Influence of fair-weather cumulus clouds on isoprene chemistry

Authors


Abstract

[1] Fair-weather cumulus clouds are not resolved in regional- and global-scale atmospheric chemistry models because their horizontal extent is less than the horizontal resolution of the model. A Large-Eddy Simulation (LES) model, with finer grid resolution, can resolve the energy containing turbulent eddies and fair-weather cumulus clouds. Isoprene, which is mainly emitted from deciduous forests and plays a significant role in producing ozone, has a chemical lifetime similar to the boundary layer turbulence turnover time, indicating that turbulent transport, cloud processes, and chemistry are all potentially important for the prediction of ambient isoprene concentrations. The LES model coupled with chemistry developed in this study is an ideal tool to examine the influence of fair-weather cumulus clouds on isoprene chemistry. With a LES model that includes a moderately complex gas-phase chemical mechanism of isoprene oxidation, we find enhancement of isoprene, methacrolein, and methylvinyl ketone in the cloud layer while changes in these chemical species' mixing ratios in the subcloud layer relative to the cloud-free case vary depending on the chemical lifetimes. We demonstrate that nitrogen oxides put into the system can modulate the chemical lifetimes of isoprene and related chemical species, which in turn changes the vertical distribution of the chemical species. For high NOx conditions, ozone in the subcloud layer for the cloudy case is ∼5 ppbv lower than that for the cloud-free case, suggesting potential positive ozone bias in large-scale models that do not include fair-weather cumulus cloud processes.

1. Introduction

[2] Synoptic high pressure regions, which are conducive to the formation and accumulation of photochemically generated pollutants in the convective boundary layer (CBL), often produce fair weather cumulus clouds, also known as shallow cumulus and cumulus humilis. Globally, daytime cumulus clouds over land occur ∼25% of the reported days during June–August, and have an average cloud amount of ∼33% [Hahn and Warren, 1999] (also S. G. Warren and C. J. Hahn, Climatic Atlas of Clouds Over Land and Ocean, 2007, available at http://www.atmos.washington.edu/CloudMap/; hereinafter Warren and Hahn, online publication, 2007). Shallow cumulus clouds can potentially affect the distribution of chemical constituents in the boundary layer. The chemical species' vertical transport can be enhanced depending on the buoyancy associated with the clouds and the strength of the temperature inversion at the top of the CBL. Scattering of radiation by the clouds can influence chemical constituents by modifying photodissociation rates, while cloud shading will reduce emissions of biogenic volatile organic compounds (VOC). The cloud and raindrops also provide a medium for aqueous chemistry to occur. Because shallow cumulus clouds often occur over land, it is important to understand their role on the CBL chemistry. In this paper, we focus on the rural setting where isoprene is a dominant precursor to ozone production.

[3] Emissions of biogenic VOCs such as isoprene and monoterpenes are known to comprise 80% of the total global VOCs emissions [Olivier and Berdowski, 2001]. They are of great importance in atmospheric chemistry by serving as precursors for O3 in the presence of NOx [Trainer et al., 1987b; Chameides et al., 1988] and by controlling the oxidative capacity of the atmosphere in pristine conditions such as the tropical forests [Jacob and Wofsy, 1988]. Because the chemical lifetime of isoprene (20–40 min) with respect to OH is comparable to the turbulence turnover time in the convective boundary layer (boundary layer height/convective velocity = ∼20 min), both turbulent transport and chemistry are important for determining the distribution of isoprene. Thus, in order to examine the influence of fair-weather cumulus clouds on isoprene with a numerical model, it is necessary to have a model that represents turbulent motion, cloud processes, and chemistry accurately in a coupled way.

[4] Because shallow cumulus cloud processes are not fully resolved in global or regional atmospheric chemistry models, it is important to quantify the influence of these clouds (via enhanced vertical transport, modified photolysis rates, modified emission rates, and cloud chemistry) on CBL chemistry. There have been only a few studies that have examined the role of fair weather clouds on the transport and chemistry of chemical constituents [Greenhut, 1986; Ching and Alkezweeny, 1986; Ching et al., 1988; Thompson et al., 1994; Angevine, 2005; Vila-Guerau de Arellano et al., 2005, hereinafter VKBP05; Karl et al., 2007]. One-dimensional models (or single column models) of the cloud topped boundary layer have been useful to examine the influence of shallow cumulus clouds on pollutant transport [e.g., Angevine, 2005]. However, there are deficiencies, such as unrealistic thermodynamic profiles and values of cloud cover and cloud liquid water that are too high, in these single column models [Lenderink et al., 2004]. Because turbulence in the CBL is three-dimensional in nature, it is best to represent the CBL dynamics with a 3-dimensional model. Large-eddy simulation (LES) provides an ideal tool to investigate chemistry coupled with turbulence and cloud processes because a LES model explicitly resolves the large scale energy-containing turbulent eddies and clouds with 1–100 m model grid size. The LES technique has been used to simulate the cloud-free and the cloudy CBL chemistry and transport of chemical constituents [Schumann, 1989; Sykes et al., 1992, 1994; Krol et al., 2000; VKBP05]. However, these previous LES studies with chemical reactions adopt a chemical mechanism that is too simple to represent reasonably the isoprene chemistry.

[5] In this study, we incorporate into the NCAR LES model [Moeng, 1984] a moderately complex chemistry mechanism that includes isoprene oxidation similar to what is used in a regional or a global atmospheric chemistry model [Hess et al., 2000; Horowitz et al., 2003]. We focus on gas-phase chemistry, and do not include gas-to-liquid partitioning and aqueous phase chemistry in the simulations. The potential impact of these processes on the isoprene chemistry will be discussed. With this LES model, we demonstrate the role of buoyancy-driven cloud transport, cloud-modified photolysis rates, and isoprene emission rate on the chemical species distributions with a focus on isoprene and its oxidation products. Two NOx regimes are considered: low and high NOx conditions, representing the influence of rural and more urban-like NOx emissions, respectively. A main objective of this manuscript is to provide the implications of representing CBL chemistry in the fair-weather cumulus regime for regional and global chemical transport models that must treat shallow cumulus with a subgrid-scale parameterization. The current study sets the basis for more detailed investigations on the interactions between turbulence and chemistry or on comparing the capabilities of boundary layer schemes in representing CBL chemistry with shallow cumulus clouds.

[6] After describing the LES model and experimental design (Sections 2 and 3), LES results of the meteorology (Section 4) are presented to provide context for the chemistry results. Next, the reactive chemistry is examined under low NOx conditions (Section 5.1) showing the variability of the chemical composition in clear and cloudy conditions and for when the clouds modify photodissociation reactions and reduce biogenic VOC emissions via cloud shading. Reactive chemistry under high NOx conditions (Section 5.2) illustrates the influence of a change in photochemical lifetimes on the vertical profiles. In this paper, we focus on average mixing ratios from the LES results. Butler et al. [2008] suggested the potentially important role of turbulence in modifying isoprene-OH chemical reactivity in their OH budget calculations, which can be examined with a LES model. To limit the length of the manuscript, the turbulence statistics, such as variances, fluxes, and co-variances and turbulent fluctuation fields, of the chemical constituents will be discussed in a second manuscript.

2. Model Description

[7] The large eddy simulation used for this study is derived from Moeng [1984] and Moeng and Wyngaard [1984] with modifications described in Sullivan et al. [1996, 1994]. The code has been modified to run on massively parallel machines using the Message Passing Interface (MPI) as described by Patton et al. [2005]. A summary of the model formulation is given here.

[8] The LES uses a sharp wave-cutoff filter to define the resolvable-scale variables and parameterizes the subgrid-scale effects with a turbulence energy model [Deardorff, 1980]. The model solves the Navier-Stokes equations for the three resolved-scale velocity components, the conservation equation for potential temperature, the Poisson equation for the pressure field, and the subgrid-scale kinetic energy equation. Horizontal advection is calculated using the pseudospectral method [Fox and Orzag, 1973] and vertical advection using a second-order centered finite difference scheme. To transport temperature and all other scalars vertically, a monotone scheme is employed [Koren, 1993]. The fields are advanced in time using a third-order Runge-Kutta scheme [Spalart et al., 1991]. The horizontal boundary condition is periodic, while a radiation boundary condition is used for the upper boundary [Klemp and Durran, 1983]. At the surface, no-slip conditions are imposed and the stress is determined from Monin-Obukhov similarity theory.

[9] The chemistry mechanism used in the NCAR HANK model [Hess et al., 2000] is updated following the NCAR global chemical transport model MOZART2.2 [Horowitz et al., 2003]. The mechanism contains 142 gas-phase reactions for which 52 species are predicted (see Appendix A, Tables A1 and A2). The gas chemistry represents daytime chemistry of methane (1700 ppbv), carbon monoxide (200 ppbv), ozone, odd nitrogen species, peroxides, aldehydes, non-methane hydrocarbons (ethane, ethene, propylene, isoprene, monoterpene), methyl vinyl ketone (MVK), methacrolein (MACR), and peroxy radicals. For the study presented here, we do not include gas-to-liquid partitioning and aqueous phase chemistry.

[10] Two passive tracers, similar to those described by Moeng and Wyngaard [1984], are also represented in the LES. The first tracer corresponds to a bottom-up tracer that has higher values in the boundary layer than those in the free atmosphere and is emitted only at the surface. The second tracer represents a top-down tracer whose higher values above the boundary layer are entrained into the boundary layer.

3. Experimental Design

3.1. Configuration of Model

[11] The LES code is configured for the conditions observed at the Southern Great Plains site of the Atmospheric Radiation Measurement program on 21 June 1997 [Brown et al., 2002] and is configured in the same manner as Brown et al. [2002] set forth in their intercomparison. This case is also similar to what VKBP05 used. A time-varying surface forcing of both sensible and latent heating is prescribed. The geostrophic wind (Ug, Vg) is set to (10, 0) m s−1 and the initial horizontal wind (u, v) is set similarly (10, 0) m s−1. The initial profiles of potential temperature and water vapor are the same as those of the standard intercomparison case in Brown et al. [2002]. The initial profile of potential temperature and consequently simulated clouds in this study are slightly different from those in VKBP05 who modified the potential temperature profile to produce cloud more quickly.

