1. Introduction
It is well known that carbon-induced global warming by human activities that use fossil-fuel combustion has become a serious issue for climate change. About 30%–40% of anthropogenic greenhouse gases (GHGs) are emitted from urban areas (Seto et al. 2014), and these emissions are associated with an increase of average temperature, higher intensity and occurrence of severe weather, and an increase and/or decrease of precipitation (IPCC et al. 2013). Given its high concentration and emissions, carbon dioxide (CO2) is the GHG most in need of monitoring, despite its lower global warming potential (GWP) relative to some other GHGs (e.g., methane, nitrous oxide, and sulfur hexafluoride respectively have 25, 298, and 22 800 times the GWP of CO2 over a 100-yr time horizon; Forster et al. 2007).
To understand better the effect of all emissions of CO2 on the atmospheric carbon cycle, a wide variety of studies have been carried out using both bottom-up and top-down approaches. In the bottom-up approach, inventory statistics of fossil-fuel consumption from each source sector, including the residential, commercial, industrial, and transportation sectors, are combined with estimates of CO2 emissions with carbon contents of each fuel type. From this approach, estimated CO2 emissions showed an uncertainty of approximately ±5% uncertainty at the global scale (Le Quere et al. 2014) and 50%–200% at the urban scale (Turnbull et al. 2011; Asefi-Najafabady et al. 2014). In contrast, the top-down approach includes atmospheric flux measurements and inversion modeling. Microscale flux-measurement studies have been conducted in vegetation-dominated areas for decades to quantify and understand the interactions between the surface and the atmosphere in the carbon budget, mainly focusing on the net ecosystem exchange (NEE) of CO2; these studies include “AmeriFlux” and “FLUXNET” (Baldocchi et al. 2001). The accuracy of the flux measurements is generally limited by the local features of land use, emission sources, and climate, however. Global-scale inverse modeling has been carried out to retrieve NEE from observed concentrations (Gurney et al. 2009; Peylin et al. 2005; Tans et al. 1990). In addition to the fact that the lack of spatially explicit a priori flux estimates can cause large uncertainties in this approach, however, the resolution of global inverse modeling is too coarse to capture the spatiotemporal resolution of microscale flux measurements. Several studies have been conducted to diagnose the mismatch between global-scale models and flux measurements (Gerbig et al. 2003; Pérez-Landa et al. 2007; van der Molen and Dolman 2007).
To fill the scale gap between the global inverse modeling and the microscale measurements, we require modeling of regional-scale CO2 transport with finer spatiotemporal resolution. Accurate simulations of mesoscale transport are important to estimate the CO2 transport, variability, and budget, which can only be established with high-resolution mesoscale models that include the transport of CO2 and the CO2 exchange flux between the biosphere and the atmosphere. Ahmadov et al. (2007, 2009) coupled a meteorological model, the Weather Research and Forecasting (WRF) Model, with the Vegetation Photosynthesis and Respiration Model (VPRM) and conducted CO2 modeling over Europe. The results showed significant improvement in capturing mesoscale circulations and observed CO2 spatiotemporal variation, which had not been represented well in global models. Pillai et al. (2011) also validated the coupled model over mountain terrain by comparing it with tall-tower measurements at Mount Ochsenkopf in Germany. Jamroensan (2013) conducted VPRM parameter optimization for the newly added vegetation classes over the U.S. Midwest and estimated the interactions between the areas growing soybeans and the atmosphere. These previous studies have been mostly conducted over vegetation-dominated and relatively heterogeneous areas. In urban areas, anthropogenic emission sources and vegetation sinks of CO2 are complexly mixed. Despite the difficulties encountered in setting up flux-measurement platforms in city areas, urban flux-measurement studies have recently been conducted that focus on the CO2 cycle, driven by both anthropogenic emission sources and biospheric uptake (Park and Schade 2016; Velasco et al. 2013; Kotthaus and Grimmond 2012).
In the United States, urban areas have been developing rapidly by population increase and land-use modification during the last five decades (Auch et al. 2004). The main anthropogenic CO2 emission sources are land-use changes affecting interactions between the surface and the atmosphere and fossil-fuel combustion for energy, transportation, and industrial processes, as reported in the U.S. GHG inventory report of the Environmental Protection Agency (http://www.epa.gov/climatechange/ghgemissions/usinventoryreport.html). Carbon dioxide emissions account for approximately 82% of all U.S. GHG emissions from human activity (National Research Council 2010). California is one of the most significant carbon emitters in the United States, and there have been many efforts (including Assembly Bill 32: the “Global Warming Solutions Act of 2006”) and reports addressing reduction of carbon emissions. The main sources of CO2 in California are the transportation (~39%), industrial (~23%), electric-power (~19%), agricultural (~8%), residential (~6%), and commercial (~5%) sectors, according to the 2015 state energy-related CO2-emission report of the Energy Information Administration (EIA; https://www.arb.ca.gov/cc/inventory/data/data.htm).
The Southern California Air Basin (SoCAB), designated by the state government of California in 1969 for the purpose of air-quality management in Southern California, includes all of Orange County and the nondesert regions of Los Angeles (LA), Riverside, and San Bernardino Counties. This region is bounded on the south and the west by the Pacific Ocean, on the north by the San Gabriel (3068 m) and San Bernardino (3505 m) Mountains, on the northwest by Mount Santa Susana (1142 m) and the Simi Hills (652 m), and on the east by Mount San Jacinto (3302 m) and Mount Santa Rosa (2657 m). This topography causes unique meteorological flows in SoCAB, combining with the sea breeze that is linked with “Catalina” eddies. When low pressure forms over the desert in southern California and Arizona and high pressure builds over the Pacific Ocean off the coast at the same time during spring, northerlies flow down to the California coast, turning toward the coast of Southern California, interacting with the islands and the shape of the coast, and forming a center of eddies near Catalina Island, the so-called Catalina eddies. The swirling winds off the coast of Southern California transport large amounts of polluting emissions and finally accumulate them within SoCAB because the mountains block ventilation out of the basin. Thus, SoCAB shows relatively higher concentration of tracers when compared with areas outside SoCAB (Ryerson et al. 2013; Conil and Hall 2006; Ulrickson and Mass 1990). In 2010, therefore, the California Research at the Nexus of Air Quality and Climate Change (CalNex) campaign was conducted in and over LA and Sacramento to research issues related to air quality and climate change (Angevine et al. 2012).
