Relationships of coastal nocturnal boundary layer winds and turbulence to Houston ozone concentrations during TexAQS 2006



[1] Measurements made with a ship-based Doppler wind lidar during the summertime 2006 Texas Air Quality Study are used to study the relationship between lower-tropospheric vertical structure and winds and ozone (O3) concentrations in Houston, Texas, under two different flow regimes. We observed that strong southerly flow regimes, dominated by the subtropical anticyclone (Bermuda high) off the Atlantic coast of the United States, resulted in strong (i.e., high wind speed) onshore nocturnal low-level jets (LLJ) and low O3 and oxidant Ox (where Ox = O3 + NO2) concentrations at night and the following afternoon. In contrast, periods dominated by northerly or easterly flow resulted in relatively weak (low wind speed), but still onshore, nocturnal LLJs associated with higher concurrent and next-day concentrations of O3 and Ox. We present lidar data from 24 h example periods for each of these conditions and demonstrate how each type of flow regime is related to in situ ship-based ozone measurements. We expanded the study to include all days during the study when the ship was near Houston, to demonstrate how the strength of the meridional winds aloft show a better relationship to concurrent ship-measured Ox concentrations than the winds near the surface do. We found a strong relationship between a parameterization of the observed nocturnal jets, which reflect the synoptic conditions, to peak hourly O3 measured the next day at the ship and averaged throughout the Houston/Galveston/Brazoria continuous ambient monitoring stations monitoring network, indicating potential applications for planning air quality.

1. Introduction

[2] For protection of public and environmental health, summertime air pollution forecasting in the Houston, Texas, region demands a clear understanding of the effects of atmospheric transport and mixing on pollution concentrations and accurate prediction of ozone precursor emissions and meteorological conditions. Previous research [Davis et al., 1998; Ryerson et al., 2003; Banta et al., 2005; Darby, 2005; Nielsen-Gammon et al., 2005a, 2005b] demonstrated that, when the local meteorology is characterized by poor ventilation, combinations of background levels [Rappenglück et al., 2008; Langford et al., 2009] and summertime industrial and vehicular emissions in the Houston region frequently result in ozone concentrations that exceed regulatory values. Prior to March 2008 the United States Environmental Protection Agency defined an ozone exceedance as an 8 h averaged ozone surface concentration greater than 0.08 ppm (effectively, 84 ppbv with rounding). Meteorological factors that include mixing and transport can affect whether or not Houston will experience an ozone exceedance. For example, strong onshore flow, which tends to transport relatively clean marine air into the region, helps to dilute and disperse the Houston plume, while continental flow brings air masses that typically have higher levels of O3 than air masses from the Gulf of Mexico. Depending on wind speed, nonsoutherly synoptic flow can increase the chance of an ozone violation via stagnation or can transport the Houston plume out over the Gulf of Mexico or to areas east or west of Houston.

[3] To further investigate meteorological and chemical factors affecting air pollution in the Houston-Galveston Bay area, the Second Texas Air Quality Study (TexAQS 2006) took place during summer of 2006. During this experiment, scientists from the University of Colorado's Cooperative Institute for Research in Environmental Sciences (CIRES), the University of Washington, NOAA's Earth System Research Laboratory, and NOAA's Pacific Marine Environmental Laboratory deployed an array of air quality research instruments on NOAA's research vessel (R/V) Ronald H. Brown to study pollution in the Houston region and thus provide feedback for improving modeling capabilities for the region [Bates et al., 2008]. From 1 August to 11 September 2006, the ship traversed areas that included the Houston Ship Channel, Galveston Bay, and the Gulf of Mexico. Processed data from NOAA's high-resolution Doppler lidar (HRDL) [Grund et al., 2001], located on the aft deck of the Brown during this study, provided information about coastal atmospheric boundary layer (BL) winds and turbulence that affect pollution concentration in this region, including high vertical and temporal resolution profiles of BL winds and turbulence, low-level jets (LLJ), wind shear, vertical mixing, and BL mixing heights (MH). This transport and mixing information, in conjunction with in situ ship chemistry and aerosol measurements, provides a new perspective on current understanding of how synoptic and mesoscale coastal flow patterns affect Houston/Galveston air quality.

[4] During the 42 d ship-based portion of TexAQS 2006, on 9 of the days Houston's near-surface ozone network detected, at one or more stations, an ozone exceedance of either the effective 8 h National Ambient Air Quality Standard of 84 ppbv (with rounding, prior to 2008) or the 1 h standard of 124 ppbv (categorized by Texas Commission on Environmental Quality [TCEQ] as “unhealthy for sensitive groups”). These 9 d were listed by the TCEQ as “Houston High Ozone” days ( This paper presents the vertical structure of the flow and turbulence observed with the HRDL near Houston on two different days during the TexAQS 2006 study period: one with strong, persistent southerly flow under the influence of the subtropical (Bermuda) high, and the other with large-scale northerly flow and higher corresponding ozone concentrations. Continuously throughout the cruise, ship-based HRDL measurements provided vertical profiles of atmospheric winds and turbulence with high temporal (15 min) and spatial (5 to 30 m vertical) resolution, which have traditionally been difficult to obtain. Here we use these results to document the evolution of the vertical structure of the flow over the diurnal cycle for the 2 d and for two types of flow evolution, and to relate these flow conditions to ozone and oxidant (Ox = O3 + NO2) concentrations measured at the ship. Use of the quasi-conserved (on short time scales) Ox provides separation of the effects of local sources from any short-term meteorological influences.

[5] Additional background information on the meteorological conditions encountered in the Houston region during this study is presented in section 2. Descriptions of the measurements used in the analysis are provided in section 3. Section 4 contains 24 h example observations of BL winds, turbulence, and Ox concentrations under two types of conditions: (1) those under the influence of the subtropical high, during which HRDL observed flow with a persistent southerly component and a strong-speed (relative to the stationary surface) nocturnal LLJ, and (2) those where the synoptic flow was more easterly or northerly. The flow categorization is then extended to the rest of the near-Houston periods during the 42 d ship-based portion of TexAQS 2006 (section 5), and diurnal averages of various atmospheric boundary layer parameters for the different conditions are presented. The last part of this study (section 6) compares the relationships between nocturnal BL dynamical conditions observed at the ship, concurrent nocturnal Ox concentrations measured at the ship, and hourly averaged O3 concentrations measured the next afternoon, both at the ship and at continuous ambient monitoring stations (CAMS, maintained under the auspices of the TCEQ) throughout the Houston-Galveston-Brazoria (HGB) region.