[12] While the meteorology configuration follows an observed situation, the chemical environment is prescribed for a typical southeastern U.S. environment so that basic, important parameters or processes can be examined for the isoprene chemistry. Climatologically, daytime cumulus clouds over the southeastern U.S. land area occur on a third to a half of the reported days during summer, and have an average cloud amount of ∼30% [Hahn and Warren, 1999; Warren and Hahn, online publication, 2007]. Thus, fair weather cumulus clouds can potentially affect vertical distributions of chemical species in this region. Chemical species with short chemical lifetimes are initialized with a photochemical box model, which is integrated from midnight to 0830 LT (local time). The initial mixing ratios within and above the CBL for many of the species are listed in Table A1 in Appendix A and cyclic lateral chemical boundary conditions are adopted. The horizontally homogeneous surface emissions of isoprene (maximum 5.04 mg m−2 hr−1) and monoterpene (maximum 0.504 mg m−2 hr−1) follow a diurnal profile (Figure 1). Because the simulations are idealized, NO emissions are given a constant value to maintain approximately constant NOx mixing ratios during the simulation. Dry deposition is calculated for all species as the product of the concentration of the species at the lowest grid cell and its deposition velocity. The deposition velocities (Table A1 in Appendix A) are appropriate for summertime, deciduous forest conditions. The 0.05 m s−1 deposition velocity is set as the aerodynamic velocity. The individual chemical species deposition velocities also include the surface resistance, which is based mostly from the values given in Wesely [1989]. Photolysis rates have diurnal variations and are appropriate for 36°N and summer solstice conditions.

Figure 1.

Diurnal variation of isoprene (solid line) and monoterpene (dashed line) fluxes used in this study.

[13] The dynamics, cloud physics, and chemistry are integrated over a model domain of 6.4 km × 6.4 km × 4.4 km (96 × 96 × 96 grid points) with periodic boundary conditions in x and y. Dynamics and chemistry are integrated together with time step Δt = 1.5 s. The simulation of the dynamics and physics begins at 0530 LT, while the simulation of the chemistry starts at 0830 LT when the turbulent flow is established and cumulus clouds start to form. The whole simulation time with chemistry is 6 h. The chemical mechanism (Table A2 in Appendix A) is solved with an Euler backward iterative approximation using a Gauss-Seidel method with variable iterations. A convergence criterion of 0.01% is used for all the species.

3.2. Description of Experiments

[14] Several LES experiments (Table 1) are conducted for this study. The CLOUDY case has well-developed fair-weather cumulus as in Brown et al. [2002], chemistry with surface emissions of isoprene, monoterpene, and nitric oxide as given in Figure 1 and Table 1, and no modifications of photolysis frequency and isoprene emission by the clouds. In the CLEAR case the formation of clouds is suppressed by turning off the impact of changes in buoyancy due to latent heat of condensation. When comparing the CLEAR case with the CLOUDY case, the effect of cloud transport on the isoprene chemistry is revealed. The CLJ case, described below in more detail, is the same as the CLOUDY case except for the inclusion of cloud-modified photolysis frequencies of 31 reactions. The CLISOP case is the same as the CLJ case except for the inclusion of isoprene emission reductions due to clouds. Details about the modifications in the CLISOP cases are given below. For all scenarios, two NOx regimes (low and high NOx emissions) are examined.

Table 1. Description of Experimentsa
CaseCloud ExistenceNitrogen Oxide (NO) Emission (kg m−2 s−1)Cloud Impacts on Photolysis FrequencyCloud Impacts on Isoprene Emission
  • a

    Maximum surface isoprene flux: 1.4 × 10−9 (kg m−2 s−1). Maximum surface monoterpene flux: 8.4 × 10−11 (kg m−2 s−1).

CLOUDYYes NoNo
   Low NOx Emiss. 5.0 × 10−11  
   High NOx Emiss. 5.0 × 10−10  
 
CLEARNo NoNo
   Low NOx Emiss. 5.0 × 10−11  
   High NOx Emiss. 5.0 × 10−10  
 
CLJYes YesNo
   Low NOx Emiss. 5.0 × 10−11  
   High NOx Emiss. 5.0 × 10−10  
 
CLISOPYes YesYes
   Low NOx Emiss. 5.0 × 10−11  
   High NOx Emiss. 5.0 × 10−10  

[15] Under the presence of clouds, the photolysis frequency below, in, and above the clouds is different from the clear sky value [Madronich, 1987]. As in Chang et al. [1987], the ratio of cloudy sky to clear sky photolysis frequency is calculated to determine the photolysis frequencies below and above the clouds. The factor F (= jcloudy/jclear) above the cloud is defined as:

display math

[16] While, below the cloud, F is defined as:

display math

[17] Here tr, is the energy transmission coefficient for normally incident light (see Joseph et al. [1976] and Stephens [1984] for its definition), χ0 is the solar zenith angle, and α is a reaction dependent coefficient [Chang et al., 1987], which is set to 1.2, 0.7, 1.3, 0.9 for NO2, O3, NO3, and aldehydes, respectively. For other reactions, α = 1 is applied. Linear interpolation is assumed inside the cloud. The factor defined above is applied to regions above and below clouds in the CLJ case. For regions to the sides of clouds, clear-sky photolysis is assumed.

[18] Changes in isoprene emissions by clouds are modeled by modification of the canopy environment part (γCE) of the emission activity factor [see Guenther et al., 2006]. γCE is defined as

display math

where γLAI, γP, and γT account for variations associated with leaf area index, photosynthetic photon flux density (PPFD), and temperature. When clouds are present, the γP is modified. The emission activity factor associated with PPFD is defined as

display math

where Pmonth is the monthly average above-canopy PPFD (= 526.6 μmol m−2 s−1 in this study), a is the solar angle (angle of the sun above the horizon, degrees), and ϕ is the above-canopy PPFD transmission (non-dimensional). In this study, tr defined above is used to replace ϕ. To get the cloud-modified isoprene emissions for the CLISOP case, the surface isoprene flux (Figure 1) is scaled by the cloudy case γP normalized by the clear-sky case γP. As mentioned, the CLISOP case includes both changes to the isoprene emissions and photolysis rates. Quantitative analysis of isoprene emission reduction due to cloud shading is given in the context of simulated clouds in next section (Figure 2).

Figure 2.

Time evolution of (a) cloud fraction and (b) cloud top (solid line) and base (dashed line), and (c) changes in isoprene emission by cloud shade effects (= (CLISOP-CLJ)/CLJ %). In Figure 2c, solid (dashed) line represents full domain (cloudy points) average.

4. Meteorology Results

[19] The observed and simulated meteorology results for the cloud case studied here were discussed by Brown et al. [2002] and VKBP05. To set the simulated chemistry into the context of the meteorological variables, in particular clouds, we give a short summary of the meteorological setting. In the model simulation, clouds begin to form at 0830 LT (Figure 2) and become abundant (cloud fraction > 0.2) after 1030 LT. The maximum cloud depth reaches approximately 2000 m at 1400 LT when cloud top (CT) is ∼2700 m. Cloud base (CB) increases slowly with time from 700 m to 1200 m. Cloud base is close to the height of minimum heat flux, Zi, which is often used as a definition of the boundary layer (BL) height for the CLEAR case or subcloud layer height for the CLOUDY case. In this study, we assume CB ≈ Zi. The BL height (h) is defined following VKBP05 as the height at which the bottom-up passive scalar emitted from the surface reaches 0.5% of its surface value. This definition of BL height is useful for examining the extension of the BL volume due to clouds. Figure 2c shows the reductions in the isoprene emission influenced by cloud shading. Maximum reduction in the emission occurs after 12 LT. At these times, the change in the emission (= (CLISOP-CLJ)/CLJ•100) is −5%–10% in horizontal domain average and is −30%–40% in cloudy point average.

[20] The time evolution of potential temperature, total water mixing ratio, liquid water mixing ratio, and turbulent kinetic energy vertical profiles are shown in Figure 3 for the CLOUDY (solid lines) and CLEAR (dashed lines) cases. Figure 3c shows that appreciable amounts of liquid H2O are seen only at 12 LT and 14 LT. In the CLEAR case, the mean potential temperature and total water profiles indicate that the vertical growth of the boundary layer is limited, relative to that for the cloudy boundary layer. Differences in mean potential temperature between CLEAR and CLOUDY cases noticeably appear at 12 LT and 14 LT in the plot (the dashed lines sometimes overlap with the solid lines and cannot be seen in the figure). The presence of fair-weather cumulus makes the air warmer in the subcloud layer and cooler in the mid-to-upper cloud layer compared to the cloud-free case because of the enhanced exchange between the subcloud and cloudy layers and latent heating of condensation in the lower part of the cloudy region. Total water mixing ratio in the CLOUDY case is lower in the subcloud layer and higher in the mid-to-upper cloud layer compared to the cloud-free case (Figure 3b) because the buoyancy-enhanced vertical transport in the CLOUDY case extends the total water mixing ratio to higher altitudes. As a result, the cloud transport dilutes the subcloud water mixing ratio. The mean liquid water profile indicates that vertically well-developed clouds occur in the afternoon. Turbulent kinetic energy (TKE) in the mixed layer increases from 1 to 2 m2s−2 with time (Figure 3d). In the cloud-free case, TKE decreases sharply above the mixed layer. In the CLOUDY case, TKE is quite high above the mixed layer in the afternoon indicating increased energy in the system brought about by latent heat of condensation.

Figure 3.

Vertical profiles of (a) potential temperature, (b) total water mixing ratio, (c) liquid water mixing ratio, and (d) turbulent kinetic energy averaged for the time (one hour) and horizontal plane at 09 LT (red), 10 LT (blue), 12 LT (green), and 14 LT (black). Solid lines stand for CLOUDY case, while dashed lines denote CLEAR case. The vertical structure of cloudy boundary layer (mixed or subcloud layer, cloud layer, and free atmosphere) is depicted in the potential temperature profile at 14 LT. Cloud top (CT) and cloud base (CB) at 14 LT are marked with solid horizontal lines.

5. Chemistry Results

[21] To understand the role of BL dynamics and chemistry on affecting the vertical profiles of isoprene chemistry species, the time scales of the turbulent mixing and the chemical lifetime of each species need to be considered. The turbulence mixing time scale in this study is determined by the boundary layer height divided by the characteristic velocity of convective mixing. Recall, the boundary layer (BL) height, h in this study is defined as the height where the horizontally averaged mixing ratio of the bottom-up passive scalar becomes 0.5% of the horizontally averaged mixing ratio in the subcloud layer. The BL height here is close to the cloud top height (subcloud layer plus cloud layer) that the chemical species emitted at the surface reach by rising motions in the thermals in the subcloud layer and the clouds. Three velocity scales are calculated, (a) the traditional convective velocity as a function of boundary layer height and the sensible heat flux, (b) the convective velocity calculated using both sensible and latent heat fluxes, and (c) the velocity scale based on the integral of the virtual temperature flux suggested for a fair-weather cumulus topped boundary layer [Frisch et al., 1995]. In this study, an average of the three velocity scales is used as the characteristic velocity to calculate the turbulence mixing time scale. For the mid-day results, the turbulent time scale is ∼18 min. The chemical lifetime is based on the species reaction rate with the OH radical and varies depending on the low versus high NOx scenarios.