Brioude et al. (2013) recently performed inverse modeling over SoCAB for the CalNex 2010 campaign and showed 31%–44% higher CO2 emissions in 2010 as posterior than in 2002. Newman et al. (2013) used CO2 ground measurements during the same study period and estimated that local fossil-fuel combustion contributed up to ~50% overnight and ~100% near midday. Feng et al. (2016) more recently conducted sensitivity studies in terms of the impact of the model frame and the fossil-fuel CO2 emissions priors’ spatial resolution [emissions from the “Vulcan” database at 10-km resolution vs LA emissions from the Hestia Project (Hestia-LA) at 1.3 km] on the simulated CO2 concentration using the WRF Model coupled with a chemistry model (WRF-Chem) over the LA megacity during the CalNex 2010 period. They concluded that the higher-resolution simulation better resolved the vertical gradient of meteorological variables and that higher resolution of the fossil-fuel CO2 emission data produced a more improved CO2 simulation.
In this study, the method for the modeling system is similar to the study described by Feng et al. (2016), but our motivation and approach are different. First, we applied the newly updated Hestia-LA emissions data (version 2.1) combined with the Fossil Fuel Data Assimilation System (FFDAS) data and the National Oceanic and Atmospheric Administration (NOAA) CarbonTracker CO2 mole fraction data from the “CT2015” release. Second, because of the significance of the parameters of photosynthesis and respiration in estimating biogenic contributions in specific regions, we optimized the VPRM parameters with flux measurements at each representative vegetation site of FLUXNET. Third, we evaluated the model’s performance by comparing with aircraft measurements as well as ground observations. Last, we estimated the regional CO2 budget over SoCAB.
2. WRF-VPRM model
a. WRF modeling system
A diagnostic vegetation model that computes biospheric CO2 fluxes (VPRM) is coupled to WRF-Chem (version 3.8.1) and is hereinafter referred to as WRF-VPRM. It has been described by Ahmadov et al. (2007) and Mahadevan et al. (2008).We set up and ran WRF by two-way nesting at 36-, 12-, and 4-km resolution on three nested grids (Fig. 1) and 38 vertical layers (with 12 layers below 1.5 km) extending up to 100 hPa, where the lowest scalar-level height was ~27 m. Initial and boundary conditions for meteorological fields and soil initialization fields for the WRF modeling were taken from the 3-h North American Regional Reanalysis dataset with 32-km spatial resolution (Mesinger et al. 2006). For sea surface temperature (SST) fields, the 6-h National Centers for Environmental Prediction SST dataset with 8-km horizontal resolution (
(left) Domains for WRF-VPRM modeling, and (right) land-use distributions for the finest domain, with the locations of CIT (red dot), PV (orange dot), and the FLUXNET sites including coastal sage (dark-blue triangle), grassland (green triangle), and oak/pine forest (light-blue triangle). The border of SoCAB is indicated by the black boundary line. The solid blue line indicates the cross-sectional path for Fig. 4.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
b. Vegetation Photosynthesis and Respiration Model
VPRM is a simple diagnostic biospheric CO2 flux parameterization that uses high-resolution land-use and satellite data along with simulated radiation and air temperature by the WRF Model (Mahadevan et al. 2008). This enables simulation of biospheric CO2 uptake and release fluxes at high spatiotemporal resolution. In several studies (Ahmadov et al. 2007, 2009; Pillai et al. 2011; Feng et al. 2016), the VPRM parameterization was successfully applied to different ecosystems.
VPRM derives biospheric CO2 fluxes using flux-measurement data and MODIS (http://modis.gsfc.nasa.gov/) satellite indices, including an enhanced vegetation index and land surface water index obtained at 500-m spatial resolution with 8-day frequency. VPRM uses eight land-use categories, including evergreen forest, deciduous broadleaf forest, mixed forest, shrubland, savanna, cropland, grassland, and urban or snow/ice, that were classified on the basis of the 1-km global Synergetic Land Cover Product (SYNMAP; Jung et al. 2006). Each vegetation category has its own parameters—so-called VPRM parameters—that should be optimized on the basis of local CO2 flux measurements for each representative land-use category.
The gross ecosystem exchange (GEE) is controlled by light, temperature, humidity, CO2 concentration, soil moisture, nutrient availability, and seasonal leaf foliage, and the respiration is determined by autotrophs (vegetation) and heterotrophs (symbiotic microorganisms in soil) (Bowden et al. 1993; Ryan and Law 2005). In VPRM, GEE was calculated using the shortwave radiation (SWDOWN) generated by WRF combined with VPRM parameters that include the product of the maximum quantum yield λ and the half-saturation value of photosynthetically active radiation PAR0. The respiration rate was calculated from the model’s 2-m air temperature along with first-order linear parameters including a slope α and an intercept β (Jamroensan 2013). VPRM parameters were first optimized by Mahadevan et al. (2008) through nonlinear least squares with the U.S. 1-km International Geosphere–Biosphere Programme classification (Belward et al. 1999), and then the NOAA WRF-Chem group released optimized values for U.S. vegetation as the “default” to the model’s source code (Table 1). The optimization is to minimize the sum of squares of error between predicted NEEs and measured NEEs. First, the measured NEE under the condition of friction velocity u* ≤ 0.1 m s−1 was removed. Then, the growing-season nighttime NEE was used to optimize α, and α was used to obtain the respiration. The measured GEE was calculated by subtracting the respiration rate from the measured NEE, and the measured GEE was used to optimize λ and PAR0. In this approach, the intercept β is set to 0. Further details of the optimization method can be found in Mahadevan et al. (2008) and Jamroensan (2013).