2. Background

[6] An important meteorological process affecting air pollution in Houston is the diurnal sea breeze oscillation (Figure 1). In the absence of other forces, the sea breeze oscillation produces a tendency for onshore flow by the afternoon and offshore flow by early morning, before sunrise. This diurnal cycle is superimposed on a larger-scale synoptic gradient wind, which generally shows little variation over a diurnal period. According to this simplified conceptual model, the resulting near-surface wind in this coastal region is produced by an interaction between the diurnally varying sea breeze component, which has an amplitude of ∼3–4 m s−1 near the coast [Nielsen-Gammon et al., 2005b], and a slowly varying large-scale wind vector. Vector diagram examples of this idealized interaction between just the sea-land breeze cycle and two amplitudes of a large-scale southerly gradient wind are demonstrated in Figures 1b and 1c (similar to Figure 17 of Nielsen-Gammon et al. [2005b]). This simplified model does not include the effects of convective turbulence on afternoon wind speeds or any potential effects of strong synoptic flow on the baroclinity of the coastal region. The latter topic certainly warrants further study but is outside the scope of this paper.

Figure 1.

(left) Simplified polar plot and (right) time-based vector diagrams demonstrating (a) the pure sea-land breeze cycle and its superposition on a (b) weak or (c) strong synoptic-scale southerly flow. Effects of convective turbulence on the wind speed within the convective boundary layer or the strong synoptic flow on the baroclinity of the coastal zone are not included in this schematic. The y axis on the time-based vector plots indicates northerly (negative) and southerly (positive) wind speed amplitudes.

[7] The large-scale winds during the summer months in the southeastern United States are sometimes dominated by a strong subtropical anticyclone centered at or to the east of the Atlantic coast of the United States (the “Bermuda high” of Nielsen-Gammon [2002]). Under these conditions, the large-scale winds along the Texas coastline, which lies under the western branch of this high-pressure system, are southerly, persistent, and stronger than the amplitude of the sea breeze cycle. As a result, the winds in at least the lowest several hundred meters of the atmosphere (and often through the lower troposphere) have a southerly component during both the day and night (Figure 1c). The simplified superposition model predicts weakening southerly flow at night and into early morning; however, observations in the early night hours reveal that the flow within the first few hundred meters above the surface accelerated at night at coastal sites, owing to frictional decoupling from cooling stabilization, and decreased surface-driven convective turbulence. As reported in mesoscale modeling studies [Nielsen-Gammon, 2002] and observations with radar wind profilers [Nielsen-Gammon et al., 2005b; Nielsen-Gammon, 2006], the combination of the baroclinity and resulting acceleration of the flow aloft led to the formation of a LLJ in, but not confined to, the Houston region. The dynamics of the near-coastal flow are thus complex, especially at night, but the result under strong Bermuda high conditions is a relatively strong southerly component flow during the day that becomes even stronger at night. Because the southerly flow air mass originates over the Gulf of Mexico, the implications for air quality are that this flow regime is generally associated with low concentrations of air pollutants (clean air).

[8] At other times during the summer months, the influence of the subtropical high weakens over the Texas coastline, and the area is instead dominated by light or even moderate northerly (i.e., offshore) or easterly component gradient flow. During midsummer these conditions may occur by an eastward or northward displacement of the subtropical high center, but later in the summer (late August into September) the influence of the anticyclone may be counteracted by the southward penetration of weak cold front activity, as described by Rappenglück et al. [2008]. Such flows are often associated with high pollutant concentrations for two reasons (at least). First, the northerly component flow brings in continental air masses that tend to be warmer and drier and have higher background concentrations of pollutants than do air masses of Gulf origin [Nielsen-Gammon et al., 2005b; Nielsen-Gammon, 2006; Rappenglück et al., 2008; Langford et al., 2009]. Second, the superposition of the offshore gradient flow and the diurnal sea breeze cycle produces a period of stagnation and a reversal from offshore to onshore flow during the afternoon hours, a situation associated with the accumulation of pollutants in the boundary layer in the coastal zone near Houston [Systems Applications International et al., 1995; Banta et al., 2005; Darby, 2005]. Depending on emissions of ozone precursors, including oxides of nitrogen and volatile organic compounds [Ryerson et al., 2003; Murphy and Allen, 2005; Webster et al., 2007; Gilman et al., 2009], and depending on the direction and magnitude of the large-scale gradient wind vector, these flow conditions can lead to very high pollutant levels. For example, in the case described by Banta et al. [2005], from the first Texas Air Quality Study, in 2000 (TexAQS2000), the routine Houston area pollutant sampling network reported hourly averaged ozone concentrations of up to 200 ppb. As with the stronger southerly flow conditions, postsunset frictional decoupling can still result in acceleration of the sea breeze flow into a shallow LLJ early in the night.

[9] Under both kinds of flow conditions, the afternoon sea breeze, or bay breeze, which occurs almost daily in Houston, typically progresses into a nocturnal low-level jet in what Nielsen-Gammon [2006] refers to as the “sea breeze low-level jet.” This jet is similar to the summertime Great Plains LLJ [Parish et al., 1988; Holton, 1967; McNider and Pielke, 1981] in the sense that both are a result of topographically driven baroclinity. However, the close relationship with the sea breeze makes the LLJs in Houston physically distinct from the Great Plains LLJs or the synoptically forced LLJs over Texas and the western Gulf of Mexico [Uccellini and Johnson, 1979; Djurić and Damiani, 1980; Igau and Nielsen-Gammon, 1998]. Some of the weaker Houston sea breeze LLJs observed during TexAQS 2006 occurred only locally, independent of these other LLJs, such that they were not observable with profilers located further inland or further out to sea.

[10] The effect of the LLJ on pollution plumes has been studied in various locations. For example, McNider and Pielke [1981], Singh et al. [1997], Gupta et al. [1997], and Banta et al. [1998] demonstrate, via modeling and observations, the relationships between the vertical wind shear from the Great Plains LLJs, the near-surface mechanically generated turbulence, and dispersion of pollution. Likewise, McNider et al. [1988] show these relationships on modeled plumes in Australia. Specific to the Houston region, Nielsen-Gammon [2002] and Banta et al. [2005] demonstrate, using modeling and trajectory analysis, respectively, that the coastal nocturnal LLJ transports and disperses the Houston plume overnight.

3. Measurements

[11] Most of the data used in this study were acquired with instruments aboard the R/V Ronald H. Brown when the ship was close to Houston. The main ship locations near Houston, indicated by the ship track in Figure 2, are Barbours Cut (a container terminal at the northwest edge of Galveston Bay) and the east–west section of the Houston Ship Channel, which leads from the entrance of Galveston Bay to a point close to downtown Houston. Both of these locations tend to experience strong land influences, rather than the typical marine atmospheric boundary layer conditions encountered when the ship was in the Gulf of Mexico [Tucker et al., 2009]. The bodies of water around the ship in these locations were small enough (<1 km across) that, except for flow directly off of Galveston Bay, air parcels moving at the lowest wind speeds (1–3 m s−1) spent on the order of only 5 min over the local water before reaching the ship. Thus the ship-based measurements in these locations are effectively land measurements.