5.1. Distributions of Chemical Species in Low NOx Case

[22] To give a sense of how isoprene is distributed and how it is related to turbulent vertical motions and clouds in our LES results, vertical cross sections of vertical velocity and isoprene through the middle of the domain at 14 LT are described for the CLEAR (Figure 4) and CLOUDY (Figure 5) cases. The liquid water content in Figure 5a indicates the location of the clouds. In the CLEAR case (Figure 4), several thermals with maximum rising speed of 2.9 m s−1 extend vertically from the surface to 1000–1300 m. Plumes of isoprene are co-located with the strong updrafts. Since isoprene is emitted from the surface, it is abundant near the surface and lower BL and it decreases sharply in the interface between the BL and the free atmosphere. Above the height of 1400 m, isoprene mixing ratios are below 0.1 ppbv. The vertical motions in the CLOUDY case (Figure 5a) are enhanced compared to the CLEAR case, with maximum 5.9 m s−1 and minimum −4.3 m s−1 for the CLOUDY case. Clouds with liquid water content greater than 0.1 g kg−1 occur in updrafts above 1000 m altitude. The buoyancy created by the clouds produce strong updrafts enhancing the vertical transport of isoprene in plumes that reach >2000 m in altitude (Figure 5b). Within the cloud, local maxima of isoprene are found. Between the clouds, low isoprene air from the free troposphere is found to be intruding into the subcloud layer (<∼1000 m) (Figure 5b). At the cloud edges, subsiding shells (i.e., descending air mass surrounding the cloud) are found because of evaporative cooling induced by lateral mixing of cloudy and environmental air [Heus and Jonker, 2008]. The transport and chemistry of isoprene associated with subsiding shells will be discussed later. The results shown in Figures 4 and 5 clearly indicate that isoprene can be strongly influenced by turbulent motions and cloud transport.

Figure 4.

Instantaneous vertical cross section of vertical velocity (W) and isoprene at ∼14 LT for low NOx CLEAR case. (Minimum, Maximum, Interval, Unit) in contours for W, isoprene, and OH are (−2.52, 2.90, 0.5, ms−1) and (0, 3.396, 0.1, ppbv), respectively. For isoprene mixing ratio > 1.2 ppbv, contours of mixing ratio 1.5, 2.0, 2.5, and 3.0 ppbv are drawn. In contours of W, red (blue) color denotes zero and positive (negative) value.

Figure 5.

Instantaneous vertical cross section of (a) vertical velocity (W) with liquid water mixing ratio (ql), (b) isoprene for low NOx CLOUDY case, and (c) isoprene for high NOx CLOUDY case at ∼14 LT. (Minimum, Maximum, Interval, Unit) in contours for W, ql, isoprene for low NOx case, and isoprene for high NOx case are (−4.33, 5.93, 0.5, m s−1), (0, 1.5, 0.1, g kg−1), (0, 3.71, 0.1, ppbv), and (0, 2.26, 0.1, ppbv), respectively. For isoprene mixing ratio > 1.2 ppbv, contours of mixing ratio 1.5, 2.0, 2.5, and 3.0 ppbv are drawn. In contours of W, red (blue) color denotes zero and positive (negative) value. Grey colored contours in the top plot represent liquid water ≥ 0.1 g kg−1.

[23] To visualize isoprene fluctuations (deviations from a horizontal domain average) in association with the vertical motions (i.e., turbulent flux of isoprene) and covariance of isoprene and other chemical species, horizontal transects of vertical velocity and fluctuations of isoprene, MACR, NOx, and OH at 500 m and 1500 m altitude from Figures 5a and 5b are plotted (Figure 6). In the subcloud layer, isoprene fluctuations are large (up to 1 ppbv) and a positive correlation between isoprene fluctuation and vertical velocity dominates (Figure 6a), implying that the sign of the turbulent flux of isoprene would be positive. MACR and NOx are positively correlated with isoprene. However, the fluctuations of NOx and MACR are much smaller than isoprene. Isoprene is negatively correlated with OH, implying that the sign of covariance of isoprene and OH would be negative. Similar co-varying patterns are found in the cloud layer in which large fluctuations of chemical species in association with cloud-enhanced vertical motions occur (Figure 6b). To examine transport and chemistry around a cloud in detail, the 3–4.6 km section of the transect is shown in Figure 6c. Within the area of the updraft in the cloud, large enhancements (reductions) of isoprene, MACR, and NOx (OH) are found. Here, the isoprene mixing ratio (horizontal domain average plus fluctuation, 1.12 ppbv) is much larger than that of NOx (0.38 ppbv). This will be compared with the case with high NOx emissions later. As in Heus and Jonker [2008], we found small subsiding shells near the cloud edges. Within these descending air masses, positive fluctuations of isoprene, MACR, and NOx are found, which can contribute to negative turbulent flux of these species (see the areas defined by the arrows in Figure 6c). The OH mixing ratio is depleted in the subsiding shells due to the increase of isoprene. Despite the evidence for subsiding shells of high isoprene occurring, their contribution to the total turbulent flux of isoprene is small compared to the flux within the updrafts. For example, the horizontally and time-averaged isoprene flux profiles decrease with height and are positive from the surface to the top of the cloud layer. Quadrant analysis (decomposition of the turbulent flux into 4 components: updraft with positive isoprene fluctuation, downdraft with positive isoprene fluctuation, updraft with negative isoprene fluctuation, and downdraft with negative isoprene fluctuation; see Sullivan et al. [1998] for the application to heat flux analysis) confirms that positive flux components (updraft of high isoprene and downdraft of low isoprene) dominate the total isoprene flux in the subcloud layer. In the cloud layer, strong updrafts of high isoprene are primarily important in determining the isoprene flux.

Figure 6.

Horizontal transects at (a) 500 m, (b) 1500 m, and (c) zoomed-in 1500 m height in the vertical cross sections shown in Figures 5a and 5b for low NOx case. Dotted line with open circle denotes vertical velocity. Red, orange, green, and blue solid lines represent the fluctuations (deviations from horizontal domain average) of isoprene, MACR, NOx, and OH, respectively. Horizontal domain averages of isoprene, MACR, NOx, and OH at 500 m (1500 m) are 0.61 (0.12) ppbv, 0.29 (0.13) ppbv, 0.30 (0.13) ppbv, and 0.28 (0.39) pptv, respectively. Arrows in Figure 5c denote the subsiding shells with positive isoprene fluctuations.

[24] Horizontal transects from Figure 5b can also provide insights for aircraft observations of isoprene. Figure 7 shows selected horizontal transects from Figure 5 and illustrate the lower limit of the measurement frequency to capture the isoprene plumes in the cloud layer (∼1500 m altitude). For an aircraft flying with speed of 100 m s−1, the measurement frequencies of 4 s, 10 s, and 20 s would then be connected to the sampling of the LES results with 400 m, 1000 m, and 2000 m in Figure 7. In the cloud layer, the sampling interval of 1000 m already shows a large error in the isoprene average that ranges from −40% to 50% (Figure 7c). With the sampling interval of 2000 m, the error in the average ranges from −60% to 140% (Figure 7d). In the subcloud layer, the sampling of the LES results with 2000 m interval exaggerates or misses the isoprene plumes significantly with an error in the average being −20% to 30% (not shown). Thus, for the aircraft speed of 100 m s−1, a measurement frequency of a few seconds or less is necessary to capture isoprene plumes both in the subcloud and the cloud layers. Commane et al. [2010] also pointed out the need for high frequency isoprene measurements in order to calculate the HOx budget in the West African CBL.

Figure 7.

Variation of simulated isoprene with x axis at ∼1500 m altitude in Figure 5b for low NOx case with (a) a LES resolution (∼60 m), (b) a sampling interval of 400 m, (c) a sampling interval of 1000 m, and (d) a sampling interval of 2000 m. Dashed lines in Figures 7c and 7d represent the model results sampled with shifts of 400 m and 1000 m, respectively. AVG and SDEV denote an average and a standard deviation, respectively.

[25] We next examine the bottom-up and top-down passive tracers to isolate the role of transport from the chemistry (Figure 8) in vertical profiles averaged over time. Similar to that found by VKBP05, the horizontally averaged tracer profiles from the CLOUDY case show that the buoyancy-enhanced vertical transport increases the vertical extent of the bottom-up passive tracer up to 2500 m altitude, whereas the results from the CLEAR case has minimal exchange between the mixed layer and the region above. BL heights at 14 LT for CLEAR (hCLEAR) and CLOUDY (hCLOUDY) cases are ∼1700 m and ∼2500 m, respectively. In the middle of the subcloud layer (∼500 m altitude), the bottom-up passive tracer for the CLOUDY case is ∼10% lower than that for the CLEAR case. At hCLEAR level, the bottom-up tracer for the CLOUDY case is about a half of the subcloud layer value, while that for the CLEAR case is almost zero. The top-down passive scalar for the CLOUDY case is ∼7% greater than that for the CLEAR case at ∼500 m altitude. This study, by contrasting the CLEAR and the CLOUDY cases (Figure 8b), demonstrates that transport of scalars in the stably stratified cloud layer into the subcloud layer is more efficient than that of scalars in the stable, free atmosphere into the clear convective boundary layer (i.e., fumigation; see Kim et al. [2005] for more information) possibly due to the enhanced vertical motions by the addition of latent heating and evaporative cooling in the CLOUDY case. Verzijlbergh et al. [2009] attributed the transport or entrainment of scalars abundant in the cloud layer into the subcloud layer to subsiding shells or wave-like motions in the stably stratified cloud layer. Previous research examined extensively the dynamical processes of the entrainment of warm air in the stable layer into the clear convective boundary layer: engulfment of air above by compensating downward motions at the interfaces of rising thermals [Sullivan et al., 1998] and Kelvin-Helmholtz instability like wave motions in association with thermals in the sheared convective boundary layer [Kim et al., 2003]. The dispersion of scalars abundant in stably stratified layer into the clear convective boundary layer is also well documented [e.g., Cai and Luhar, 2002; Kim et al., 2005, 2007]. However, for the cloudy conditions, the entrainment of scalars abundant in the cloud layer into the subcloud layer has not been systematically studied. Dynamics responsible for this process needs to be better identified. In addition, influences of this process on chemistry need to be examined.

Figure 8.