Optimized and default VPRM parameters under the condition of u* > 0.1 m s−1 for this study.
c. CO2 input data
For lateral boundary conditions and initial conditions for CO2 concentration, we used the CT2015 dataset (
In addition to the CT2015 data, hourly fossil-fuel CO2 emissions from the Hestia Project (version 2.1) gridded at 1.0 km for 2010 (Gurney et al. 2012; http://hestia.project.asu.edu/) were used for parallel sets of numerical experiments to feed surface boundary conditions. Because the Hestia emissions only cover SoCAB areas, emissions from outside the Hestia domain were derived from the total anthropogenic emissions from FFDAS (Asefi-Najafabady et al. 2014) and shipping and aviation emissions taken from the Emission Database for Global Atmospheric Research (EDGAR; Petrescu et al. 2012) for all three domains.
In this study, to improve performance and reduce simulation uncertainties induced by VPRM parameters, we optimized VPRM parameters for each vegetation type over Southern California, using CO2 flux measurements at FLUXNET sites (http://www.ess.uci.edu/~california/). The sites used were the oak/pine forest site (33°48′29″N, 116°46′19″W) for evergreen forest, the pine/juniper site (33°36′18″N, 116°27′18″W) for mixed forest, the coastal sage site (33°44′02″N, 117°41′46″W) for shrubland, the desert chaparral site (33°36′36″N, 116°27′00″W) for savanna, and the grassland site (33°44′13″N, 117°41′42″W) for grassland. For the other vegetation classes not measured by FLUXNET sites, the default values were used as in the setup of Feng et al. (2016). Table 1 reports our optimized (posterior) and the default (prior) VPRM parameters.
3. Observation data
To address research issues related to air quality and climate change, the CalNex campaign was conducted in and over Los Angeles, Bakersfield, and Sacramento during May–June 2010; specific information about CalNex 2010 has been published by Ryerson et al. (2013) and posted online by NOAA (http://www.esrl.noaa.gov/csd/calnex/). In situ continuous ground measurements were conducted on a 10-m tower at the California Institute of Technology campus (CIT; 34°08′12″N, 118°07′39″W) in Pasadena, California (Fig. 1). Meteorological variables were measured by various sensors, and the planetary boundary layer heights (PBLH) were retrieved by a Vaisala, Inc., Ceilometer CL31 model [the methods are described in Haman et al. (2012)]. Along with the measurements of chemical compounds such as aerosols and gases, CO2 was measured by a Picarro, Inc., model G1101-i isotopic CO2 analyzer (cavity ring-down spectroscopy). The 10-min averages of meteorological variables and the 15-min averages of PBLH measurements were integrated into the hourly averaged time series for comparison with the WRF-VPRM modeling results at each hour.
The NOAA P-3 aircraft was also instrumented to make vertical profiles of meteorological variables and chemical species. Measurement accuracies were estimated at 0.5°C for temperature, 5% for water vapor, 5° for wind direction, 1 m s−1 for wind speed, and 0.2 ppm for CO2 concentration (Peischl et al. 2012).
4. Results and discussion
a. Meteorological evaluation
The simulated meteorological results were compared with four ground measurements including CIT and three FLUXNET sites (Fig. 1), and their statistics are reported in Table 2. The modeling results captured day-to-day hourly variations of meteorological observations including 2-m temperature and relatively humidity (RH), 10-m winds, and surface SWDOWN at CIT (not shown). WRF-VPRM overestimated temperature by approximately 1.0°C, and the model underestimated RH by approximately 11%. The model overestimated wind speeds by approximately 0.3 m s−1. Wind directions ranged from 120° to 240°: the south and southwest directions were dominant during day- and nighttime, respectively, which are associated with the Catalina eddies off the coast and the topography of Southern California, as described in section 1. Besides temperature, RH, and winds, the agreement of SWDOWN is also critical for calculating NEE. SWDOWN is very closely correlated with photosynthesis active radiation (PAR); PAR ≈ 1.98 × SWDOWN (Mahadevan et al. 2008) and is used to compute GEE over vegetation areas. SWDOWN showed clear diurnal variations, starting from sunrise (0500 LST) and ending at sunset (1900 LST), with a midday peak around 1200 LST (not shown). Overall, the model overestimated SWDOWN by 85 W m−2 on average.
Statistics of meteorological variables at CIT and some FLUXNET sites, including oak/pine forest, coastal sage, and grassland sites. Temperature and RH at 2 m and winds at 10 m above ground level are compared.
The volume of the PBL basically determines the concentration of species such as CO2 and is fundamentally controlled by solar radiation, heating of the ground, and developing atmospheric turbulence within the PBL. The simulated PBLH started to increase around 0500 LST, reached a peak around 1200 LST, and then continually decreased until 2200–2300 LST. By comparing with the measured cloud-base heights, it was determined that the simulated PBLHs were over- and underestimated by approximately 56 and 128 m with IOAs of 0.68 and 0.55, respectively, during the day- and nighttime. Note that uncertainties may occur when directly comparing the simulated PBLHs with the cloud-base heights measured by the ceilometer, because 1) the model’s coarser resolution often cannot resolve clouds that appear on the subgrid scale and can be detected by the ceilometer and 2) the simulated PBLH is sensitive to the model’s PBL schemes, making it often hard to capture the measured PBLHs. To investigate the uncertainty of simulated PBLH, we estimated PBLH using the vertical profile of potential temperature measured by the NOAA P-3 aircraft during midday on 14, 16, and 19 May, when the aircraft flew spirals over CIT during daytime. The bias of modeled PBLH ranged approximately from 10 to 380 m (Fig. S2 in the online supplemental material). This bias unfortunately cannot be directly compared with the ground ceilometer measurements because of the fact that there were no available measured values at the same time as the aircraft measurements.