Figure 2.

Map showing coastal outlines around the Houston area. R/V Brown ship tracks (black line) made near Houston during TexAQS 2006 indicate the Houston Ship Channel. Barbours Cut is indicated with an arrow pointing to the northwest edge of Galveston Bay. Gray crosses indicate locations of all Houston-Galveston-Brazoria CAMS available during TexAQS 2006.

[12] Because of its proximity to both Houston and Galveston Bay, Barbours Cut is an ideal location for high temporal and vertical resolution observations of sea and land breeze cycles generated by the contrast between land and water conditions as well as observations of the nocturnal LLJ. For only a few periods during TexAQS 2006 was the Brown in a single location for a full 24 h, but if it was near Houston, the ship tended to overnight in one place, either Barbours Cut or the turning basin of the Houston Ship Channel, providing the opportunity to acquire continuous nocturnal boundary layer data characteristic of coastally influenced land conditions.

3.1. R/V Brown: High-Resolution Doppler Lidar Wind Measurements

[13] Detailed knowledge of wind speed and direction is essential if the gas phase and aerosol pollution measurements made aboard the Brown are to be linked to specific local sources, source regions, or general flow regimes. In most cases, because local transport is assumed to occur near the surface, wind speed and direction measured with ship-based anemometers (sonic, wind vane, etc., corrected by accounting for ship orientation) are used [Gilman et al., 2009; Bates et al., 2008; Lerner et al., 2009]. Back trajectories, which typically implement surface and radar wind profiler data to estimate larger-scale winds, have also been used to identify source regions for the ship-based chemistry measurements [Bates et al., 2008; Massoli et al., 2009].

[14] Lower-tropospheric wind profiles above the ship during TexAQS 2006 were measured with balloon radiosondes, launched once every 6 h; a 915 MHz radar wind profiler (RWP) with 1 h averaging; and the HRDL, which provided a profile once every 15 min. The sonde data suffered from issues with ship wake in the lowest 50–100 m as well as long intervals between launches, while the high minimum altitude of the RWP, ∼500 m (attributed to sea clutter by Wolfe et al. [2007]), made it difficult to correlate the data from these instruments with in situ chemistry measurements. More direct comparisons are possible with profiles from the continuous-scanning HRDL, a 2 μm wavelength, coherent Doppler lidar that provides line-of-sight wind velocity estimates with 30 m range resolution, and ∼20 cm s−1 velocity precision at a 2 Hz update rate. The HRDL provided high vertical resolution (5–30 m) profiles of mean horizontal wind speed and direction from ∼8 m above the surface to the top of the aerosol layer. The maximum range of 2–8 km depended on scan geometry, aerosol backscatter conditions, and cloud/precipitation (the lidar's near-infrared wavelength cannot penetrate clouds of optical depth greater than about 2 km). The system's motion-stabilized [Hill et al., 2008] full hemispherical scanning capability allows implementation of conical azimuth scans at various elevations, vertical slice (elevation) scans, and zenith stares. Measurements from the vertical-slice and zenith scans were used to estimate horizontal and vertical velocity variance (σh2 and σw2, respectively) with 30 m vertical resolution. The low-elevation azimuth scans and zenith stares provided two-dimensional wind and backscatter fields in the horizontal and vertical planes. Finally, the HRDL also provided 30 m vertical resolution profiles of uncalibrated aerosol backscatter. Details of the HRDL scanning and data acquisition procedures during TexAQS 2006 are described by Tucker et al. [2009]. The plots in Figure 3 (discussed in further detail in section 4) show examples of the wind speed and variance profiles measured by the HRDL system under two different flow conditions on 27 August (top row) and 2 September (bottom row).

Figure 3.

Nocturnal wind speed (left) and velocity variance (right) profiles for the time period between 0645 and 0745 UTC on (top panels) 27 August 2006 and (bottom panels) 2 September 2006. Four 15 min profiles are provided for each period. Horizontal velocity variance (dashed lines) profiles were available in the lowest 300 m and vertical velocity variance profiles (solid lines) were available from 200 m up.

3.2. R/V Brown: In Situ Chemistry Measurements

[15] Other ship-based instruments used during TexAQS 2006 included a flux tower [Fairall et al., 1997], a full in situ meteorological station, balloon-borne radiosondes, and a suite of in situ aerosol and gas-phase chemistry measurements [Bates et al., 2008; Parrish et al., 2009]. We use only the in situ O3 and NO2 measurements in this study. O3 was measured by UV absorbance with a commercial instrument (Model 49C; Thermo Environmental Instruments, Inc.) [Williams et al., 2006]. NO2 was measured with two different techniques: photolysis, followed by O3-induced chemiluminescence [Lerner et al., 2009], and pulsed cavity ring down spectroscopy [Osthoff et al., 2006]. Because NO2 concentrations made with the photolysis technique were limited to values under 50 ppbv, the data obtained by cavity ring down spectroscopy were used during periods of high NO2, or when the chemiluminescence data were missing. Depending on the ship's location, these and other chemistry instruments were able to detect the existence and extent (depending on ship track) of pollution plumes in the Houston area, Galveston Bay, or Gulf of Mexico [Bates et al., 2008].

3.3. Land-Based Measurements: Continuous Ambient Monitoring Stations

[16] Surface measurements of O3 from CAMS (described in detail at and by Langford et al. [2009]), maintained under the auspices of the TCEQ, were used to generalize the analysis. These measurements enable additional study of the relationship between the coastal LLJ and ozone concentrations measured throughout the HGB region.

4. Nocturnal Boundary Layer Conditions

[17] The nocturnal BL is characterized with meteorological parameters such as temperature, humidity, wind speed, wind direction, and velocity variance (turbulence). For this study, we start by comparing and contrasting the nocturnal BL conditions observed on two different days when the ship was near Houston overnight. We show measurements from a 24 h period (UTC day), 27 August 2006, that started with a strong onshore LLJ at night associated with low concurrent and next-day Ox concentrations. We then present measurements from a contrasting 24 h period, 2 September 2006, a night with weak onshore flow near the surface, stronger northerly flow aloft, and corresponding high Ox concentrations. In some cases, the strength of a jet refers to the perturbation in the winds relative to the background flow; in this study, however, the words “strong” and “weak,” as applied to the LLJs, refer to the absolute maximum speed of the jet relative to the stationary surface. A“day” in this study refers to the 0000 to 2400 UTC time period, where 1800 Central Standard Time (CST) is the start of the new UTC day (0000 UTC), and noon CST is 1800 UTC. Local sunrise was shortly before 1200 UTC.