Horizontally averaged passive scalar mixing ratio (arbitrary unit) profiles from CLOUDY (solid line) and CLEAR (dashed) cases: (a) bottom-up and (b) top-down scalars. These are averaged from 13:30 LT to 14:30 LT. Hereafter CT, CB, hCLEAR, hCLOUDY are cloud top, cloud base, BL height for CLEAR case, and BL height for CLOUDY case, respectively. BL height, h is defined as the height in which (Mh-MFA) becomes 0.5% of (MSFC-MFA). Here M denotes mixing ratio of bottom-up scalar and subscripts h, FA, and SFC represent BL height, free atmosphere, and surface.

[26] Vertical profiles (Figure 9) of isoprene, hydroxyl radical (OH), nitrogen dioxide (NO2), O3, MVK, MACR, formaldehyde (CH2O), and glyoxal (CHOCHO) demonstrate the impact of cloud transport on reactive scalars. Isoprene and NO2 are clearly enhanced in the cloud layer similar to the bottom up passive scalar (Figure 8). Compared to the bottom up passive tracer, isoprene and NO2 in the subcloud layer are not as vertically well mixed: slight negative vertical gradients in the middle of subcloud layer (= ∼500 m) exist. Further, isoprene and NO2 do not show the differences between the CLOUDY and CLEAR cases because of their rapid chemical reactions. Thus, isoprene, which has a chemical lifetime of ∼20 min, is controlled by both turbulent transport (time scale of 18 min) and chemistry. The LES isoprene profile is similar to observations of isoprene profiles in the southeast U.S. [Greenberg et al., 1999], which show that observed isoprene decreases as altitude increases, and that sharp gradients can occur near the surface.

Figure 9.

Horizontally averaged profiles of chemical species mixing ratio for CLEAR (black) and cloudy (blue: CLOUDY, green: CLJ, red: CLISOP) cases with low NOx emission: (a) isoprene, (b) OH, (c) NO2, (d) O3, (e) MVK, (f) MACR, (g) CH2O, and (h) CHOCHO. These are averaged from 13:30 LT to 14:30 LT. Heights of CB, CT, and hCLEAR are given in Figure 9a.

[27] The vertical profile of O3 (Figure 9) is similar to that of the top-down tracer (Figure 8). An increase (decrease) of the subcloud (cloud) layer O3 due to cloud transport is very small because of the weak O3 gradient across the subcloud and the cloud layers. A decrease in O3 in the cloud layer indicates that increases in O3 precursors such as isoprene and NO2 in this layer were not effective enough to produce O3 within hours.

[28] While the subcloud layer OH mixing ratios for the CLEAR and CLOUDY cases are quite similar, the cloud layer OH for the CLOUDY case is much lower (20–30%) than that for the CLEAR case. For the CLEAR case, OH increases across the BL inversion layer and reaches a peak at the BL top (hCLEAR). The OH chemistry budgets (Figure 10) can help explain the shapes of the simulated OH profiles. Both OH production and destruction for CLOUDY and CLEAR cases are very close below CB (<1000 m), but exhibit differences in the cloud layer. Although the CO + OH reaction loss rate is greater for the CLEAR case than the CLOUDY case between 1500 and 2500 m, the total loss rate for the CLEAR case is less than that for the CLOUDY case because the loss rates of BL species (e.g., isoprene) are an order of magnitude smaller in the CLEAR case than in the CLOUDY case for this altitude range. For the CLEAR case, the substantial decrease of the loss of OH reacting with isoprene between 1000 m and 1700 m (hCLEAR) leads to the large increase of OH mixing ratio with height. The reason that OH for the CLEAR case does not increase continuously between 1700 m and 2500 m is that OH production from the photolysis of O3 and reaction of O(1D) with H2O, the major OH production, decreases in this layer due to water vapor decrease (Figure 3) and the loss of OH reacting with CO is dominant in the free atmosphere. Constant CO and CH4 for the full domain are assumed throughout the simulations, which is not totally realistic as CO often decreases above the boundary layer. For the CLOUDY case, because of the vertical transport of isoprene to the cloud layer, OH loss reacting with isoprene still occurs in most of the cloud layer (up to ∼2200 m). Thus, OH for the CLOUDY case is lower than that for CLEAR case above CB (∼1000 m). While OH is controlled by chemistry processes, it is still indirectly affected by the BL dynamics because its major production and destruction precursors (O3, water vapor, isoprene, and NOx) are modified by the BL dynamics. Water vapor is important in determining both the cloud fields and oxidation capacity of the atmosphere.

Figure 10.

Horizontally averaged budget of OH (a) production and (b) loss. It is also averaged over time from 13:30 LT to 14:30 LT. Solid (dashed) lines represent CLOUDY (CLEAR) case. In Figure 10a, red, orange, green, blue, black, and gray colors denote OH production by O3 photolysis, reaction between HO2 and O3, H2O2 photolysis, reaction between HO2 and NO, total OH production, and ratio of summation of four OH production terms mentioned earlier to total OH production, respectively. In Figure 10b, red, orange, green, blue, light blue, black, and gray stand for OH loss reacted by CH4, CH2O, CO, Isoprene, NO2, total OH loss, and ratio of summation of five OH loss terms mentioned earlier to total OH loss, respectively.

[29] The products of isoprene oxidation, MACR, MVK, CH2O, and CHOCHO, have vertical profiles similar to the bottom-up passive tracer because their chemical lifetimes (∼60 min, ∼80 min, 3.5 h, and 3.5 h, respectively) are long compared to the turbulent time scale (∼18 min). Thus, transport primarily controls MVK and MACR mixing ratios resulting in a nearly constant vertical profile in the subcloud layer (or mixed layer for the CLEAR case). Vertically well-mixed characteristics can explain small fluctuations of MACR in Figure 6a. The simulated MVK and MACR mixing ratios are similar to those measured at Pittsboro, NC [Greenberg et al., 1999]. Recent studies [Volkamer et al., 2006; Galloway et al., 2011a] indicate that there may be larger CHOCHO yields from isoprene oxidation and that CHOCHO may have a shorter chemical lifetime (1–2 h) than those estimates in our study. In addition, uptake of glyoxal to cloud drops and aerosols [Ervens and Volkamer, 2010; Ervens et al., 2011; Galloway et al., 2011b; Washenfelder et al., 2011] may reduce CHOCHO in the cloud layer, thus altering the vertical profile as simulated for this study (Figure 9).

[30] The distribution of the chemical species can be affected by cloud scattering (which modifies the photolysis rates) and cloud shading, which reduces isoprene emissions. Modified photolysis rates can create substantial changes in instantaneous NO2 mixing ratios below, within, and above a single cloud, as was shown by VKBP05 using the NO-NO2-O3 triad chemistry. In our study, the NO2 mixing ratios below a single cloud increase by ∼20% and above the cloud decrease by ∼20%. Differences between CLJ and the CLOUDY case for OH are much larger than those in NO2 with a ∼70% decrease (increase) below (above) a single cloud. For the CLISOP case where both photolysis rates and isoprene emissions are modified by cloud, the isoprene mixing ratio decreases, especially near the surface under the cloud by ∼1 ppbv (∼40% reduction relative to CLJ case).

[31] Compared to changes seen around a single cloud, changes in the horizontal mean profiles of each chemical species due to different cloud processes are much smaller (Figure 9). Because the cloud fraction is fairly low (0.2–0.3), the horizontal domain average includes substantial clear sky regions that are not directly affected by the cloud-modified photolysis rates and isoprene emissions in this study. Isoprene in the CLJ case increases by ∼10% below CB relative to the other cases, as OH decreases due to the reduced actinic fluxes. However, the reduction of isoprene emission due to cloud shading (CLISOP case) decreases isoprene mixing ratios in the subcloud layer to the values found in the CLEAR and CLOUDY cases. NO2 mixing ratios in the subcloud layer increase by about 5% because of the reduced actinic fluxes (seen in both the CLJ and CLISOP cases). However, the reduction of isoprene emissions caused by cloud shading does not affect the profile of NO2. Similar NO2 increases for modified photolysis rates were found by VKBP05. While small differences can be seen in the O3 profiles between the different cases, their magnitude is insignificant. As expected, the OH mixing ratio decreases in the subcloud layer in the CLJ case compared to the CLOUDY case, and increases in the upper cloud and above cloud. When both the photolysis rates and isoprene emissions are modified by clouds (CLISOP case), OH mixing ratios increase slightly (∼5%) due to the reduction in isoprene mixing ratios below clouds and subsequent decreases in OH loss reacting with isoprene. MVK and MACR in the subcloud layer increase slightly (3–5%) when the reduction of photolysis frequencies due to clouds is incorporated because decreased OH destroys less MVK and MACR. When the reduction of isoprene emissions is also included (CLISOP), MVK and MACR in the subcloud layer are at their smallest mixing ratios among the 4 cases. Overall changes in CH2O and CHOCHO for CLJ and CLISOP cases are <3%.

[32] To determine the importance of fair weather cumulus clouds in the subcloud layer where regional and global models assume clear sky conditions, the fractional differences (%) are calculated between average subcloud mixing ratios in the CLEAR case and CLOUDY cases for isoprene, OH, NO2, O3, MVK, MACR, CH2O, and CHOCHO mixing ratios. Overall, as the photochemistry peaks and cloud depth thickens the differences between the CLOUDY and CLEAR cases become larger (Figure 11). The impact of cloud venting results in significant reductions in the volume-averaged mixing ratios of the species that have longer chemical lifetimes, such as MVK (∼13%), MACR (∼13%), CH2O (∼10%) and CHOCHO (∼12%), than the species with shorter chemical lifetime, e.g., isoprene, OH, and NO2 (Figure 11). The changes in photolysis frequency and isoprene emission due to clouds affect isoprene, OH, and NO2 mixing ratio substantially. Compared to the CLEAR case, reductions in photolysis frequencies along with cloud transport lead to ∼12% and ∼4% increases of isoprene and NO2 mixing ratios averaged across the subcloud layer, respectively, while it decreases OH by ∼12%. Reduction in isoprene emissions by clouds affects isoprene, OH, MVK and MACR significantly. Compared to the CLOUDY case with only photolysis modification (CLJ case), isoprene in the CLISOP case is reduced by ∼10% after 12 LT. Other species (MVK, MACR, O3) are affected <5% by the cloud-modified isoprene emissions. Thus, the implications of cloud transport and cloud modifications in photolysis and isoprene emission to regional and global scale models is that the models without these 3 cloud processes can overestimate isoprene, MVK, MACR, OH, CH2O, and CHOCHO by up to 5%, 15%, 15%, 12%, 12%, and 15%, respectively, and underestimate NO2 by 4% in the subcloud layer for the meteorological and low NOx chemical scenarios considered. In addition to the changes in the subcloud layer averages, strong enhancement of chemical species concentrations in the cloud layer indicates cloud-modifications of the fates of these species by changing environmental wind and stability and chemical regimes.

Figure 11.

