented in Fig. 11. The mean difference is low over the period (−0.21 m s−1) with a standard deviation of 2.1 m s−1. The WRF-FDDA model was able to capture the observed daily variability with no systematic over- or underestimation of both the buoyancy flux and the mean horizontal wind speed on average over the period. The observed buoyancy flux shows a larger variability because of remaining turbulent structures measured at high frequency. The temperature at 2 m shown in Fig. 12 was also overestimated in early February, contributing directly to the buoyancy term in the surface energy budget by an excess of mixing due to heat transfer. On average, the temperature shows a low bias (0.8°C) and a standard deviation of 2.4°C. As a comparison, the wind speed in the valley does not show any positive bias during the same period (see Fig. 11) and agrees with the measured wind speed, with an averaged model–data mismatch of 0.2 ±2.0 m s−1. However, only a portion of the cold wave period is affected by positive model–data mismatch. This suggests that even though the average daily emissions during the cold wave may be overestimated for 5 days, the previous and following days seem consistent and should provide reasonable flux estimates.
(a) Observed and (b) observed minus modeled friction wind velocity (modeled at 1.33-km resolution by the WRF-FDDA modeling system) at the surface over the campaign (m s−1).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
(a) Observed and (b) observed minus modeled friction wind velocity (modeled at 1.33-km resolution by the WRF-FDDA modeling system) at the surface over the campaign (m s−1).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
(a) Observed and (b) observed minus modeled friction wind velocity (modeled at 1.33-km resolution by the WRF-FDDA modeling system) at the surface over the campaign (m s−1).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and simulated (red) buoyancy fluxes at 1.33-km resolution by the WRF-FDDA modeling system over the campaign (W m−2).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and simulated (red) buoyancy fluxes at 1.33-km resolution by the WRF-FDDA modeling system over the campaign (W m−2).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and simulated (red) buoyancy fluxes at 1.33-km resolution by the WRF-FDDA modeling system over the campaign (W m−2).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Mean horizontal wind speed differences (modeled minus observed) at 1.33-km resolution by the WRF-FDDA modeling system over the campaign (m s−1). The dashed line indicates the mean difference over the period (−0.21 m s−1).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Mean horizontal wind speed differences (modeled minus observed) at 1.33-km resolution by the WRF-FDDA modeling system over the campaign (m s−1). The dashed line indicates the mean difference over the period (−0.21 m s−1).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Mean horizontal wind speed differences (modeled minus observed) at 1.33-km resolution by the WRF-FDDA modeling system over the campaign (m s−1). The dashed line indicates the mean difference over the period (−0.21 m s−1).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and simulated (red) potential temperatures at 1.33-km resolution by the WRF-FDDA modeling system at the 2-m elevation over the campaign (°C).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and simulated (red) potential temperatures at 1.33-km resolution by the WRF-FDDA modeling system at the 2-m elevation over the campaign (°C).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Observed (blue) and simulated (red) potential temperatures at 1.33-km resolution by the WRF-FDDA modeling system at the 2-m elevation over the campaign (°C).
Citation: Journal of Applied Meteorology and Climatology 52, 12; 10.1175/JAMC-D-13-038.1
Considering the WEF-2012 meeting period, the WRF model shows low model–data mismatch for the friction velocity, the buoyancy flux, and the 2-m temperature. The month of January exhibits an RMS of 0.17 m s−1 and a mean of 0.4 m s−1 in model–data mismatch of friction velocity, with no significant difference during the WEF-2012 meeting. It suggests that the decrease of the emissions during the WEF-2012 meeting relative to the emissions over the entire month of January is not due to biases in the vertical mixing. The model–data mismatch for the buoyancy flux has a mean of −0.32 W m−2 over the period and the RMS is about 12.6 W m−2, showing no systematic differences over the period (see Fig. 10). However, the low vertical mixing could induce a limited extent of the concentration footprint of the downtown site, which means that the estimated daily emissions may not represent the entire city of Davos. It seems unlikely that the concentration footprints are much smaller than 6 km2, whereas footprints from Lagrangian particle dispersion models cover several tenths of square kilometers even in stable conditions (Lauvaux et al. 2008).