4.1. Strong Nocturnal Jet/Low Ozone

[18] During two periods of the experiment, 10–14 August and 26–28 August, meteorological conditions were dominated by southerly onshore flow and strong nocturnal LLJs. The atmosphere in the lowest 2 km during these periods demonstrated strong jet-shaped nocturnal wind profiles (i.e., wind profiles with speed maxima at heights below 2 km above ground level [agl]; see Figure 3, upper left panel) that reached peak speeds between 9 and 13 m s−1 and vertical extents up to 1500 m. Using time-altitude displays, Figure 4 shows an example of wind profiles characteristic of a strong LLJ, observed while the ship was in the Houston Ship Channel on Sunday, 27 August 2006. The highest wind speeds of 10–13 m s−1 are shown as yellow to brown colors, according to the color bar on the right of the image in Figure 4a. Surface wind speeds were lower (2–3 m s−1), resulting in strong shear and corresponding turbulence in the lowest 400 m (Figure 4c). The low-level (<500 m agl) wind direction profiles (Figure 4b) demonstrate a version of the diurnal/inertial oscillation, starting at 0000 UTC with southeasterly flow that follows the previous day's afternoon bay/sea breeze [Nielsen-Gammon et al., 2005a, 2005b], then rotates clockwise through southerly (∼0900 UTC) to southwesterly (by 1200 UTC) flow throughout the night. The bay/sea breeze started again around 1700 UTC and brought the flow back to southeasterly.

Figure 4.

High-resolution Doppler lidar (HRDL) profiles of (a) horizontal wind speed and (b) wind direction for 27 August 2006 with (c) corresponding velocity variance profiles. The strong LLJ appears as a yellow to brown “blob” in Figure 4a. The colored bar at the top of Figure 4c indicates periods when the ship was in Barbours Cut (red) and in the east–west portion of the Houston Ship Channel (green). (d) Ship-measured Ox and O3 are indicated for the same 24 h period.

[19] Figure 4c shows the corresponding HRDL velocity variance (i.e., turbulence) profiles for this period, imaged (log scale) using color for amplitude. The black line plotted over the turbulence profiles indicates the BL mixing height as determined by using a threshold method described by Tucker et al. [2009]. The profiles demonstrate how after dark (∼0200 UTC), when the LLJ formed, the strong shear-generated mechanical turbulence defined the height of the shallow, surface-based, nocturnal mixing layer of 250–400 m depth [McNider and Pielke, 1981; Tucker et al., 2009; Pichugina and Banta, 2009]. This jet-induced mechanical turbulence reached up into the core of the jet, where increasing jet wind speed corresponded to greater fetches. Performance of 48 h back trajectories (not shown) using HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory and Real-time Environmental Applications and Display sYstem, data available at; see also READY (Real-time Environmental Applications and Display sYstem), details at, both websites at the NOAA Air Resources Laboratory) at 40 km resolution indicated that air masses from the surface to at least 2.5 km agl during this period came from over the Gulf of Mexico. The turbulent mixing into the core of the jet resulted in entrainment of the relatively cleaner marine air at these levels, while surface pollutants and ozone precursors were mixed up and into the jet, where they could be transported a significant distance away before morning (e.g., a 10 m s−1 wind speed in the jet implies a potential transport of 72 km within 2 h).

[20] After sunrise (shortly after 1200 UTC), the jet wind speeds slowed as surface heating again led to the growth of the convective BL, entraining the relatively clean air aloft. The synoptic conditions that led to formation of the strong nocturnal jet persisted through the day so that wind speeds within the new convective BL were relatively high, compared with those on higher ozone days, and still contained a southerly component. There was no land breeze, offshore flow, or stagnation during this period, consistent with the simplified conceptual model in Figure 1c. Correspondingly, ozone and Ox concentrations (Figure 4d) remained relatively low the rest of the day. The afternoon sea breeze on this day occurred as a relatively quick (less than 30 min) shift in the mean winds from southwesterly to southeasterly in the midafternoon around 1900 UTC.

4.2. Weak Nocturnal Jet/High Ozone

[21] In contrast to 27 August, the dominant/synoptic flow on 2 September 2006 was mostly northerly, with strong northerly gradient winds of 12 m s−1 and greater above about 2 km agl. These winds opposed the baroclinity of the sea breeze LLJ system resulting in a weak, shallow, onshore LLJ that veered from southerly to westerly and finally gave way during the night to a land breeze and offshore flow by 1030 UTC. Correspondingly, this day demonstrated higher than average nocturnal Ox and peak afternoon O3 concentrations at the ship and in Houston. This day did not result in an ozone exceedance; however, the peak hourly averaged O3 concentration in Houston was 90 ppbv at Manvel Croix Park and the peak 8 h averaged O3 concentration was 83 ppbv (2 ppbv below the effective exceedance level) at Smith Point Hawkins Camp. The peak hourly averaged O3 measured at the ship was 87 ppbv. All of these levels are considered moderate for Houston.

[22] HRDL wind speed (Figure 5a) and direction (Figure 5b) measurements made in Barbours Cut on this night revealed a weak and shallow (below 500 m) nocturnal jet between ∼0100 and 1100 UTC. The ∼5–7 m s−1 maximum speed of the weak jet, combined with the short period of southerly flow (compared with the jet on 27 August), meant that less relatively clean air was brought into the region on this night. Still, the O3 and Ox concentrations (Figure 5d) both dropped as the convective mixing decreased and speed of the jet increased (between 0000 and 0300 UTC). These concentrations then leveled off until after the development of the land breeze around 1100 UTC. The further drop in Ox and O3 at this time, with the weak northerly/northeasterly land breeze, indicates a change in air mass to one with more chemical losses, possibly a result of more surface deposition from vegetated areas north of the ship.

Figure 5.

HRDL-derived profiles of (a) horizontal wind speed, (b) wind direction, and (c) velocity variance for 2 September 2006. The weak jet appears as a light green region under 500 m in Figure 5a between 0100 and 0900 UTC. The stagnant layer above the jet appears as a blue layer around/above 500 m. The colored bar along the top of Figure 5c indicates when the ship was in Barbours Cut (red), in the east–west portion of the Houston Ship Channel (green), crossing Galveston Bay (blue), and in the Gulf of Mexico (white, after 2200 UTC). (d) Ship-measured 1 min concentrations of Ox and O3 are indicated from the same 24 h period.

[23] In situ chemistry measurements on the ship [Lerner et al., 2009; Williams et al., 2009] indicated that the weak flow during the jet may have also transported pollutants (e.g., sulfur compounds, NOx, CO) from the industry/roads/shipping channels south and west of Barbours Cut toward the ship's location with little of the dispersion that would occur with higher mixing [Gilman et al., 2009] and stronger jet speeds. Results for 24 h back trajectories for winds in the lowest 500 m showed that the air mass coming in with the southerly LLJ was a recirculated offshore flow from Beaumont and the Texas/Louisiana border.