Fractional differences (= (MCLOUDY-MCLEAR)/MCLEAR•100, M mixing ratio) averaged in the subcloud layer of isoprene, OH, NO2, O3, MVK, MACR, CH2O, and CHOCHO between CLOUDY and CLEAR cases for low NOx conditions. Solid, dashed, and dash-dotted lines represent the CLOUDY, CLJ, and CLISOP deviations from the CLEAR case, respectively.

[33] Because photolysis rates, isoprene emissions, and cloud development depend on solar radiation and heating, and therefore vary with the time of day, the isoprene chemistry also varies with time. We generally do not see differences between the CLOUDY and CLEAR cases until late morning (1000 LT or later) when the cloud fraction reaches 20% and cloud depth thickens (see Figures 2 and 3). The differences can be larger when modified photolysis rates and cloud-modified isoprene emissions are invoked, but substantial differences are not seen until 1130 LT when the cloud fraction is ∼30% and cloud depth is 1000 m on average.

[34] As mentioned, gas-to-liquid partitioning and aqueous phase chemistry in cloud droplets are not included in this study. Previous modeling studies [Lelieveld and Crutzen, 1990; Liang and Jacob, 1997; Frost et al., 1999; Barth et al., 2002, 2003] showed that cloud chemistry affects tropospheric chemistry by decreasing O3, OH and HO2. Measurements also showed reductions of OH and HO2 in clouds [Mauldin et al., 1997; Olson et al., 2004; Commane et al., 2010]. Results from a preliminary LES run that includes aqueous chemistry, which represented chemistry following Barth et al. [2002] plus the oxidation of organic aldehydes, showed small differences from the CLOUDY case in terms of domain-averaged profiles. However, simulated OH and HO2 along a horizontal transect through individual clouds showed large reductions as found in the previous measurements [Mauldin et al., 1997; Olson et al., 2004; Commane et al., 2010]. Thus, LES with cloud chemistry would be useful for interpreting aircraft measurements with high frequency. It will be important to explore more thoroughly cloud chemistry within the LES framework in the future.

5.2. Distributions of Chemical Species in High NOx Case

[35] As in the low NOx case, isoprene in the high NOx case is vertically transported and is enhanced near the clouds. However, the magnitude of isoprene in the high NOx case (Figure 5c) is much smaller than that in the low NOx case (Figure 5b). At 1500 m height, isoprene mixing ratios in the low NOx case reach ∼1.2 ppbv while those in the high NOx case reach only ∼0.2 ppbv despite having the same isoprene emissions for both NOx cases. The reduction of isoprene with increasing NOx emissions will be discussed further below.

[36] Vertical velocity and fluctuations of isoprene, MACR, NOx, and OH along the horizontal transects at 500 m and 1500 m altitude from Figure 5c are given in Figure 12. Similar to the low NOx case, fluctuations of isoprene, MACR, and NOx are positively correlated with vertical velocity. Thus, positive isoprene fluxes in the subcloud layer and the cloud layer are expected, which is similar to, yet smaller in magnitude than, the low NOx case. Fluctuations of MACR at 500 m are increased compared to the low NOx case due to its shorter chemical lifetime with high NOx mixing ratios which leads to MACR being less well-mixed in the subcloud layer. In the cloud layer (∼1500 m altitude), isoprene is strongly positively correlated with OH. Here NOx concentration in strong updrafts is much higher than that of isoprene. This suggests potentially positive covariance of isoprene and OH in the cloud layer with NOx being the main driver to produce more OH in the high NOx case. These results do not include aqueous chemistry, which would decrease OH within cloud [e.g., Mauldin et al., 1997]. Thus, further measurement and modeling studies of OH in cloudy, high NOx environments should be pursued. As in the low NOx case, subsiding shells at the edges of a shallow cumulus cloud can be found where enhancements of isoprene, MACR, NOx, and OH are co-located with the downdrafts along the side of the cloud (see the areas defined by the arrows in Figure 12c).

Figure 12.

The same as in Figure 6 except for high NOx case. Horizontal domain averages of isoprene, MACR, NOx, and OH at 500 m (1500 m) are 0.14 (0.01) ppbv, 0.13 (0.04) ppbv, 1.40 (0.52) ppbv, and 0.97 (0.91) pptv, respectively.

[37] In Figure 13, vertical profiles of isoprene, OH, NO2, O3, MVK, MACR, CH2O and CHOCHO averaged from 1330 LT to 1430 LT are shown. Similar to the low NOx case, isoprene exhibits a strong gradient near the surface where its source is located. Compared to the low NOx case (Figure 9a), the isoprene mixing ratio is reduced substantially in the high NOx case (Figure 13a), in which the NOx emissions are a factor of 10 higher and the average NO2 mixing ratio at 500 m is a factor of 5 higher. The NO2 mixing ratio also has a sharp gradient near the surface up to about 200m followed by a more moderate decrease throughout the remainder of the BL. With the additional NO emission, the NO + HO2 reaction increases OH which is responsible for the decrease in isoprene mixing ratios. For the low to high NOx cases, the chemistry is NOx-limited and follows the pattern that when more NOx is added to the system, more ozone and OH are produced causing depletion in hydrocarbons like isoprene. The relationship between NOx and isoprene is consistent with previous findings in the relationship between NOx and OH as predicted by models [Logan et al., 1981; Trainer et al., 1987a] and as seen in comparisons of models and ambient OH measurements [McKeen et al., 1997].

Figure 13.

The same as in Figure 9 except for high NOx case. Axis limits for NO2 and O3 are different from those in Figure 9.

[38] An examination of the CLOUDY case at high NOx emissions indicates that cloud-enhanced vertical transport of BL species behaves similarly in both the high and low NOx cases. Because the isoprene is depleted in the high NOx case (compared to the low NOx case), the dilution (increase) of isoprene in the subcloud (cloud) layer is not as substantial.

[39] For the high NOx case, O3 production is very efficient (Figure 13d). There is a 40 ppbv increase during five hours of simulation. Further, the O3 profile is shaped like a bottom-up passive scalar (Figure 8) except that O3 decreases near the surface due to dry deposition.

[40] The vertical profile of OH in the high-NOx case (Figure 13b) shows that OH decreases from the surface to 100 m and then increases from 100 m until it starts to decrease again near the top of the BL. Below CB, the OH for the CLOUDY case is slightly lower than that for the CLEAR case, while above it in the mid-to-upper cloud layer OH for the CLOUDY case is much higher than that for the CLEAR case. To explain the shapes of the simulated OH profiles in the high NOx regime and the differences between the two NOx regimes, OH budgets for the high NOx case are calculated (Figure 14). Both OH production and loss terms for the CLOUDY and CLEAR cases are very similar in the subcloud layer (<1000 m), but exhibit differences between 1000 m and 2500 m. The total OH production rate in the high NOx case is about three times larger than that in the low NOx case indicating more active photochemistry. While O3 photolysis is the largest source of OH in the low NOx case, the most dominant OH source in the high NOx case is the reaction between HO2 and NO, which reaches ∼4.5 pptv s−1 near the surface. OH production due to ozone photolysis in the high NOx case is still about 50% higher than that in the low NOx case in the subcloud layer. Thus, it affects isoprene depletion. Because the source of NO emission is located at the surface, OH production by the NO+HO2 reaction is the highest near the surface. This explains why horizontally averaged OH increases as the height decreases below 100 m for the high NOx case.

Figure 14.

The same as in Figure 10 except for high NOx case.

[41] The most dominant OH loss process below ∼100 m is the OH reaction with isoprene. Above 200 m altitude, OH destruction by CO and CH2O become the two major loss terms. The OH reaction with NO2 to form HNO3 is as important to OH loss as the reaction with isoprene in the high NOx regime while this term is negligible in the low NOx regime. The height where the OH mixing ratio starts to increase matches with the height (∼100 m) where the OH reaction with isoprene drops substantially.

[42] The reason that OH starts to decrease above 1000 m is that the production of OH from the reaction between HO2 and NO starts to decrease sharply around this altitude because NO decreases (similar to NO2 decrease in Figure 13). For the CLOUDY case, the OH production from HO2 + NO decreases slowly in the cloud layer and O3 photolysis, the second largest source of OH, is higher than that at the same height in the CLEAR case, which leads to a higher OH mixing ratio in this layer. The CLOUDY case has less OH production than the CLEAR case between 1000 m and 1500 m, while it has more OH production between 1500 m and 2500 m, which is associated with cloud transport of NO similarly to NO2 (Figure 13c).

[43] For the high NOx regime, MVK and MACR vertical profile differences between the CLOUDY and CLEAR cases (Figures 13e and 13f) are similar to those of NO2. The increased OH concentration leads to the reduction of the chemical lifetimes of MACR (20 min) and MVK (40 min) in the subcloud layer where the NOx and isoprene emissions are directly released. Thus, the profiles of MACR and MVK are more sensitive to the chemistry for the high NOx regime than for the low NOx regime, especially in the subcloud layer, and are similar in shape to isoprene for the low-NOx case (Figure 9a). This is consistent with the fluctuations of MACR along the 500 m altitude transect (Figure 12). The effect of cloud transport on the CH2O and CHOCHO vertical profiles (Figures 13g and 13h) for the high NOx case is similar to the bottom-up passive tracer.

[44] In a high NOx environment, the cloud effects on chemistry via modified photolysis rates and modified isoprene emissions due to cloud shading have a very small effect on isoprene, but have larger effects on the other species shown in Figure 13. The OH radical is influenced the most by the modified photolysis rates (CLJ case) with a 15% decrease in the horizontally averaged OH mixing ratios from the surface to ∼1500 m. The decreased OH affects MVK, MACR, CH2O, and CHOCHO, which are all greater by 5–15% in magnitude. The effects of the reduced isoprene emissions (CLISOP case) are similar to those described for the low NOx case.

[45] To highlight the implications for regional and global-scale chemistry transport models of not including fair weather cumulus clouds in their BL parameterizations, the diurnal variations of the fractional differences (%) in the subcloud averaged mixing ratios between the 3 cloudy simulations and the CLEAR case are examined for the high NOx case. Most species respond to the clouds similarly to those for the low NOx case except for O3, MVK, MACR, and CH2O (Figure 15). Cloud enhanced vertical exchange alone without modification of photolysis and emission changes leads to ∼4% (∼4 ppbv) reduction of O3 averaged across the subcloud layer at 14 LT compared to the CLEAR case. Suppressed photochemistry by reductions in photolysis frequencies and isoprene emissions causes slightly smaller O3 production by ∼5% at 14 LT (∼5 ppbv) relative to the CLEAR case. Modification of MVK and MACR due to clouds for high NOx case follows changes in isoprene. More active photochemistry in the high NOx regime reduces the chemical lifetime of MVK and MACR, which makes MVK and MACR more sensitive to changes in cloud properties. Reductions in CH2O averaged across the subcloud layer relative to the CLEAR case are smaller for the high NOx regime than those for the low NOx regime. Increases in the NO2 subcloud average in the CLJ and CLISOP cases compared to the CLOUDY case for the high NOx environment are slightly larger (up to ∼6%) than those for low NOx condition (up to ∼4%). Thus, the implications of cloud transport and cloud modifications in photolysis and isoprene emission to regional and global scale models is that the models without these 3 cloud processes can either overestimate or underestimate isoprene, MVK, MACR by up to 5% depending on the time of the day, overestimate OH, CH2O, and CHOCHO by up to 15%, 6%, and 15%, respectively, and underestimate NO2 by up to 6% in the subcloud layer for the meteorological and chemical conditions (high NOx emission) in this study.