The diurnal cycle composite of the CO/CO2 ratio that corresponds to the contributions of house heating and traffic is consistent with previous studies in urban environments (van der Laan et al. 2010; Levin and Karstens 2007; Vogel et al. 2010). The CO/CO2 ratio showed clear evidence of changes in source contributions, between the traffic and the house heating components. Considering that the CO/CO2 ratio decreased during the WEF-2012 meeting, the decrease in the emissions may be related to a decrease in traffic emissions in the area. Because of the lack of available information about traffic intensity during the meeting, this assumption remains valid but cannot be confirmed.
In addition, the daily estimates were performed using different threshold values of friction velocity, to assure the well-mixed criteria and better modeling performance. Despite removing several days of observations over the campaign based on the threshold values for the friction velocity (e.g., >0.3 m s−1), the observed variability in emissions remained unchanged, that is, an increase during the cold wave event in early February and a decrease after the cold wave. The presence of a heliport during the WEF-2012 meeting may impact the stability conditions in the valley, with an increased entrainment of air from the free troposphere into the PBL. The vertical mixing due to helicopters taking off near the city may explain part of the apparent decrease in emissions.
b. Applicability of the method
The campaign deployment and the daily emission estimates provided several insights to emissions at very finescale and high temporal resolution. For the first time, a large-scale event in a limited-size city was monitored in real time. The decrease in emissions seems counterintuitive at the first order and raised several questions and concerns about the capability of the present inverse system. However, the evaluation of the modeling system and the use of CO as an additional tracer for fossil fuel emissions seem in agreement with the observed decrease of the emissions during the WEF-2012 meeting. The increase during the cold wave showed that major changes can be detected by our system and was confirmed by the observed lower CO/CO2 ratios. Still, several limitations remain, such as the size of the concentration footprints in stable conditions or the potential changes in fossil fuel sources (e.g., closure of public places located in the city for the duration of the WEF-2012 meeting). The proximity of the background site, which may be impacted by emissions from Davos in the afternoon (uplift of air from the valley as indicated by elevated CO concentrations) or by air masses coming from other cities in the area, may affect our emission estimates. A second background site may have helped to identify CO2 air masses from other valleys, but other observations tend to indicate that changes in the economic activity level may explain the observed decrease with limited traffic around the city because of security measures. Finally, no clear evidence contradicts our current findings, that is, model errors or CO/CO2 ratios, but further investigations would be needed to confirm the causes of the decrease during the WEF-2012 meeting, including external information on the traffic and additional measurement locations to cross evaluate the present findings.
Acknowledgments
Thanks are given to Pr. Ariane Walz from the Potsdam Institute for Climate Impact Research (PIK) for fruitful discussions, Michael Lehning and Marc Ruesch from the Schnee- und Lawinenforschung (SLF) for their support during field deployment, Colm Sweeney from the National Oceanic and Atmospheric Administration (NOAA/ESRL/GMD division) for discussions about the deployment, the canton of Davos for hosting the instruments, and Gian-Paul Calonder for support during field deployment and fruitful discussions about this study. This work was funded by The Pennsylvania State University, in collaboration with Picarro, Inc.; Climmod Engineering; and Sigma Space Corp.
REFERENCES
Bishop, G., and G. Stedman, 2008: A decade of on-road emissions measurements. Environ. Sci. Technol., 42, 1651–1656.
Bocquet, M., 2005: Grid resolution dependence in the reconstruction of an atmospheric tracer source. Nonlinear Processes Geophys., 12, 219–233, doi:10.5194/npg-12-219-2005.
Bousquet, P., P. Peylin, P. Ciais, C. L. Quéré, P. Friedlingstein, and P. P. Tans, 2000: Regional changes in carbon dioxide fluxes of land and oceans since 1980. Science, 290, 1342–1346.