[24] Above about 1000 m the winds transitioned with increasing altitude to the stronger synoptic northerly/northeasterly flow. Starting with the afternoon sea breeze from the previous day, as the weak southerly (turning to southwesterly) nocturnal LLJ formed underneath the northerly/easterly continental flow, a stagnant layer formed above 500 m (observed as a dark blue layer between 500 and 1000 m in the HRDL wind speed profiles in Figure 5a). The height of the stagnant layer was within the ∼2 km depth of the afternoon convective mixing layer from the previous day (1 September), meaning that ozone and ozone precursors lofted to those levels the previous day were not removed far from the area overnight. An ozone-sonde launched from the University of Houston (29.72°N, 95.34°W, about 30 km west of Barbours Cut) revealed residual ozone concentrations of up to 70 ppbv within the stagnant layer [Morris et al., 2006, 2009]. The high-speed northerly/northeasterly winds above about 1500 m throughout the day were representative of the continental flow that brought relatively ozone rich (∼60–80 ppbv) continental air into the region during this period as discussed by Rappenglück et al. [2008] and Langford et al. [2009]. After sunrise, the growing convective BL (observed in the variance profiles in Figure 5c) eventually reached the height of the ozone-rich stagnant layer from the previous night and then the height of the continental flow layer, mixing these air masses down toward the surface.

[25] The wind direction profiles in the lowest 500 m on this day (Figure 5b) demonstrated the full diurnal/inertial oscillation discussed by Nielsen-Gammon [2002] and Nielsen-Gammon et al. [2005b], starting with a weak and variable southeasterly flow that grew into the weak jet, veered through southerly to southwesterly, and then became westerly. As previously mentioned, around 1100 UTC the flow in the lowest 150 m became abruptly northerly then northeasterly because of a land breeze formation that is discussed in further detail by Day et al. [2009]. The afternoon bay/sea breeze, observed from the ship's positions crossing Galveston Bay between 1700 and 2100 UTC, brought the flow, though weak, back to southeasterly, completing the oscillation. Darby [2005] and Banta et al. [2005] describe how high ozone levels often follow periods such as this: an offshore morning flow that pushes the Houston plume out over Galveston Bay can be followed by an afternoon sea breeze, a breeze that after a period of stagnation can recirculate the plume over the city to mix with local pollution. As shown in Figure 5d, the highest ship-measured, 1 min-averaged ozone concentration on this afternoon, measured over Galveston Bay, was 100 ppbv.

4.3. Turbulence and Mixing Height

[26] As with the observed large differences in winds within the nocturnal LLJ and aloft on the strong versus the weak jet night examples, the mixing heights and strength of turbulence in the boundary layer on these nights exhibited differences that illuminate how the nocturnal boundary layer dynamics affected Ox concentrations. Rappenglück et al. [2008] conclude that the relationship between mixing height and ozone concentration is still unclear. We aim to clarify this relationship for nocturnal conditions through the use of turbulence profiles, although the matter is admittedly often complicated by changing wind directions and transport at the top of the boundary layer. To understand the effect of turbulence and mixing height on Ox concentrations, it is important to differentiate the ways in which turbulence manifested in this region on these nights. On the strong LLJ night (27 August, Figure 4), the 200–400 m deep nocturnal boundary layer turbulence was mechanically shear-driven, similar to those LLJ-driven stable boundary layer conditions over the Great Plains of the United States [McNider and Pielke, 1981] and in Kansas and eastern Colorado [Banta et al., 2002, 2003, 2006; Pichugina and Banta, 2009]. On the weak jet night (2 September, Figure 5), the nocturnal boundary layer in this region exhibited many of the properties characteristic of a stable boundary layer, including weak turbulence, as described by Mahrt and Vickers [2006], Banta et al. [2007], and references therein.

[27] Using vertical slice or “elevation” scans, the HRDL data can be used to generate low-level profiles of horizontal velocity variance. This was done for the TexAQS 2006 data by first removing the horizontal mean wind field from each horizontal velocity estimate in the low-elevation scans, binning the residual data into 30 m vertical bins and calculating the atmospheric variance for the lowest 300 m as described by Tucker et al. [2009]. Above 300 m, vertical velocity variance profiles, calculated using zenith stare data (ibid.), were used to observe the BL turbulence. Images incorporating these two types of mixing profile for the strong LLJ night example (27 August) and the weak LLJ night example (2 September) are provided in Figure 4c and Figure 5c, respectively. Turbulence in very stable nocturnal boundary layers tends to be weak and sporadic compared to that of shear-driven boundary layers under strong nocturnal LLJs – and this is evident in the comparison of these examples. Compared to the profiles from 27 August (compare Figure 3), the turbulence profiles acquired on 2 September (Figure 5c) demonstrated weaker and shallower atmospheric boundary layer mixing. The reduced mixing meant less dilution of nocturnally and locally emitted precursors, enhancing the effects of less transport from weaker winds. Because of the reduced levels of transport and lower mixing heights, higher ozone precursor (Ox) concentrations of 40–60 ppbv near the surface (inside the mixing layer) were present before the daytime production began, as compared with only ∼20 ppbv on 27 August.

[28] Mixing heights (MH) too, were different on these two nights. Knowledge of MH values is important for correctly inferring emission rates with in situ sensors during TexAQS 2006 [Gilman et al., 2009] as well as for producing accurate output of transport and dispersion models as described by White et al. [2009]. The nocturnal mixing heights were slightly higher (on average) on the strong jet night because the shear was stronger and the shear layer was thicker in height. The high variability in MH on 2 September is the result of weak sporadic turbulence as well as the land breeze induced changes in wind direction.

[29] To illustrate the role that the synoptically modified jet winds play in reducing nocturnal surface pollution and precursor concentration in Houston, we provide a simplified approach to the transport and mixing time scales, both at the surface and aloft. Figure 3 contains sample profiles of wind speed and turbulence for the two example nights: 27 August to represent the strong LLJ data and 2 September to represent the weak LLJ. On 27 August (top row), the average near-surface meridional wind speed was approximately 3 m s−1. Assuming 75 km of distance between the near Houston location, with its previous day and overnight emissions, and the Gulf of Mexico, with its relatively cleaner air masses, approximately 7 h of transport time would have been required at the 3 m s−1 speed for marine air to reach the Houston location if no mixing occurred between the near-surface layer and layers above it. Likewise, to transport the Houston plume 75 km north or northeast of Houston would have required ∼7 h. Aloft, however, the meridional winds near the bottom of the jet, at around 350 m agl, had an average magnitude of about 11 m s−1, requiring less than 2 h for air masses to travel the same distances.