Figure 15.

The same as in Figure 11 except for high NOx case.

6. Conclusions

[46] The influence of fair weather cumulus clouds on isoprene chemistry under different NOx regimes is examined in terms of cloud transport, cloud-modification of photolysis, and isoprene emission reductions below clouds using a large-eddy simulation (LES) model coupled with moderately complex chemistry. Our simulations show that cloud transport modifies the vertical distributions of chemical constituents substantially by transporting boundary layer species vertically ∼1000 m higher due to the buoyancy associated with clouds. Globally, daytime cumulus clouds over land occur ∼25% of the reported days during June–August, and have an average cloud amount of ∼33% [Hahn and Warren, 1999; Warren and Hahn, online publication, 2007]. Thus, the role of clouds should be considered in boundary layer parameterizations.

[47] To understand the role of BL dynamics and chemistry on affecting the vertical profiles of isoprene-chemistry species, the timescale of the turbulent mixing and the chemical lifetime of each species are compared. Isoprene, which has a chemical lifetime similar to the turbulence turnover timescale, does not show reductions in the subcloud layer, but increases significantly in the cloud layer. For MVK, MACR, CH2O, and CHOCHO, chemical species more abundant in the BL than the free atmosphere with chemical timescales longer than the turbulence turnover timescale, enhanced exchange between the subcloud layer and the cloud layer results in reductions of the species mixing ratios in the subcloud layer and substantial increases of those in the cloud layer.

[48] For the chemical species that are more plentiful in the free atmosphere than in the BL and that react slowly compared to the turbulence mixing timescale, e.g., O3 in the low NOx case, the clouds increase the mixing ratios in the subcloud layer due to the net transport from the cloud layer to the subcloud layer. Because OH has a much shorter chemical timescale than the turbulence timescale, chemistry controls the OH mixing ratio. However, OH mixing ratios are still indirectly affected by BL dynamics because the main OH production and loss rates are affected by species (O3, water vapor, isoprene, and NOx) influenced by the dynamics. During midday, isoprene has a chemical timescale similar to the turbulence timescale and thus its vertical distribution is influenced both by chemistry and turbulence. More details on the interaction between turbulence and isoprene chemistry will be discussed further in the next paper.

[49] Our study shows that different NOx levels change the lifetime of isoprene, MACR, MVK and other VOCs as a result of the modulation of OH concentration by NOx. In a low NOx regime, MACR and MVK are controlled by transport because their chemical timescales are large compared with the turbulent timescale. However, in a high NOx regime MACR and MVK have a chemical timescale comparable to the turbulence timescale and therefore both chemistry and transport control their distribution in the BL. Isoprene in the mixed layer decreases as NOx emission increases. Thus, the isoprene profile is determined by NOx emission as well as isoprene emission. Studies that use the observed isoprene profile to estimate the isoprene emission flux should therefore include information on NOx concentrations as well.

[50] Cloud-modified photolysis rates and the reduction of isoprene emissions due to cloud shading affect isoprene, OH, and NOx significantly for a single cloud. To examine the implications of fair-weather cumulus processes for regional or global scale chemical transport models, the horizontally averaged changes are mainly discussed. Cloud-modified photolysis rates increase the subcloud isoprene averaged over the horizontal domain by more than 10% due to reduced isoprene oxidation, compared to the CLEAR and CLOUDY cases. However, when including the reduction of the isoprene emissions by cloud shading, isoprene mixing ratios decrease back to the magnitude found for the CLEAR and CLOUDY cases. For low NOx conditions, the cloud-modified photolysis rates increase NO2 and decrease O3 mixing ratios by up to 5% in the subcloud layer averages. However, for high NOx conditions, the NO2 averaged across the whole boundary layer (subcloud layer + cloud layer) increase by 10% due to fair-weather cumulus clouds.

[51] For the high NOx emission case in this study, O3 is ∼100 ppbv in the mixed layer. Compared to the cloud-free case, O3 in the cloudy cases is 5% (∼5 ppbv) less, which implies that an air quality model without a proper implementation of fair-weather cumulus process can have a positive bias of simulated O3 by ∼5 ppbv.

[52] By sub-sampling the LES results, we can provide insight on the ability for aircraft measurements to sample the detailed structure of isoprene in the boundary layer. By sampling the LES along a horizontal transect at 400 m, 1000 m, and 2000 m intervals, we find up to −60 to 140% error in the average of “simulated” measurements, relative to the original high-resolution simulation results. Thus, a measurement frequency of a few seconds or less with an aircraft flying with a speed of 100 m s−1 is necessary to capture isoprene plumes.

[53] While this study has addressed boundary layer chemistry and dynamics for a typical convective boundary layer, there are several processes that still need to be addressed. The three-dimensional effects of clouds on photolysis rates and aqueous-phase chemistry were not considered here. The photolysis rates and their modification by clouds could be represented better by accounting for the angle of the direct beam of the sun and the scattering off the sides of the clouds. Lantz et al. [1996] showed that the NO2 photolysis rate at the surface either decreased by 50% or increased by 30% depending on whether the sun was blocked, or not. Aircraft measurements [Palancar et al., 2011] show a 10% effect of clouds on actinic flux in the free troposphere, although most of those measurements were made above cloud. Aqueous-phase chemistry is another important process to be represented, because of its effects on radical concentrations and their reservoirs such as H2O2 and CH2O and its potential impacts on isoprene nitrates and O3 [Treves et al., 2000; Horowitz et al., 2007]. The role of isoprene reaction products to the formation of secondary organic aerosols via aqueous phase chemistry is also an important research topic [Ervens et al., 2008; Ervens and Volkamer, 2010], for which a LES model that includes both detailed gas-phase as well as aqueous phase chemistry would be the right research tool. With the recent increase of measurements of isoprene and its products in different conditions of boundary layers, it will be useful to apply the LES model tool to interpret the observations from field campaigns. This can be fully achieved only when good meteorological data are obtained with the chemistry measurements.

Appendix A:: Tables of Initial Conditions and Dry Deposition Velocities of Chemical Species and a Chemical Mechanism Used in This Study

[54] In this section, we provide tables that introduce chemical mechanism and conditions used in this study. Initial conditions in the boundary layer and the free atmosphere and dry deposition velocities of chemical species are given in Table A1. Chemical reactions and rates are given in Table A2.

Table A1. Initial Conditions and Dry Deposition Velocities of Chemical Species
SpeciesBoundary Layer (ppbv)Free Atmosphere (ppbv)Deposition Velocity (m s−1)
O34.548 × 1017.000 × 1018.333 × 10−3
NO27.889 × 10−12.695 × 10−17.143 × 10−3
NO2.140 × 10−17.607 × 10−29.980 × 10−5
HNO37.991 × 10−11.166 × 1005.000 × 10−2
H2O21.646 × 1008.904 × 10−15.000 × 10−2
CH3OOH2.500 × 10−14.124 × 10−25.000 × 10−2
CH2O2.914 × 1003.267 × 10−11.149 × 10−2
ISOP(C5H8)3.996 × 1001.010 × 10−1-
C2H69.894 × 10−19.923 × 10−2-
C3H69.649 × 10−29.144 × 10−3-
TERP(C10H16)7.145 × 10−21.003 × 10−3-
C4H101.004 × 1009.196 × 10−2-
SO29.629 × 10−107.142 × 10−3
C2H42.111 × 10−19.823 × 10−2-
OH2.607 × 10−43.016 × 10−45.000 × 10−2
CH3CO31.190 × 10−31.029 × 10−45.000 × 10−2
MCO3(CH2CCH3CO3)2.663 × 10−43.940 × 10−55.000 × 10−2
HNO45.781 × 10−32.126 × 10−2-
HO22.431 × 10−21.909 × 10−25.000 × 10−2
N2O55.289 × 10−58.135 × 10−5-
NO31.240 × 10−44.432 × 10−5-
C2H5O27.191 × 10−41.455 × 10−45.000 × 10−2
ISOPO2(HOCH2COOCH3CHCH2)9.460 × 10−36.461 × 10−35.000 × 10−2
PO2(C3H6OHO2)5.691 × 10−43.142 × 10−45.000 × 10−2
MACRO2(CH3COCHO2CH2OH)1.419 × 10−34.154 × 10−45.000 × 10−2
CH3O25.488 × 10−33.263 × 10−35.000 × 10−2
PAN(CH3CO3NO2)3.565 × 10−12.192 × 10−25.000 × 10−2
MPAN(CH2CCH3CO3NO2)7.188 × 10−28.113 × 10−35.000 × 10−2
MACR(CH2CCH3CHO)2.911 × 10−12.813 × 10−25.000 × 10−2
MVK(CH2CHCOCH3)4.042 × 10−14.223 × 10−2-
CH3CHO6.767 × 10−14.106 × 10−25.000 × 10−2
POOH(C3H6OHOOH)2.043 × 10−25.049 × 10−35.000 × 10−2
CH3COOOH7.443 × 10−21.015 × 10−35.000 × 10−2
CHOCHO6.950 × 10−23.631 × 10−45.000 × 10−2
C2H5OOH2.475 × 10−23.307 × 10−35.000 × 10−2
GLYALD(HOCH2CHO)4.887 × 10−11.542 × 10−25.000 × 10−2
CH3COCHO2.205 × 10−19.050 × 10−35.000 × 10−2
ISOPNO3(CH2CHCCH3OOCH2ONO2)1.511 × 10−52.803 × 10−65.000 × 10−2
ONITR(CH2CCH3CHONO2CH2OH)1.048 × 10−11.673 × 10−25.000 × 10−2
HYDRALD(HOCH2CCH3CHCHO)2.030 × 10−14.427 × 10−25.000 × 10−2
CH3OH3.385 × 10−37.410 × 10−45.000 × 10−2
TERPO2(C10H17O3)6.159 × 10−42.198 × 10−5-
CH3COCH32.110 × 1001.370 × 10−4-
TERPOOH(C10H18O3)1.154 × 10−18.419 × 10−45.000 × 10−2
MACROOH(CH3COCHOOHCH2OH)1.427 × 10−22.391 × 10−35.000 × 10−2
HYAC(CH3COCH2OH)1.044 × 10−11.093 × 10−2-
RO2(CH3COCH2O2)5.472 × 10−58.283 × 10−95.000 × 10−2
ROOH(CH3COCH2OOH)1.196 × 10−36.460 × 10−85.000 × 10−2
XO2(HOCH2COOCH3CHCHOH)8.895 × 10−45.779 × 10−45.000 × 10−2
XOOH(HOCH2COOHCH3CHCHOH)2.604 × 10−25.152 × 10−35.000 × 10−2
ISOPOOH(HOCH2COOHCH3CHCH2)1.268 × 10−16.568 × 10−25.000 × 10−2
Table A2. Chemical Reactions and Rates
 Reactionk298 or jainline image
  • a

    Units for the photolysis frequencies are s−1, for the second-order reaction rate constants are cm3 molecules−1 s−1, and for the third-order reaction rate constants are cm6 molecules−2 s−1. Photolysis rate values are noontime values for June 21 at 36 °N; the simulation follows a diurnal profile. Second-order reaction rate constants are of the form inline image. Third-order reaction rate constants are of the form k = k0[M]/(1. + κ)Fcϕ where κ = k0[M] / k, ϕ = ( [1 + log 10(κ)]2 )−1 and [M] is the air concentration.