The modeling results were also compared with NOAA P-3 aircraft observations. Among 18 days in total of the NOAA P-3 flights during CalNex 2010, we chose the days of 14, 16, 19, and 21 May for daytime and 29 and 30 May and 2 and 3 June for nighttime, when the aircraft flew over SoCAB. Each day’s vertical profile of potential temperature and wind speed is displayed in Fig. 2. WRF-VPRM results showed good agreement with observed potential temperature along flight altitudes; they underestimated potential temperature by 0.1°C in daytime and overestimated by 1.4°C in nighttime. The model also captured the vertical wind profile, except at higher altitude (>3 km) on 19 May; it underestimated wind by 0.6 m s−1 (IOA = 0.89; RMSE = 3.3 m s−1) during daytime and 0.2 m s−1 (IOA = 0.68; RMSE = 2.6 m s−1) during nighttime.
Vertical profiles of (top) potential temperature and (bottom) wind speed for each of the eight flights during the study period. Red and black lines indicate WRF-VPRM results and NOAA P-3 aircraft observations, respectively.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
b. CO2 concentration
1) Background CO2 concentrations
To investigate the performance of the model’s background CO2 concentration, we used observational data measured at the Palos Verdes site (PV; 33°42′88″N, 118°18′70″W) in Fig. 1. This site is located at the southern end of SoCAB, on a steep hillside near the Pacific Ocean ~0.3 km above sea level, at which the concentrations of tracers transported by sea breezes to SoCAB can be monitored. Newman et al. (2013) measured CO2 concentration and assumed the daily minimum hourly values at PV to be a constant background concentration of 393.1 ppm. We also assumed that the measured concentrations at this site can represent the “local” background concentration for SoCAB. Other CO2 monitoring sites were not known to us during the study period. Our simulated result was close to the measured background concentration value with the offset of 0.1 ± 2.4 ppm.
2) Spatial distributions
The horizontal and vertical distributions of CO2 concentration are associated with the diurnal variation of the sea breeze, emissions, and PBLHs along with the topographical features of SoCAB—21 May was selected as an example of a typical day. Its diurnal pattern of CO2 concentrations on horizontal and vertical cross-sectional distributions is displayed in Figs. 3 and 4, respectively. At 0600 LST in the early morning, the highest CO2 concentration appeared to be due mainly to the combination of the smallest volume of PBL and the beginning of morning rush-hour vehicle CO2 emissions. The heights of surrounding mountains are much higher than the average PBLH most of the time, preventing ventilation. Around 1200–1400 LST when the PBLH grew and the vegetation uptake increased, the sea breeze pushed the tracers over the mountains, and the concentration remained low until late afternoon (~1800 LST). After this time, CO2 concentration started to increase again as a result of the evening rush-hour emissions along with the gradually decreasing PBLH and the increasing vegetation respiratory contribution, and high concentrations were maintained through the night until early the next morning. This typical diurnal pattern of tracers for SoCAB agrees well with the results of other previous pollution studies (e.g., Chen et al. 2013).
Horizontal distribution of CO2 concentrations (ppm) of the finest-scale domain (d03) on 21 May, displayed in 6-h intervals.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
Vertical cross section of CO2 concentrations along the line indicated in Fig. 1 for 21 May, displayed in 6-h intervals.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
3) Comparison with CIT ground observations
The time series of the hourly variation of CO2 concentration at CIT during the whole study period is shown in Fig. 5. WRF-VPRM captured the variation of CO2 concentrations. The model overestimated by 9.2 ppm with IOA = 0.53 and RMSE = 17.0 ppm during daytime, and it overestimated by 1.6 ppm with IOA = 0.49 and RMSE = 1.8 ppm during nighttime.
Time series of CO2 concentrations from simulations (red line) and ground observations (black dots) at CIT.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
To facilitate discussion of the variation of CO2 concentrations together with emissions, the averaged diurnal variation of CO2 emissions and concentrations at CIT and over SoCAB is displayed in Fig. 6, together with that of PBLHs. Over SoCAB, the anthropogenic CO2 emissions increased during the morning rush hour (from 0500 to 0900 LST) and then increased again after 1300 LST until 1700 LST, showing a daily maximum peak. The larger second rush-hour peak was a typical emission pattern in metropolitan areas, such as in Houston, Texas (e.g., Park et al. 2010). The daytime emissions at CIT were about 2 times those averaged over SoCAB.
Averaged diurnal variation of anthropogenic CO2 emissions, concentrations, and PBLH at CIT and over SoCAB during the study period. Gray bars indicate 1 std dev for the observations.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
The averaged CO2 concentrations at CIT were approximately 9 ppm higher than those over SoCAB, as expected. The CO2 concentrations at CIT began to increase at 0500 LST and continued to rise until 0800 LST when the dominant morning rush hour ended, and values remained high until noon. In the afternoon, the concentrations decreased until 1600–1700 LST, associated with the highest PBLH and biospheric contribution, and then gradually increased again through the night until early the next morning as the PBLH decreased and the plant respiratory contribution increased.