[30] Meanwhile, at the ship location near Houston, HRDL line-of-sight velocity estimates in the lowest 200–300 m agl demonstrate an average horizontal velocity variance during the jet of σh2 = 0.2 m2 s−2 (higher values of 0.5 m2 s−2 were not uncommon during this study, i.e., see Figures 3 and 9). Minimum range restrictions mean that HRDL data cannot be used to estimate σw2 below 200 m, but according to Banta et al. [2006] and references therein, vertical variances near the surface during a LLJ are on the order of less than half the horizontal variances, or in this case σw2 < 0.1 m2 s−2 (σw < 0.32 m s−1). This means that in an idealized model a packet of polluted air at the surface could have reached the 350 m height h in as short a time as h/σw > 18 min, and likewise, a packet of marine air in the jet could reach the surface to dilute surface concentrations in this amount of time. In reality, turbulent conditions between the surface and 350 m altitude would exist as far upwind as the jet begins, mixing the relatively clean air aloft with near-surface air all along the trajectory from the Gulf of Mexico.

[31] In contrast, on 2 September the maximum speed inside the jet was ∼5 m s−1 (bottom row of Figure 3), requiring 4.2 h of constant southerly flow to potentially advect marine air the 75 km from the Gulf to the ship's position in the Houston Ship Channel. The period of southerly jet flow, however, lasted for only a few hours and the indication from back trajectories was that the flow originally came from offshore near Beaumont rather than from out in the Gulf of Mexico. This flow was relatively clean, however, compared to the concentration of near-surface emitted pollutants that would build up in this area in the absence of transport and mixing. During the jet, the average σw2 in the lowest 200 m on this and similar nights would have been less than 0.04 m2 s−2 (0.2 m s−1). Again, in an idealized situation, a minimum of 17 min would have been required to mix surface pollutants up to the 200 m level of the highest winds inside the shallower jet.

[32] In either case, reductions in surface concentrations at night did not occur only by transport away of surface pollutants and advection of cleaner air in the lowest 100–200 m (where wind speeds were typically ∼3 m s−1 on both nights) but also by the combination of advection of relatively cleaner air in the jet aloft, turbulent mixing of that air toward the surface, and turbulent mixing of pollutants away from the surface.

5. Other Days in the Study

[33] Using the two example cases of 27 August and 2 September we have looked at the different types of atmospheric boundary layer kinematics related to the southerly versus northerly/northeasterly flow and the corresponding strong versus weak onshore nocturnal LLJs near Houston. We now expand the study to look at all times when the ship was taking data near Houston at night, during TexAQS 2006 between 1 August and 11 September 2006. These 18 nights are listed in Table 1 along with the peak speed and height of the observed nocturnal jet, the peak 1 h-averaged ozone measurement on the ship for the following afternoon, as well as the average of all the 1 h peak HGB CAMS ozone measurements for that afternoon. Of the nine days during the study period when HGB reported an exceedance of the ozone standard(s), the ship was near Houston or Galveston during the previous night for five of them. In the afternoons, the ship was near Houston around the time of each exceedance for seven of the nine exceedance days.

Table 1. Low-Level Jet and Ozone Measurements for All UTC Days During the Ship-Based Portion of the 2006 Texas Air Quality Experiment When the Ship Was Near Houston or Galveston Overnighta
DateDay of WeekShip Location at NightHRDL Peak Jet Speed (m s−1)Wind Direction at Jet Peak (degrees)Height of Peak Speed (m)Jet ClassificationPeak Ship 1 h O3 Near Houston (ppbv)Average of All 1 h Peak O3 HGB CAMS (ppbv)
  • a

    UTC day, day from 0000 to 2400; HRDL, high-resolution Doppler lidar; HGB, Houston-Galveston-Brazoria; CAMS, continuous ambient monitoring stations; HSC, Houston Ship Channel; BC, Barbours Cut; NC, not classified.

  • b

    Listed by the Texas Commission on Environmental Quality as a day on which one or more of the ozone standards were exceeded.

3 AugThurs.HSC9.7160730Strong3136
5 AugSat.BC7.7140345NC5467
11 AugFri.HSC10.2222830Strong4139
12 AugSat.HSC11.2216625Strong2532
13 AugSun.BC10.0195545Strong2627
14 AugMon.BC9.1197706Strong3439
15 AugTues.BC9.2224310NC5461
16 AugWed.Offshore Galveston6.8237214Weak7478b
17 AugThurs.BC6.4128154Weak112101b
26 AugSat.BC11.1198809Strong3125
27 AugSun.BC12.72171001Strong2928
28 AugMon.HSC11.1214830Strong 29
31 AugThurs.BC7.3/4.32113140Weak7996b
1 SepFri.Offshore Galveston4.1185154Weak169113b
2 SepSat.BC5.8223244Weak8774
7 SepThurs.HSC7.3206113Weak11295b
8 SepFri.BC5.4144184Weak9081
10 SepSat.BC7.5123330Weak2345

[34] In Table 1, the 1 min resolution ozone data from R/V Brown are smoothed by a 1 h running average for comparison to the hourly averaged CAMS data. Because the ship is often moving, longer averaging is avoided (a crossing of Galveston Bay typically required 2 h).

[35] Most of the days in this study exhibited some form of an onshore nocturnal low-level jet, with a period of southerly flow within or underneath the larger-scale gradient wind. On several days the southerly synoptic flow resulted in nocturnal LLJs that were strong (>9 m s−1) and deep, such as those observed in the 27 August example. Those days, classified as “Strong” in Table 1, include 3 August, 11–14 August, and 26–28 August. The HRDL wind profiles from all of these nights exhibited a clear jet shape with the strong peak jet speed between 0000 and 1200 UTC (1800 to 0600 CST). Although all of the strong LLJs observed by HRDL during TexAQS 2006 occurred during August, a number of days in August exhibited northerly and northeasterly flow aloft, with weaker jets near the surface, and by September most of the southerly jet-like events observed with HRDL were weak, shallow, and more likely to give way to a land breeze or offshore flow. Nights with HRDL wind speed profiles similar to the weak jet example of 2 September include 16–17 August, 31 August, 1 September, 7–8 September and 10 September. These jets are classified as “weak.” The weak jet nights of 31 August, 1 September, 3 September, and 7 September corresponded to the postfrontal conditions described by Rappenglück et al. [2008], but the weak jet nights of 16 and 17 August corresponded to strong easterly flow associated with a high-pressure system centered over the southeastern United States.