  • b

    O3 photolysis rate is computed as image2 × 10−10[H2O]/( 2.2 × 10−10[H2O] + 2.9 × 10−11[M] )) where H2O is the water vapor concentration.

  • c

    Here, the rate constant is of the form k = (ka + kb[M]) ( 1 + kc[H2O]).

  • d

    Here, the rate constant is of the form k = ka + kb[M]/(1 + kb[M]/kc).

  • e

    The CO + OH reaction rate constant has the form 1.5 × 10−13(1. + 0.6Patm) where Patm is the pressure (atm). Reaction rate coefficients are from Hess et al. [2000], Horowitz et al. [2003], and MOZART 2.2 chemical mechanisms (personal communications) and other references.

(R1)O3 + hv → 2.00 OH4.9 × 10−5b0.0
(R2)HO2 + O3 → OH2.0 × 10−15500.0
(R3)HO + O3 → HO26.8 × 10−14940.0
(R4)HO2 + OH → H2O + O21.1 × 10−10−250.0
(R5)OH + OH + O2 → O3+ H2O1.9 × 10−12240.0
(R6)OH + OH + M → H2O2 + Mk0 = 6.9 × 10−31 (T/300)−0.8
k = 1.5 × 10−11
Fc = 0.6
(R7)H2 + HO → HO2 + H2O6.7 × 10−152000.0
(R8)HO2 + HO2 → H2O2 + O2ka = 1.7 × 10−12c−600.0
kb = 4.9 × 10−32−1000.0
kc = 2.2 × 10−18−2200.0
(R9)H2O2 + hv → 2.00 OH9.8 × 10−60.0
(R10)H2O2 + OH → HO2 + H2O1.7 × 10−12160.0
(R11)NO + O3 → NO21.8 × 10−141400.0
(R12)HO2 +NO → OH + NO28.6 × 10−12−250.0
(R13)NO2 + hv → NO + O31.0 × 10−2500.0
(R14)OH + NO2 + M → HNO3 + Mk0 = 2.6 × 10−30 (T/300)−3.2
k = 2.4 × 10−11 (T/300)−1.3
Fc = 0.6
(R15)HNO3 + hv → OH + NO27.0 × 10−70.0
(R16)HNO3 + OH → NO3ka = 7.2 × 10−15d−785.0
kb = 1.9 × 10−33−725.0
kc = 4.1 × 10−16−1440.0
(R17)HO2 + NO2+M → HNO4 + Mk0 = 1.8 × 10−31 (T/300)−3.2
k = 4.7 × 10−12 (T/300)−1.4
Fc = 0.6
(R18)HNO4 + M → HO2 + NO2 + Mk0 = 1.8 × 10−31 (T/300)−3.2
k = 4.7 × 10−12 (T/300)−1.4
Fc = 0.6 (backward)
(R19)HNO4 + hv → 0.67 HO2 + 0.33 OH + 0.33 NO3 + 0.67 NO29.9 × 10−60.0
(R20)HNO4 + OH → NO24.6 × 10−12−380.0
(R21)NO2 + O3 → NO33.2 × 10−172500.0
(R22)NO3 + hv → 0.92 NO2 + 0.08 NO + 0.92 O32.6 × 10−10.0
(R23)NO3 + HO2 → 0.40 HNO3 + 0.60 OH + 0.60 NO23.5 × 10−120.0
(R24)NO3 + NO3 → 2.00 NO22.3 × 10−162400.0
(R25)NO3 + NO2 + M → N2O5 + Mk0 = 2.2 × 10−30 (T/300)−3.9
k = 1.5 × 10−12 (T/300)−0.7
Fc = 0.6
(R26)N2O5 + M → NO3 + NO2 + Mk0 = 2.2 × 10−30 (T/300)−3.9
k = 1.5 × 10−12 (T/300)−0.7
Fc = 0.6 (backward)
(R27)N2O5 + hv → NO3 + NO26.4 × 10−50.0
(R28)H2O + N2O5 → 2.00 HNO32.0 × 10−20.0
(R29)CH4 + OH → CH3O26.3 × 10−151800.0
(R30)CH3O2 + NO → CH2O + NO2 + HO27.7 × 10−12−180.0
(R31)CH3O2 + HO2 → CH3OOH5.6 × 10−12−800.0
(R32)CH3O2 + CH3O2 → 1.40 CH2O + 0.80 HO2 + 0.60 OACD4.7 × 10−13−190.0
(R33)CH3OOH + hv → CH2O + OH + HO29.2 × 10−60.0
(R34)CH3OOH + OH → 0.70 CH3O2 + 0.30 CH2O + 0.30 OH7.4 × 10−12−200.0
(R35)CH2O + hv → CO + 2.00 HO23.8 × 10−50.0
(R36)CH2O + hv → CO + H25.3 × 10−50.0
(R37)CH2O + OH → CO + HO21.0 × 10−110.0
(R38)CH2O + NO3 → CO + HNO3 + HO25.8 × 10−162900.0
(R39)CO + OH → HO22.4 × 10−13e0.0
(R40)SO2 + OH + M → SO4 + HO2 + Mk0 = 3.0 × 10−31 (T/300)−3.3
k = 1.5 × 10−12
Fc = 0.6
(R41)C2H6 + OH → C2H5O22.4 × 10−131100.0
(R42)C4H10 + OH → 2.00 C2H5O22.6 × 10−121100.0
(R43)C2H5O2 + NO → CH3CHO + NO2 + HO28.7 × 10−120.0
(R44)C2H5O2 + HO2 → C2H5OOH8.0 × 10−12−700.0
(R45)C2H5OOH + OH → 0.30 C2H5O2 + 0.70 CH3CHO + 0.70 OH1.1 × 10−11−200.0
(R46)C2H5O2 + C2H5O2 → 1.60 CH3CHO + 1.20 HO2 + 0.80 OALC6.8 × 10−140.0
(R47)C2H5O2 + CH3O2 → 0.74 CH2O + 0.74 CH3CHO + 0.96 HO2 + 0.78 OALC3.6 × 10−130.0
(R48)C2H5OOH + hv → CH3CHO + OH + HO29.2 × 10−60.0
(R49)C3H6 + OH + M → PO2 + Mk0 = 8.0 × 10−27 (T/300)−3.5
k = 3.0 × 10−11
Fc = 0.5
(R50)C3H6 + O3 → 0.087 CH4 + 0.56 HO2 + 0.307 CH3O2 + 0.39 CO + 0.393 OH + 0.533 CH2O + 0.0633 H2 + 0.5 CH3CHO + 0.333 OACD1.2 × 10−171900.0
(R51)PO2 + HO2 → POOH8.0 × 10−12−700.0
(R52)POOH + OH → 0.30 PO2 + 0.70 OH + 2.10 OALC1.3 × 10−11−200.0
(R53)POOH + hv → CH3CHO + OH + CH2O + HO29.2 × 10−60.0
(R54)C3H6 + NO3 → 3.00 ONIT9.4 × 10−150.0
(R55)PO2 + NO → CH3CHO + NO2 + CH2O + HO27.7 × 10−12−180.0
(R56)C2H4 + OH + M → 0.66 PO2 + Mk0 = 1.0 × 10−28 (T/300)−0.8
k = 8.8 × 10−12
Fc = 0.6
(R57)C2H4 + O3 → 0.44 CO + CH2O + 0.13 H2 + 0.49 HO2 + 0.37 OH + 0.37 OACD1.7 × 10−182600.0
(R58)CH3CHO + hv → CH3O2 + CO + HO28.0 × 10−60.0
(R59)CH3CHO + OH → CH3CO31.4 × 10−11−250.0
(R60)CH3CHO + NO3 → CH3CO3 + HNO32.4 × 10−151900.0
(R61)CH3CO3 + NO → CH3O2 + NO22.4 × 10−110.0
(R62)CH3CO3 + NO2 + M → PAN + Mk0 = 2.7 × 10−28 (T/300)−7.1
k = 1.2 × 10−11 (T/300)−0.9
Fc = 0.3
(R63)CH3CO3 + HO2 → 0.75 CH3COOOH + 0.25 O3 + 0.37 OACD1.3 × 10−11−1000.0
(R64)CH3CO3 + CH3CO3 → 2.00 CH3O21.6 × 10−11−550.0
(R65)CH3O2 + CH3CO3 → 0.75 CH3O2 + 1.00 CH2O + 0.75 HO2 + 0.25 OACD1.1 × 10−11−640.0
(R66)CH3COOOH + hv → CH3O2 + OH2.7 × 10−60.0
(R67)C2H5OOH + OH → CH3CO31.3 × 10−11100.0
(R68)PAN + hv → CH3CO3 + NO21.6 × 10−60.0
(R69)PAN + OH → 0.60 CH2O + NO2 + 0.80 OACD4.0 × 10−140.0
(R70)PAN + M → CH3CO3 + NO2 + Mk0 = 2.7 × 10−28 (T/300)−7.1
k = 1.2 × 10−11 (T/300)−0.9
Fc = 0.3 (backward)
(R71)GLYALD + hv → 2.00 HO2 + CO + CH2O2.4 × 10−50.0
(R72)GLYALD + OH → HO2 + 0.20 CHOCHO + 0.80 CH2O1.0 × 10−110.0
(R73)CHOCHO + hv → CH2O + CO9.7 × 10−60.0
(R74)CHOCHO + hv → 2.00 CO + H26.5 × 10−50.0
(R75)CHOCHO + OH → 2.00 CO + HO21.1 × 10−110.0
(R76)CHOCHO + NO3 → 2.00 CO + HO2 + HNO31.3 × 10−152700.0
(R77)CH3COCHO + OH → CH3CO3 + CO + H2O1.4 × 10−11−830.0
(R78)CH3COCHO + hv → CH3CO3 + CO + HO21.5 × 10−40.0
(R79)ISOP + OH → ISOPO29.9 × 10−11−410.0
(R80)ISOP + O3 → 0.40 MACR + 0.20 MVK + 0.07 C3H6 + 0.27 OH + 0.06 HO2 + 0.60 CH2O + 0.30 CO + 0.10 O3 + 0.20 MCO3 + 0.20 CH3COOH1.3 × 10−172000.0