During the early daytime (0600–1300 LST), WRF-VPRM overestimated the CO2 concentrations by ~14 ppm on average (Fig. 6). The uncertainties of the model can be due to vertical mixing, advection, emissions, initial fields, and VPRM parameters, as described in previous studies (Ahmadov et al. 2007; Pillai et al. 2011). The overestimate of daytime CO2 concentrations may be caused by overestimates either of the first peak of emissions at CIT or of the advection from CO2 source areas to CIT. To examine daytime emission sources affecting CIT by advection from specific wind directions, frequency of counts by wind direction of CO2 concentration was compiled. The results are represented in Fig. 7. This exercise revealed that south and southwest, where downtown LA is located, were dominant daytime CO2 source directions. The averaged simulated daytime CO2 concentration in the dominant wind directions (150°–240°) was 421.8 ppm, which is comparable to the observed mean values of 411.9 ± 8.3 ppm. The Hestia emissions data used in this study are considered to be the “climatology” of emissions rather than the “weather” of emissions. This means that the emissions data would not represent “true” day-to-day hourly-basis variations in real environments. From the relatively low bias of the meteorological conditions, the uncertainties of emissions data in or around downtown LA may cause the overestimate of the simulated CO2 concentrations during daytime. We still need comparisons at multiple measurement points for further investigation, however.
Frequency of counts by wind direction (%) of simulated CO2 concentration at CIT.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
To extract anthropogenic and biospheric signals from the total CO2 concentrations, two isotopic tracer radiocarbon (∆14C) measurements were conducted at CIT during the study period by Newman et al. (2013), resulting in a finding that ~100% and ~50% of emissions were from fossil-fuel combustion during day- and nighttime, respectively. We compared the measured signals with “tagged” tracers built into WRF-VPRM, which can identify the contribution of different CO2 sources between anthropogenic and biospheric signals (Ahmadov et al. 2009; Pillai et al. 2011). The simulated results showed positive values most of the time with a range of <2 ppm (Fig. S3 in the online supplemental material), which implied that our model successfully showed that the influence of fossil-fuel combustion was significant near CIT. Our model failed to capture the negative trend during the first half of the study period and underestimated the positive values by a factor of 3–5 during the second half period. This discrepancy may not be decreased simply by making model resolutions higher, as shown in the sensitivity tests in Feng et al. (2016). It seems that further investigation for both intensive measurements and improving models is needed in future studies.
4) Comparison with aircraft observations
Here, we discuss the comparison of the model results with aircraft observations, which was missing in the previous study of Feng et al. (2016). The observed and simulated CO2 vertical profiles and time series for all eight flights are displayed in Figs. 8 and 9, respectively, and each flight’s statistics are reported in Table 3. The model captured the vertical gradient of CO2 concentrations, except at lower altitudes (<1 km) on 19 May and 3 June, on which days it significantly overestimated and underestimated the concentrations, respectively (Fig. 8). The relatively large discrepancies on the two noted days were not directly caused by the model’s meteorological-simulation performance (Fig. 4). No clear correlation with wind directions was found (not shown).
As in Fig. 2, but for CO2 concentrations.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
Time series of CO2 concentrations from simulations (red line) and aircraft observations (black dots), along with time series of aircraft altitudes (black dashed line) and simulated PBLHs (blue thick line). Land-cover types at the aircraft locations are represented by color codes as in Fig. 1.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
Statistics for CO2 concentrations between modeling results and aircraft observations.
The aircraft moved in and out of the PBL, flying over various land-cover types (Fig. 9). Overall, lower concentrations appeared over the ocean and above the PBL and higher concentrations occurred over the urban area and within the PBL. The model underestimated concentration by 1.8 ppm, with IOA = 0.81 and RMSE = 6.1 ppm, and by 0.3 ppm, with IOA = 0.74 and RMSE = 6.7 ppm, during day- and nighttime flight days, respectively. During daytime, CO2 concentrations within the PBL were ~7 ppm higher than those above the PBL, on average, without regard to land cover. The model’s results showed a better agreement with observations above the PBLH, where there are fewer direct effects of heterogeneous emission sources. During nighttime, it appears that the aircraft moved above the nighttime PBL most of the time. High concentrations (>400 ppm) appeared at lower altitudes (<1 km) even above the nighttime PBLHs in both observations and simulations. This phenomenon could be due either to CO2 concentrations in residual layers or to advection at high altitudes. For further investigations of the uncertainties at higher altitudes, inverse modeling as in the work of Brioude et al. (2013) can be used in future studies.
c. CO2 fluxes
On the basis of the vegetation fraction retrieved by SYNMAP, the total vegetation area over SoCAB was estimated to be approximately 77%, mostly in the mountains, of which shrubland was the most dominant vegetation class (~62%) followed by evergreen forest (~33%), grassland (~4%), and savanna (~1%). To see by how much the simulated NEE was improved by optimization of VPRM parameters at each corresponding vegetation class site, we compared the averaged diurnal variations of simulated NEE and observed CO2 flux, assuming that the simulated NEE is equal to the measured CO2 flux. The results are displayed in Fig. 10. In comparing with the results from the prior VPRM parameters, it is seen that the statistics with posterior parameters were much improved.
Averaged diurnal variation and scatterplots of simulated NEEs and observed CO2 fluxes at three FLUXNET sites, including oak/pine forest, coastal sage, and grassland. NEE and “NEE-default” are the simulated result with posterior and prior VPRM parameters, respectively. The black dashed line indicates a 1-to-1 line, and the red line indicates the linear-regression line for daytime only (black dots). Nighttime data (gray circles) are excluded from statistics.