[36] The wind speed in the nocturnal jet on 15 August peaked at 9.1 m s−1 but the height of this peak was only 310 m, which was relatively shallow for this study. Profiles from that night (not shown) reveal stagnation at around 500 m, similar to that seen in the 2 September example. On the other hand, 5 August did not display a well-defined nocturnal LLJ, but the nocturnal BL demonstrated weak mixing and offshore flow by morning. Because these two dates exhibit a mixture of the two types of conditions with corresponding intermediate ozone levels, they are rated “NC” for “not classified” and are excluded from the portion of this analysis that contrasts the two types of day.

[37] To compare boundary layer characteristics under the two types of flow it is helpful to look at diurnal averages of BL parameters including mixing heights (MH), horizontal velocity variance, Ox, and air temperature (Figure 6). Although MH were provided in Figures 4c and 5c for the strong and weak jet examples, the variability in MH on the weak jet night can make comparison difficult. Average diurnal cycles in MH for the two different types of night are shown in Figure 6a. The mean nocturnal MH remained higher on strong jet nights (∼230 m between 0500 and 1000 UTC) than on weak jet nights (∼130 m) by a factor of 1.8. The differences were smaller shortly after sunset, when MH were still dominated by leftover convection and again shortly before sunrise, because the jet typically gave way to the land breeze between 1000 and 1100 UTC on weak jet nights [see Day et al., 2009]. During the land breezes, the shear at ∼200 m increased the MH to approximately that level. On the strong jet nights, the jet lasted until shortly after sunrise when surface heating initiated formation of the convective BL. On the weak jet nights, cooler nocturnal temperatures (Figure 6d) associated with the more stable boundary layers and offshore flow correspond to delayed formation of the convective boundary layer compared to the warmer strong jet nights. Figure 6b shows the average horizontal velocity variance (σh2) measured in the lowest 200 m, demonstrating the weaker average turbulence on the weak jet nights. The variance and MH are more similar during the convective afternoon periods. The plot in Figure 6c shows the corresponding diurnal cycles in ship-measured Ox concentrations for the different types of flow. In addition to the transport effects of emissions that reached the top of the strong jet driven BL, the turbulence and MH differences had strong implications on surface concentrations of late afternoon, nighttime, and early morning emitted emissions when MH were generally lower [Gilman et al., 2009]. MH differences were also enhanced after ∼2200 UTC when the sea breeze was fully established. On days prior to weak LLJ nights, the sea breeze pushed under the synoptic winds resulting in a capping of the BL at around 500 m. In contrast, on strong LLJ days the speed and direction of the bay/sea breeze did not contrast greatly with synoptic conditions so even after onset of the sea breeze, which tended to reduce the local MH slightly around 2000 UTC, convective mixing could still dominate the MH and the mixing layer was not capped by directional shear.

Figure 6.

(a) Average HRDL-measured mixing height versus time of day, calculated using a 2 h sliding average for days that started with a strong nocturnal LLJ (dark line) versus those that started with a weak or no jet (light, dashed line). (b) Average HRDL-measured horizontal velocity variance (σh2) in the lowest 200 m, (c) average ship-measured Ox, and (d) average ship-measured temperatures for those same days. Shaded areas indicate the variance on each of the averages.

6. Relationship Between Nocturnal Boundary Layer Conditions and Air Quality

[38] Having characterized the nocturnal BL conditions (including the strength and height of the LLJ, strength of turbulence, and mixing heights) for two different kinds of flow, with a 24 h example of each type of day, and average diurnal cycles for those days with similar flow, we next present direct comparisons of those measurements to concurrent O3 and Ox concentrations measured both on the ship and throughout the HGB region. In order to compare two-dimensional (time and altitude) wind and turbulence profiles to one-dimensional (time) in situ ship-surface chemistry measurements, the HRDL profiles are reduced to time series of wind speed, wind direction, and velocity variance in different altitude bins at a 15 min sampling interval. The speed and direction data were used to calculate the northerly/southerly (meridional) wind component, V, a proxy for offshore/onshore flow in the near-Houston ship locations. The Ox data were then averaged for each 15 min time interval, centered on the wind profile time, and the results compared to the time series of HRDL data from each altitude bin.

6.1. Nocturnal Ox Versus Wind Speed

[39] Figure 7a contains a scatterplot of nocturnal (0200 to 1200 UTC) 15 min averaged Ox concentration versus concurrent meridional wind speed V measured between the surface and 20 m agl. Using V helps to separate the effects of wind speed amplitude from onshore versus offshore flow [Nielsen-Gammon et al., 2005b]. Markers indicate whether or not a strong jet was observed in the HRDL profiles during the night: dots indicate data from strong LLJ nights and x's indicate data from weak jet nights. A clear relationship between Ox levels and surface mean wind speeds is not evident.

Figure 7.

Ship-measured nocturnal surface Ox (15 min average) versus concurrent HRDL-measured meridional mean wind speed average (a) between 0 and 20 m above ground level (agl) and (b) between 650 and 700 m agl.

[40] Plotting surface Ox versus average V from a higher altitude bin (i.e., 650–700 m agl), however, demonstrates a clearer relationship (Figure 7b). The 650–700 m agl altitude bin was chosen because it represents an average height range for the vertical center of most of the observed strong southerly jets and tends to be in the stagnant layer above the maximum height of the weak jets, thus providing good contrast between the two. Values of V from altitude bins below and above this range were also compared with Ox concentrations on each type of night but did not show as large a difference. Wind speed at the top of the mixing layer may also be used with similar results but this requires estimation of mixing height (available with the HRDL data but not with most profiler data). For simplicity, and for eventual extension of the analysis to radar wind profiler data, the 650–700 m range bin is used in this study.

[41] Relative to weak jet nights, the relationship of winds to reduced ozone concentrations on strong jet nights is clarified when the Ox concentrations are compared to winds aloft. For both altitude bins the Ox concentrations tended to fall off with increasing amplitude of V, from both northerly and southerly directions; however this is difficult to observe with the negative (northerly) nocturnal V as they were never as strong during this study as the southerly winds in these altitude bins.