Chen, H., and Coauthors, 2010: High-accuracy continuous airborne measurements of greenhouse gases (CO2 and CH4) using the cavity ring-down spectroscopy (CRDS) technique. Atmos. Meas. Tech.,3, 375–386, doi:10.5194/amt-3-375-2010.
Deng, A., and D. Stauffer, 2006: On improving 4-km mesoscale model simulations. J. Appl. Meteor., 45, 361–381.
Deng, A., N. Seaman, G. Hunter, and D. Stauffer, 2004: Evaluation of interregional transport using the MM5-SCIPUFF system. J. Appl. Meteor., 43, 1864–1886.
Deng, A., and Coauthors, 2009: Update on WRF-ARW end-to-end multi-scale FDDA system. Extended Abstracts, 10th WRF Users’ Workshop, Boulder, CO, National Center for Atmospheric Research, 1.9. [Available online at http://www.mmm.ucar.edu/wrf/users/workshops/WS2009/abstracts/1-09.pdf.]
Deng, A., T. Lauvaux, K. Davis, N. Miles, S. Richardson, and D. Stauffer, 2012a: A WRF-Chem realtime modeling system for monitoring CO2 emissions. Extended Abstracts, 13th Annual WRF Users’ Workshop, Boulder, CO, P33. [Available online at https://www.regonline.com/AttendeeDocuments/1077122/43387256/43387256_1045173.pdf.]
Deng, A., D. Stauffer, B. Gaudet, and G. Hunter, 2012b: A rapidly relocatable high-resolution WRF system for military-defense, aviation and wind energy. Extended Abstracts, 13th Annual WRF Users’ Workshop, Boulder, CO, 2.6. [Available online at https://www.regonline.com/AttendeeDocuments/1077122/43387256/43387256_1045166.pdf.]
Energy Information Administration, 2007: Documentation for emissions of greenhouse gases in the United States 2005. U.S. Dept. of Energy Rep. DOE/EIA-0638, 240 pp. [Available online at http://www.eia.gov/oiaf/1605/ggrpt/documentation/pdf/0638(2005).pdf.]
Enting, I., 1998: Attribution of greenhouse gas emissions, concentrations and radiative forcing. CSIRO Atmospheric Research Tech. Paper 38, 27 pp. [Available online at http://unfccc.int/files/meetings/workshops/other_meetings/application/pdf/enting_2000c.pdf.]
Gaudet, B., D. Stauffer, N. Seaman, A. Deng, J. Pleim, R. Gilliam, K. Schere, and R. Elleman, 2009: Modeling extremely cold stable boundary layers over interior Alaska using a WRF FDDA system. Preprints, 13th Conf. on Mesoscale Processes, Salt Lake City, UT, Amer. Meteor. Soc., 18.1. [Available online at https://ams.confex.com/ams/pdfpapers/155043.pdf.]
Göckede, M., D. P. Turner, A. M. Michalak, D. Vickers, and B. E. Law, 2010: Sensitivity of a subregional scale atmospheric inverse CO2 modeling framework to boundary conditions. J. Geophys. Res.,115, D24112, doi:10.1029/2010JD014443.
Grell, G., S. E. Peckham, R. Schmitz, S. A. McKeen, G. Frost, W. C. Skamarock, and B. Eder, 2005: Fully coupled online chemistry within the WRF model. Atmos. Environ., 39, 6957–6975.
Gurney, K. R., D. Mendoza, Y. Zhou, M. Fischer, C. Miller, S. Geethakumar, and S. de le Rue du Can, 2009: High resolution fossil fuel combustion CO2 emission fluxes for the United States. Environ. Sci. Technol., 43, 5535–5541, doi:10.1021/es900806c.
Gurney, K. R., I. Razlivanov, Y. Song, Y. Zhou, B. Benes, and M. Abdul-Massih, 2012: Quantification of fossil fuel CO2 emissions on the building/street scale for a large U.S. city. Environ. Sci. Technol., 46, 12 194–12 202, doi:10.1021/es3011282.