Citation: Journal of Applied Meteorology and Climatology 57, 6; 10.1175/JAMC-D-17-0358.1
Errors of simulated NEEs are likely attributed to the calculation of respiration by 2-m air temperature, as described by Ahmadov et al. (2009) and Pillai et al. (2011), and to the optimization approaches (Jamroensan 2013). In comparing with the results from prior VPRM parameters, it is seen that the model with the posterior parameters reduced the bias of simulated NEE by 40%, 48%, and 34% during daytime and 94%, 41%, and 26% during nighttime, at the oak/pine forest, coastal sage, and grassland sites, respectively. Note that the direct comparison of NEEs with fluxes of CO2 measured by the eddy covariance method can naturally include uncertainties, because the fluxes include a CO2 storage term within the canopy (Aubinet et al. 2012). In this study, however, we kept our assumption because of the facts that there is lack of storage-term information in FLUXNET data and that estimates of the storage term are outside the bounds of this study. Overall, the posterior VPRM parameters played an important role in improving the simulation results of NEE, and therefore we conclude that the optimization of VPRM parameters is essential for regional CO2 modeling. Thus, the optimization can improve the estimate of the CO2 budget over SoCAB (section 4d), in which multiple vegetation types are mixed.
d. Estimate of CO2 budget
An estimate of the CO2 budget is important for corrections to or updates of current environmental policy in relation to reductions in fossil-fuel carbon emissions that are mandated by California state law (Assembly Bill 32). In this section, we estimated the CO2 budget over SoCAB during the CalNex 2010 period. Here we define the CO2 budget over the whole area of SoCAB (Fig. 1) as the ratio of the total simulated NEEs to the total Hestia 2010 anthropogenic CO2 emissions at all surface grid points. The amount of biospheric contribution was estimated at approximately −23% (daytime) and approximately +9% (nighttime) of the total anthropogenic CO2 emissions during the study period. This vegetation contribution rate is higher than the value calculated with the prior VPRM parameters by approximately a factor of 2. Note that the study period was part of the vegetation growing season, and therefore a further long-term (~1 yr) simulation study, taking into account the seasonal leaf foliage, should be followed to assess the annual CO2 budget.
During the process of optimization, another uncertainty from the VPRM parameters could emerge from a cutoff criterion of the u* threshold value for the neighborhood-scale flux measurements. The u* threshold value varies from site to site and from season to season, and it is critical to filter underestimated measured CO2 fluxes, especially during transition and nighttime. To evaluate the effects of the threshold value, we carried out sensitivity tests with additional cutoff criteria of u* < 0.2 and u* < 0.0 m s−1. These two values are widely used ranges in the micrometeorology literature. Results showed that the biospheric contribution relative to the anthropogenic emissions over SoCAB during the CalNex 2010 period is estimated to be in the range from −24% to −20% during daytime and from +8% to +9% during nighttime, in consideration of the uncertainty from cutoff criteria. On the basis of the land-cover distribution, it appears that the biospheric CO2 uptake mainly occurred in the vegetation-dominated mountain regions in SoCAB.
5. Summary and conclusions
In this study, a coupled WRF-VPRM model was applied over the anthropogenic-CO2-emission-rich region of SoCAB during the CalNex 2010 period, in contrast to most previous studies, which were mainly focused on vegetation-dominated areas. Discriminating from a similar previous study of Feng et al. (2016), here 1) VPRM parameters were optimized for better performance of the biospheric module and 2) the newly updated version of Hestia-LA combined with the FFDAS anthropogenic CO2 emission data and the CT2015 CO2 mole-fraction data was used. We reported the performance of WRF-VPRM for CO2 transport and temporal variability in comparison with the observations and also discussed the model’s improvement from the VPRM parameter optimization. Last, we presented the estimated CO2 budget over SoCAB.
The WRF-VPRM successfully recreated the meteorological variables for both ground- and aircraft-based measurements. The model also captured the diurnal variation of CO2 concentrations at the ground sites but slightly overestimated daytime CO2 concentrations. The analysis of daytime CO2 concentrations by wind direction implied that the uncertainty of local emission sources located in the south and southwest directions, where downtown LA is located, affected the model overestimation of CO2 concentrations at the ground measurement site. The model also matched the vertical profile and times series of CO2 concentrations, in comparison with the NOAA P-3 aircraft measurements, but the modeled CO2 concentrations were over- and underestimated at lower altitudes on 19 May and 3 June, respectively.
After validating the improvement of NEE calculation with posterior VPRM parameters, the total biospheric contribution rate was calculated over SoCAB, resulting in a range from −24% to −20% (in daytime) and from +8% to +9% (in nighttime) of the total anthropogenic CO2 emissions. The uncertainty of the CO2 budget was reduced by approximately a factor of 2 relative to the results from the prior VPRM parameters. Overall, SoCAB played a role as a net emission source of CO2 during the study period.
The output of this modeling study can be used to validate remote sensing instruments and as the priori for regional inverse modeling, as well as to support environmental policy in relation not only to emission control of fossil-fuel carbon but also to vegetation management in the state of California. As the first goal of the study, the slant column concentration of CO2, the total number of absorbing molecules per unit area along the sun–Earth–instrument optical path, over LA can also be compared with and used to calibrate the California Laboratory for Atmospheric Remote Sensing facility currently set up on Mount Wilson for the continuous monitoring of air pollution and greenhouse gases in SoCAB (Wong et al. 2015). The study’s total column concentrations from WRF-VPRM results can be compared with data from the Orbiting Carbon Observatory-2 (OCO-2; Eldering et al. 2017) or the Japanese Greenhouse Gas Observing Satellite (GOSAT), as in the work of Hedelius et al. (2017).
Acknowledgments
Author CP acknowledges financial support by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2017R1D1A1B03034021). CarbonTracker CT2015 results were provided by NOAA/ESRL and are available online (http://carbontracker.noaa.gov).
REFERENCES
Ahmadov, R., C. Gerbig, R. Kretschmer, S. Koerner, B. Neininger, A. J. Dolman, and C. Sarrat, 2007: Mesoscale covariance of transport and CO2 fluxes: Evidence from observations and simulations using the WRF-VPRM coupled atmosphere–biosphere model. J. Geophys. Res., 112, D22107, https://doi.org/10.1029/2007JD008552.
Ahmadov, R., C. Gerbig, R. Kretschmer, S. Korner, C. Rodenbeck, P. Bousquet, and M. Ramonet, 2009: Comparing high resolution WRF-VPRM simulations and two global CO2 transport models with coastal tower measurements of CO2. Biogeosciences, 6, 807–817, https://doi.org/10.5194/bg-6-807-2009.