(R81)ISOP + NO3 → ISOPNO36.6 × 10−13450.0
(R82)ISOPO2 + NO → 0.08 ONITR + 0.92 NO2 + HO2 + 0.51 CH2O + 0.23 MACR + 0.32 MVK + 0.37 HYDRALD4.0 × 10−12−180.0
(R83)ISOPO2 + NO3 → HO2 + NO2 + 0.60 CH2O + 0.25 MACR + 0.35 MVK2.4 × 10−110.0
(R84)ISOPO2 + CH3O2 → 0.25 CH3OH + HO2 + 1.20 CH2O + 0.19 MACR + 0.26 MVK + 0.30 HYDRALD1.9 × 10−12−400.0
(R85)ISOPO2 + CH3CO3 → CH3O2 + HO2 + 0.60 CH2O + 0.25 MACR + 0.35 MVK + 0.40 HYDRALD1.4 × 10−110.0
(R86)ISOPO2 + HO2 → ISOPOOH8.4 × 10−12−700.0
(R87)ISOPOOH + hv → 0.40 MVK + 0.29 MACR + 0.69 CH2O + HO29.2 × 10−60.0
(R88)ISOPOOH + OH → 0.50 XO2 + 0.50 ISOPO27.4 × 10−12−200.0
(R89)ISOPNO3 + NO → 1.21 NO2 + 0.79 HO2 + 0.72 CH2O + 0.17 MACR + 0.04 MVK + 0.79 ONITR9.4 × 10−12−360.0
(R90)ISOPNO3 + NO3 → 1.21 NO2 + 0.07 CH2O + 0.17 MACR + 0.04 MVK + 0.79 ONITR + 0.79 HO22.4 × 10−120.0
(R91)ISOPNO3 + HO2 → XOOH + 0.21 NO2 + 0.79 HO2 + 0.01 CH2O + 0.17 MACR + 0.04 MVK + 0.79 ONITR8.4 × 10−12−700.0
(R92)TERP + OH → TERPO24.6 × 10−11−400.0
(R93)TERP + O3 → 0.70 OH + MVK + MACR + HO28.6 × 10−17730.0
(R94)TERP + NO3 → TERPO2 + NO26.2 × 10−12−490.0
(R95)TERPO2 + NO → 0.10 CH3COCH3 + HO2 + MVK + MACR7.6 × 10−12−180.0
(R96)TERPO2 + HO2 → TERPOOH7.9 × 10−12−700.0
(R97)TERPOOH + hv → OH + 0.10 CH3COCH3 + HO2 + MVK + MACR (MOZ 2.2)9.2 × 10−60.0
(R98)TERPOOH + OH → TERPO27.4 × 10−12−200.0
(R99)MVK + hv → 0.70 C3H6 + 0.70 CO + 0.30 CH3O2 + 0.30 CH3CO34.4 × 10−50.0
(R100)MVK + OH → MACRO21.9 × 10−11−450.0
(R101)MVK + O3 → 0.80 CH2O + 0.95 CH3COCHO + 0.08 OH + 0.20 O3 + 0.06 HO2 + 0.05 CO + 0.04 CH3CHO2.4 × 10−12−360.0
(R102)MACR + hv → 1.34 HO2 + 0.66 MCO3 + 1.34 CH2O + 1.34 CH3CO35.1 × 10−70.0
(R103)MACR + hv → 0.66 OH + 1.34 CO5.1 × 10−70.0
(R104)MACR + OH → 0.50 MACRO2 + 0.50 H2O + 0.50 MCO33.5 × 10−11−180.0
(R105)MACR + O3 → 0.80 CH3COCHO + 0.28 HO2 + 0.20 CO + 0.20 O3 + 0.70 CH2O + 0.22 OH1.0 × 10−182500.0
(R106)MACRO2 + NO → NO2 + 0.47 HO2 + 0.25 CH2O + 0.25 CH3COCHO + 0.53 CH3CO3 + 0.53 GLYALD + 0.22 HYAC + 0.22 CO9.4 × 10−12−360.0
(R107)MACRO2 + NO → ONITR4.4 × 10−13−360.0
(R108)MACRO2 + NO3 → NO2 + 0.47 HO2 + 0.25 CH2O + 0.25 CH3COCHO + 0.53 CH3CO3 + 0.53 GLYALD + 0.22 HYAC + 0.22 CO2.4 × 10−120.0
(R109)MACRO2 + HO2 → MACROOH8.4 × 10−12−700.0
(R110)MACRO2 + CH3O2 → 0.73 HO2 + 0.88 CH2O + 0.11 CO + 0.24 CH3COCHO + 0.26 GLYALD + 0.26 CH3CO3 + 0.25 CH3OH + 0.23 HYAC1.9 × 10−12−400.0
(R111)MACRO2 + CH3CO3 → 0.25 CH3COCHO + CH3O2 + 0.22 CO + 0.27 HO2 + 0.53 GLYALD + 0.22 HYAC + 0.25 CH2O + 0.53 CH3CO31.4 × 10−110.0
(R112)MACROOH + OH → 0.50 MCO3 + 0.20 MACRO2 + 0.10 OH + 0.20 HO24.5 × 10−11−200.0
(R113)MCO3 + NO → NO2 + CH2O + CH3CO31.8 × 10−11−360.0
(R114)MCO3 + NO3 → NO2 + CH2O + CH3CO35.0 × 10−120.0
(R115)MCO3 + HO2 → 0.30 O3 + 0.30 CH3COOH + 0.70 CH3COOOH + 0.70 O21.2 × 10−11−1000.0
(R116)MCO3 + CH3O2 → 2.00 CH2O + HO2 + CO2 + CH3CO31.1 × 10−11−640.0
(R117)MCO3 + CH3CO3 → 2.00 CO2 + CH3O2 + CH2O + CH3CO32.7 × 10−11−530.0
(R118)MCO3 + MCO3 → 2.00 CO2 + 2.00 CH2O + 2.00 CH3CO31.4 × 10−11−530.0
(R119)MCO3 + NO2 + M → MPAN + Mk0 = 2.7 × 10−28 (T/300)−7.1
k = 1.2 × 10−11 (T/300)−0.9
Fc = 0.3
(R120)MPAN + hv → MCO3 + NO21.6 × 10−60.0
(R121)MPAN + M → MCO3 + NO2 + Mk0 = 2.7 × 10−28 (T/300)−7.1
k = 1.2 × 10−11 (T/300)−0.9
Fc = 0.3 (backward)
(R122)MPAN + OH → 0.50 HYAC + 0.50 NO3 + 0.50 CH2O + 0.50 HO2k0 = 8.0 × 10−27 (T/300)−3.5
k = 3.0 × 10−11
Fc = 0.5
(R123)CH3COCH3 + hv → CH3CO3 + CH3O21.5 × 10−60.0
(R124)CH3COCH3 + OH → RO2 + H2Ok = 8.8 × 10−12 exp (−1320/T) + 1.7 × 10−14 exp (423/T) (MOZ 2.2)
(R125)RO2 + NO → CH3CO3 + CH2O + NO27.7 × 10−12−180.0
(R126)RO2 + HO2 → ROOH + O27.9 × 10−12−700.0
(R127)ROOH + hv → CH3CO3 + CH2O + OH9.2 × 10−60.0
(R128)ROOH + OH → RO2 + H2O7.4 × 10−12−200.0
(R129)ONITR + hv → HO2 + CO + NO2 + CH2O8.0 × 10−60.0
(R130)ONITR + OH → 0.50 CO + 0.50 CH2O + HYDRALD + NO2 + HO21.5 × 10−110.0
(R131)ONITR + NO3 → HO2 + NO2 + HYDRALD2.4 × 10−151900.0
(R132)OH + HYDRALD → XO23.5 × 10−11−180.0
(R133)XO2 + NO → NO2 + 1.5 HO2 + CO + 0.25 HYAC + 0.25 CH3COCHO + 0.25 GLYALD9.4 × 10−12−360.0
(R134)XO2 + NO3 → NO2 + 1.5 HO2 + CO + 0.25 HYAC + 0.25 CH3COCHO + 0.25 GLYALD2.4 × 10−120.0
(R135)XO2 + CH3O2 → 0.30 CH3OH + HO2 + 0.70 CH2O + 0.40 CO + 0.10 HYAC + 0.10 CH3COCHO + 0.10 GLYALD1.9 × 10−12−400.0
(R136)XO2 + CH3CO3 → CO + CH3O2 + 1.5 HO2 + 0.25 HYAC + 0.25 CH3COCHO + 0.25 GLYALD1.1 × 10−11−640.0
(R137)XO2 + HO2 → XOOH8.4 × 10−12−700.0
(R138)XOOH + hv → OH9.2 × 10−60.0
(R139)XOOH + OH → H2O + XO23.6 × 10−12−190.0
(R140)XOOH + OH → H2O + OHK = 7.69 × 10−17 T2 exp (253/T) (MOZ 2.2)
(R141)HYAC + hv → CH3CO3 + HO2 + CH2O8.0 × 10−60.0
(R142)HYAC + OH → CH3COCHO + HO23.0 × 10−120.0

Acknowledgments

[55] Our thanks go to Kenneth Davis, Chin-Hoh Moeng, Peter Sullivan, and Ned Patton for initiation of the study and the development of NCAR LES model. Authors thank Alex Guenther and Christine Wiedinmyer for radiation data used to calculate the cloud shade effect. Authors thank Sungsu Park for a reference about cloud atlas. Helpful reviews by Don Lenschow and anonymous reviewers are greatly appreciated. A part of this work was supported by the EPA STAR grant R825379. The National Center for Atmospheric Research is sponsored by the National Science Foundation. The authors acknowledge support from NOAA's Health of the Atmosphere.