Hanna, S. R., R. Paine, D. Heinold, E. Kintigh, and D. Baker, 2007: Uncertainties in air toxics calculated by the dispersion models AERMOD and ISCST3 in the Houston ship channel area. J. Appl. Meteor. Climatol.,46, 1372–1382.
IEA, 2008: World Energy Outlook 2008.OECD/IEA, 578 pp.
Lac, C., and Coauthors, 2012: CO2 dispersion modelling over paris region within the CO2-MEGAPARIS project. Atmos. Chem. Phys. Discuss., 12, 28 155–28 193, doi:10.5194/acpd-12-28155-2012.
Lauvaux, T., and Coauthors, 2008: Mesoscale inversion: First results from the CERES campaign with synthetic data. Atmos. Chem. Phys., 8, 3459–3471, doi:10.5194/acp-8-3459-2008.
Lauvaux, T., and Coauthors, 2009: Bridging the gap between atmospheric concentrations and local ecosystem measurements. Geophys. Res. Lett.,36, L19809, doi:10.1029/2009GL039574.
Lauvaux, T., and Coauthors, 2012: Constraining the CO2 budget of the corn belt: Exploring uncertainties from the assumptions in a mesoscale inverse system. Atmos. Chem. Phys., 12, 337–354, doi:10.5194/acp-12-337-2012.
Levin, I., and U. Karstens, 2007: Inferring high-resolution fossil fuel CO2 records at continental sites from combined 14CO2 and CO observations. Tellus,59B, 245–250, doi:10.1111/j.1600-0889.2006.00244.x.
Lowry, D., C. W. Holmes, N. D. Rata, P. O’Brien, and E. G. Nisbet, 2001: London methane emissions: Use of diurnal changes in concentration and δ13C to identify urban sources and verify inventories. J. Geophys. Res., 106, 7427–7448.
Marland, G., A. Brenkert, and J. Olivier, 1999: CO2 from fossil fuel burning: A comparison of ORNL and EDGAR estimates of national emissions. Environ. Sci. Policy, 2, 265–273, doi:10.1016/S1462-9011(99)00018-0.
Masson, V., C. Grimmond, and T. Oke, 2002: Evaluation of the town energy balance (TEB) scheme with direct measurements from dry districts in two cities. J. Appl. Meteor., 41, 1011–1026.
Mays, K. L., P. B. Shepson, B. H. Stirm, A. Karion, C. Sweeney, and K. R. Gurney, 2009: Aircraft-based measurements of the carbon footprint of Indianapolis. Environ. Sci. Technol., 43, 7816–7823, doi:10.1021/es901326b.
Nisbet, E., and R. Weiss, 2010: Top-down versus bottom-up. Science, 328, 1241–1243, doi:10.1126/science.1189936.
Office of Air Quality Planning and Standards, 1995a: Fuel oil combustion. Stationary Point and Area Sources, Vol. 1, Compilation of Air Pollutant Emission Factors, Tech. Publ. AP-42, 5th ed., U.S. Environmental Protection Agency, 1.3-1–1.3-35. [Available online at http://www.epa.gov/ttnchie1/ap42/ch01/final/c01s03.pdf.]
Office of Air Quality Planning and Standards, 1995b: Natural gas combustion. Stationary Point and Area Sources, Vol. 1, Compilation of Air Pollutant Emission Factors, Tech. Publ. AP-42, 5th ed., U.S. Environmental Protection Agency, 1.4-1–1.4-10. [Available online at http://www.epa.gov/ttn/chief/ap42/ch01/final/c01s04.pdf.]
Olivier, J., A. Bouwman, J. Berdowski, C. Veldt, J. Bloos, A. Visschedijk, C. van der Maas, and P. Zandveld, 1999: Sectoral emission inventories of greenhouse gases for 1990 on a per country basis as well as on 1° × 1°. Environ. Sci. Policy, 2, 241–263, doi:10.1016/S1462-9011(99)00027-1.