Angevine, W. M., L. Eddington, K. Durkee, C. Fairall, L. Bianco, and J. Brioude, 2012: Meteorological model evaluation for CalNex 2010. Mon. Wea. Rev., 140, 3885–3906, https://doi.org/10.1175/MWR-D-12-00042.1.
Asefi-Najafabady, S., and Coauthors, 2014: A multiyear, global gridded fossil fuel CO2 emission data product: Evaluation and analysis of results. J. Geophys. Res. Atmos., 119, 10 213–10 231, https://doi.org/10.1002/2013JD021296.
Aubinet, M., C. Feigenwinter, B. Heinesch, Q. Laffineur, D. Papale, M. Reichstein, J. Rinne, and E. Van Gorsel, 2012: Nighttime flux correlation. Eddy Covariance: A Practical Guide to Measurement and Data Analysis, M. Aubinet, T. Vesala, and D. Papale, Eds., Springer, 133–157.
Auch, R., J. Taylor, and W. Acevedo, 2004: Urban growth in American cities: Glimpses of U.S. urbanization. U.S. Geological Survey Circular 1252, 52 pp., https://pubs.er.usgs.gov/publication/cir1252.
Baldocchi, D., and Coauthors, 2001: FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull. Amer. Meteor. Soc., 82, 2415–2434, https://doi.org/10.1175/1520-0477(2001)082<2415:FANTTS>2.3.CO;2.
Belward, A. S., J. E. Estes, and K. D. Kline, 1999: The IGBP-DIS global 1-km land-cover data set DISCover: A project overview. Photogramm. Eng. Remote Sensing, 65, 1013–1020.
Bernardet, L., and Coauthors, 2008: The Developmental Testbed Center and its winter forecasting experiment. Bull. Amer. Meteor. Soc., 89, 611–628, https://doi.org/10.1175/BAMS-89-5-611.
Bowden, R. D., K. J. Nadelhoffer, R. D. Boone, J. M. Melillo, and J. B. Garrison, 1993: Contributions of aboveground litter, belowground litter, and root respiration to total soil respiration in a temperate mixed hardwood forest. Can. J. For. Res., 23, 1402–1407, https://doi.org/10.1139/x93-177.
Brioude, J., and Coauthors, 2013: Top-down estimate of surface flux in the Los Angeles Basin using a mesoscale inverse modeling technique: Assessing anthropogenic emissions of CO, NOx and CO2 and their impacts. Atmos. Chem. Phys., 13, 3661–3677, https://doi.org/10.5194/acp-13-3661-2013.
Chen, D., and Coauthors, 2013: WRF-Chem simulation of NOx and O3 in the L.A. basin during CalNex-2010. Atmos. Environ., 81, 421–432, https://doi.org/10.1016/j.atmosenv.2013.08.064.
Conil, S., and A. Hall, 2006: Local regimes of atmospheric variability: A case study of Southern California. J. Climate, 19, 4308–4325, https://doi.org/10.1175/JCLI3837.1.
Eldering, A., and Coauthors, 2017: The Orbiting Carbon Observatory-2: First 18 months of science data products. Atmos. Meas. Tech., 10, 549–563, https://doi.org/10.5194/amt-10-549-2017.
Feng, S., and Coauthors, 2016: Los Angeles megacity: A high-resolution land–atmosphere modelling system for urban CO2 emissions. Atmos. Chem. Phys., 16, 9019–9045, https://doi.org/10.5194/acp-16-9019-2016.
Forster, P., and Coauthors, 2007: Changes in atmospheric constituents and in radiative forcing. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 129–234.
Gerbig, C., J. C. Lin, S. C. Wofsy, B. C. Daube, A. E. Andrews, B. B. Stephens, P. S. Bakwin, and C. A. Grainger, 2003: Toward constraining regional-scale fluxes of CO2 with atmospheric observations over a continent: 2. Analysis of COBRA data using a receptor-oriented framework. J. Geophys. Res., 108, 4757, https://doi.org/10.1029/2003JD003770.
Gurney, K. R., D. L. Mendoza, Y. Y. Zhou, M. L. Fischer, C. C. Miller, S. Geethakumar, and S. de la Rue du Can, 2009: High resolution fossil fuel combustion CO2 emission fluxes for the United States. Environ. Sci. Technol., 43, 5535–5541, https://doi.org/10.1021/es900806c.
Gurney, K. R., I. Razlivanov, Y. Song, Y. 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, https://doi.org/10.1021/es3011282.
Haman, C. L., B. Lefer, and G. A. Morris, 2012: Seasonal variability in the diurnal evolution of the boundary layer in a near-coastal urban environment. J. Atmos. Oceanic Technol., 29, 697–710, https://doi.org/10.1175/JTECH-D-11-00114.1.
Hedelius, J. K., and Coauthors, 2017: Emissions and topographic effects on column CO2 () variations, with a focus on the Southern California megacity. J. Geophys. Res. Atmos., 122, 7200–7215, https://doi.org/10.1002/2017JD026455.
Hong, S. Y., Y. Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341, https://doi.org/10.1175/MWR3199.1.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.
Jamroensan, A., 2013: Improving bottom-up and top-down estimates of carbon fluxes in the midwestern USA. Ph.D. thesis, University of Iowa, 121 pp., http://ir.uiowa.edu/etd/2530/.
Jimenez, P. A., and J. Dudhia, 2012: Improving the representation of resolved and unresolved topographic effects on surface wind in the WRF Model. J. Appl. Meteor. Climatol., 51, 300–316, https://doi.org/10.1175/JAMC-D-11-084.1.
Jung, M., K. Henkel, M. Herold, and G. Churkina, 2006: Exploiting synergies of global land cover products for carbon cycle modeling. Remote Sens. Environ., 101, 534–553, https://doi.org/10.1016/j.rse.2006.01.020.