Peters, W., and Coauthors, 2007: An atmospheric perspective on north american carbon dioxide exchange: CarbonTracker. Proc. Natl. Acad. Sci. USA, 104, 18 925–18 930, doi:10.1073/pnas.0708986104.
Reen, B., D. R. Stauffer, K. Davis, and A. R. Desai, 2006: A case study on the effects of heterogeneous soil moisture on mesoscale boundary-layer structure in the southern Great Plains, U.S.A. Part II: Mesoscale modelling. Bound.-Layer Meteor., 120, 275–314, doi:10.1007/s10546-006-9056-6.
Rella, C. W., and Coauthors, 2012: High accuracy measurements of dry mole fractions of carbon dioxide and methane in humid air. Atmos. Meas. Tech. Discuss.,5, 5823–5888, doi:10.5194/amtd-5-5823-2012.
Rogers, R. E., A. Deng, D. R. Stauffer, B. J. Gaudet, Y. Jia, S.-T. Soong, and S. Tanrikulu, 2013: Application of the Weather Research and Forecasting model for air quality modeling in the San Francisco Bay area. J. Appl. Meteor. Climatol.,52, 1953–1973.
Schuh, A. E., A. S. Denning, K. D. Corbin, I. T. Baker, M. Uliasz, N. Parazoo, A. E. Andrews, and D. E. J. Worthy, 2010: A regional high-resolution carbon flux inversion of North America for 2004. Biogeosciences, 7, 1625–1644, doi:10.5194/bg-7-1625-2010.
Schuh, A. E., and Coauthors, 2013: Evaluating atmospheric CO2 inversions at multiple scales over a highly inventoried agricultural landscape. Global Change Biol.,19, 1424–1439, doi:10.1111/gcb.12141.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/docs/arw_v3.pdf.]
Stauffer, D., and N. Seaman, 1994: Multiscale four-dimensional data assimilation. J. Appl. Meteor., 33, 416–434.
Strong, C., C. Stwertka, D. R. Bowling, B. B. Stephens, and J. R. Ehleringer, 2011: Urban carbon dioxide cycles within the Salt Lake Valley: A multiple-box model validated by observations. J. Geophys. Res., 116, D15307, doi:10.1029/2011JD015693.
Traynor, G. W., M. G. Apte, and G.-M. Chang, 1996: Pollutant emission factors from residential natural gas appliances: A literature review. Lawrence Berkeley National Laboratory Rep. LBL-38123, 20 pp. [Available online at http://escholarship.org/uc/item/49z4g2xr.]
Turnbull, J. C., and Coauthors, 2011: Assessment of fossil fuel carbon dioxide and other anthropogenic trace gas emissions from airborne measurements over Sacramento, California in spring 2009. Atmos. Chem. Phys., 11, 705–721, doi:10.5194/acp-11-705-2011.
van der Laan, S., U. Karstens, R. Neubert, I. van der Laan-Luijkx, and H. Meijer, 2010: Observation-based estimates of fossil fuel-derived CO2 emissions in the Netherlands using Δ14C, CO, and 222Radon. Tellus,62B, 389–402, doi:10.1111/j.1600-0889.2010.00493.x.
Vogel, F., S. Hammer, A. Steinhof, B. Kromer, and I. Levin, 2010: Implication of weekly and diurnal 14C calibration on hourly estimates of CO-based fossil fuel CO2 at a moderately polluted site in southwestern Germany. Tellus,62B, 512–520, doi:10.1111/j.1600-0889.2010.00477.x.
Walz, A., G.-P. Calonder, F. Hagedorn, C. Lardelli, C. Lundström, and V. Stöckli, 2008: Regional CO2 budget, countermeasures and reduction aims for the Davos region, Switzerland. Energy Policy, 36, 811–820.
West, T., G. Marland, N. Singh, B. Bhaduri, and A. B. Roddy, 2009: The human carbon budget: An estimate of the spatial distribution of metabolic carbon consumption and release in the United States. Biogeochemistry,94, 29–41, doi:10.1007/s10533-009-9306-z.