Kotthaus, S., and C. S. B. Grimmond, 2012: Identification of micro-scale anthropogenic CO2, heat and moisture sources—Processing eddy covariance fluxes for a dense urban environment. Atmos. Environ., 57, 301–316, https://doi.org/10.1016/j.atmosenv.2012.04.024.
Le Quere, C., and Coauthors, 2014: Global carbon budget 2013. Earth Syst. Sci. Data, 6, 235–263, https://doi.org/10.5194/essd-6-235-2014.
Mahadevan, P., and Coauthors, 2008: A satellite-based biosphere parameterization for net ecosystem CO2 exchange: Vegetation Photosynthesis and Respiration Model (VPRM). Global Biogeochem. Cycles, 22, GB2005, https://doi.org/10.1029/2006GB002735.
Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343–360, https://doi.org/10.1175/BAMS-87-3-343.
Newman, S., and Coauthors, 2013: Diurnal tracking of anthropogenic CO2 emissions in the Los Angeles basin megacity during spring 2010. Atmos. Chem. Phys., 13, 4359–4372, https://doi.org/10.5194/acp-13-4359-2013.
National Research Council, 2010: Advancing the Science of Climate Change. The National Academies Press, 526 pp.
Park, C., and G. W. Schade, 2016: Anthropogenic and biogenic features of long-term measured CO2 flux in north downtown Houston, Texas. J. Environ. Qual., 45, 253–265, https://doi.org/10.2134/jeq2015.02.0115.
Park, C., G. W. Schade, and I. Boedeker, 2010: Flux measurements of volatile organic compounds by the relaxed eddy accumulation method combined with a GC-FID system in urban Houston, Texas. Atmos. Environ., 44, 2605–2614, https://doi.org/10.1016/j.atmosenv.2010.04.016.
Peischl, J., and Coauthors, 2012: Airborne observations of methane emissions from rice cultivation in the Sacramento Valley of California. J. Geophys. Res., 117, D00V25, https://doi.org/10.1029/2012JD017994.
Pérez-Landa, G., and Coauthors, 2007: Mesoscale circulations over complex terrain in the Valencia coastal region, Spain—Part 2: Modeling CO2 transport using idealized surface fluxes. Atmos. Chem. Phys., 7, 1851–1868, https://doi.org/10.5194/acp-7-1851-2007.
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, https://doi.org/10.1073/pnas.0708986104.
Petrescu, A. M. R., R. Abad-Vinas, G. Janssens-Maenhout, V. N. B. Blujdea, and G. Grassi, 2012: Global estimates of carbon stock changes in living forest biomass: EDGARv4.3—Time series from 1990 to 2010. Biogeosciences, 9, 3437–3447, https://doi.org/10.5194/bg-9-3437-2012.
Peylin, P., and Coauthors, 2005: Daily CO2 flux estimates over Europe from continuous atmospheric measurements: 1, inverse methodology. Atmos. Chem. Phys., 5, 3173–3186, https://doi.org/10.5194/acp-5-3173-2005.
Pillai, D., and Coauthors, 2011: High-resolution simulations of atmospheric CO2 over complex terrain—Representing the Ochsenkopf mountain tall tower. Atmos. Chem. Phys., 11, 7445–7464, https://doi.org/10.5194/acp-11-7445-2011.
Ryan, M. G., and B. E. Law, 2005: Interpreting, measuring, and modeling soil respiration. Biogeochemistry, 73, 3–27, https://doi.org/10.1007/s10533-004-5167-7.
Ryerson, T. B., and Coauthors, 2013: The 2010 California Research at the Nexus of Air Quality and Climate Change (CalNex) field study. J. Geophys. Res., 118, 5830–5866, https://doi.org/10.1002/jgrd.50331.
Satterthwaite, D., 2008: Cities’ contribution to global warming: Notes on the allocation of greenhouse gas emissions. Environ. Urbanization, 20, 539–549, https://doi.org/10.1177/0956247808096127.
Seto, K. C., and Coauthors, 2014: Human settlements, infrastructure and spatial planning. Climate Change 2014: Mitigation of Climate Change, O. Edenhofer et al., Eds., Cambridge University Press, 923–1000.
Tans, P. P., I. Y. Fung, and T. Takahashi, 1990: Observational constraints on the global atmospheric CO2 budget. Science, 247, 1431–1438, https://doi.org/10.1126/science.247.4949.1431.
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, https://doi.org/10.5194/acp-11-705-2011.
Ulrickson, B. L., and C. F. Mass, 1990: Numerical investigation of mesoscale circulations over the Los Angeles Basin. Part I: A verification study. Mon. Wea. Rev., 118, 2138–2161, https://doi.org/10.1175/1520-0493(1990)118<2138:NIOMCO>2.0.CO;2.
van der Molen, M. K., and A. J. Dolman, 2007: Regional carbon fluxes and the effect of topography on the variability of atmospheric CO2. J. Geophys. Res., 112, D01104, https://doi.org/10.1029/2006JD007649.
Velasco, E., M. Roth, S. H. Tan, M. Quak, S. D. A. Nabarro, and L. Norford, 2013: The role of vegetation in the CO2 flux from a tropical urban neighbourhood. Atmos. Chem. Phys., 13, 10 185–10 202, https://doi.org/10.5194/acp-13-10185-2013.
Willmott, C. J., 1982: Some comments on the evaluation of model performance. Bull. Amer. Meteor. Soc., 63, 1309–1313, https://doi.org/10.1175/1520-0477(1982)063<1309:SCOTEO>2.0.CO;2.
Wong, K. W., and Coauthors, 2015: Mapping CH4 : CO2 ratios in Los Angeles with CLARS-FTS from Mount Wilson, California. Atmos. Chem. Phys., 15, 241–252, https://doi.org/10.5194/acp-15-241-2015.