[42] Figure 7b suggests that on nights with strong positive meridional wind speeds aloft the low Ox concentrations were controlled by the synoptically driven ventilation with clean marine air (see Langford et al. [2010], for a similar conclusion) whereas at lower wind speeds, when air parcels have a longer residence time over source areas, higher precursor concentrations are possible. At higher positive values of V (i.e., 9–12 m s−1), Ox levels did not continue to decrease but instead leveled out around 15–22 ppbv. These values correspond to the Ox (and ozone) levels of 15–28 ppbv measured in the Gulf of Mexico under southerly flow conditions during TexAQS 2006 implying that the air mass at the surface near Houston during these nights was largely made up of relatively clean marine air. Given the long (∼70 km) distance from the Gulf of Mexico to Houston, and low near surface wind speeds as well as the presence of local surface-based land and ship emissions at night throughout the region [Williams et al., 2009], it seems likely that most of this relatively clean marine air was transported in the jet flow aloft and mixed down to the surface via shear-induced turbulence as previously discussed in section 4.3. The low meridional wind speeds in Figure 7b, which indicate little to no ventilation with marine air, corresponded to higher variability in Ox concentrations. Insight into this variability requires knowledge of a number of other factors including regular and episodic high emissions of NOx and volatile organic compounds in the Houston/Galveston region [Ryerson et al., 2003; Murphy and Allen, 2005; Webster et al., 2007; Gilman et al., 2009], background O3 levels [Langford et al., 2009; Rappenglück et al., 2008], dynamic stability and convective venting [Langford et al., 2010], deposition, etc., as well as variability in wind direction. In addition, easterly or westerly winds may contain a small meridional component even if the wind speeds, and thus potential ventilation, may be higher.

[43] To further elucidate the differences between the wind speeds during these two types of flow regime (and thus jet speeds), Figure 8 contains distributions of average V in the lowest 20 m (Figure 8a) and the average V from 650 to 700 m (Figure 8b). These distributions overlap in the near-surface winds, but show no overlap at the 650–700 m range for 0.5 m s−1 bin widths. The plots in Figure 7a and Figure 8a demonstrate how the meridional surface winds on strong jet nights and weak jet nights (or even nights without a jet) can be quite similar, even though the boundary layer dynamics and stability on these two types of night are quite different. Differences in the distributions of the winds aloft, however, is much greater, and serves as a better indicator of the synoptic flow conditions and Ox concentrations.

Figure 8.

Distributions of the nocturnal wind speed data used in the plots in Figure 7. Shading indicates the same strong and weak LLJ categorizations.

[44] Several other chemical species and meteorological parameters measured on the ship during TexAQS 2006, including volatile organic compounds, sulfates, temperature, relative humidity, surface solar radiation, etc., similarly exhibited distinct differences in their average values between days with strong and weak jets, however we have limited the comparison of the HRDL data in this paper to O3 and Ox measurements for simplicity.

6.2. Nocturnal Ox Versus Turbulence

[45] The velocity variance profiles in Figure 4c and Figure 5c demonstrate not only differences in mixing height on strong jet and weak jet nights, but also differences in the strength of the boundary layer turbulence. Horizontal variance profiles from all times the ship was near Houston, averaged over the lowest 200 m, form a variance parameter that, under stable conditions, is related to average turbulent kinetic energy (TKE) as explained by Banta et al. [2006] and Pichugina et al. [2008]. Concurrent nocturnal Ox measurements are plotted against this parameter in Figure 9, with symbols used to denote the categorization of the samples according to the type of jet observed.

Figure 9.

Ship-measured Ox (15 min average) versus concurrent HRDL-measured horizontal velocity variance, σh2, averaged over the lowest 200 m, from 0200 to 1200 UTC (2000 to 0600 CST). Dots indicate data acquired during strong nocturnal LLJ; crosses indicate data acquired during nights with a weak LLJ.

[46] On nights when a strong LLJ was present (dots), shear-induced turbulence (x axis) reached higher values. On weak LLJ nights, however, the tendency was toward lower variance values, and generally higher Ox concentrations. The cluster of low σh2 values between 0.04 and 0.1 m2 s−2 correspond to Ox levels that never fell below 30 ppbv suggesting that this weak turbulence, which corresponded to shallower mixing heights (not shown) resulted in less upward transport, little downward mixing of air masses aloft, and less opportunity for deposition. This trapping of emissions, combined with very little ventilation from the low-surface wind speeds, resulted in relatively higher precursor concentrations from nocturnal emissions [Gilman et al., 2009]. Figure 9 demonstrates that on the weak jet nights the tendency was toward relatively lower Ox levels with increasingly stronger turbulence, while ventilation appears to dominate the effects of turbulence on the strong jet nights. The overlap of the turbulence values in these two categories, however, demonstrates that mixing is not the only factor in determination of Ox concentrations: other factors such as emissions and the transport of relatively clean air aloft must also be considered.

6.3. Nocturnal and Daytime Winds Versus Maximum Daytime Ozone

[47] The discussion in section 6.2 demonstrated observed effects of nocturnal wind speeds aloft on concurrent surface ozone concentrations and the difference between nights with strong versus weak LLJs. How these winds correlate to air quality the following afternoon is now addressed by expanding the comparisons to Ox for the daytime periods, and then by comparing nocturnal jet characteristics to the peak hourly averaged ozone value measured the following afternoon, either at the ship, or averaged throughout the HGB region.

[48] Figure 10a contains a plot of ship-based Ox measurements versus HRDL-measured meridional speed V in the 650–700 m bin for those daytime periods following nocturnal LLJ's. These data represent the daytime (i.e., >1200 UTC) version of those plotted in Figure 7, where the dots indicate data acquired during the daytime following a strong nocturnal LLJ and x's indicate those acquired after a weak LLJ. This plot demonstrates that the wind speeds during the day following a strong LLJ tended to also be higher and more likely onshore, on average, than wind speeds following a weak LLJ. Conversely, weak jets tended to exist under northerly/northeasterly synoptic conditions, which lead to weaker afternoon flow, sometimes resulting in stagnation before the sea breeze initiation. The increased overall scatter in the daytime plots can again be attributed to several factors including emissions, ozone production rates, dynamics (including the amount of convective activity in the area [see Langford et al., 2010]), background O3 levels, and the vector average of mean daytime boundary layer winds as discussed by Banta et al. [2009]. Overall, however, greater variability in Ox was observed on the weaker wind days than on the days ventilated by strong southerly synoptic winds following a strong nocturnal jet. As was shown for nocturnal winds in Figure 8, Figure 10b shows how the distribution of these daytime wind speeds, categorized by the strength of the jet from the previous night again demonstrates distinct separation for the two flow regimes.

Figure 10.

(a) Ship-measured surface Ox (15 min average) versus concurrent HRDL-measured average horizontal mean wind speed between 650 and 700 m agl for all days after a nocturnal LLJ when the ship was close to Houston. (b) Distributions of the daytime wind speed data from Figure 10a. Shading indicates the same strong and weak LLJ categorizations.

[49] With the objective of finding a simple relationship between local nocturnal winds and next-afternoon Houston ozone concentrations during TexAQS 2006, the daily maxima of each are compared. Figure 11 contains a plot of the maximum 1 h-averaged afternoon ozone measured on the ship (gray filled circles) and the average of all maximum 1 h CAMS ozone measurements from the HGB monitoring network (open diamonds) versus the maximum speed of the nocturnal jet Vmax, (m s−1), measured by HRDL the previous night, for each period that the ship was near Houston. The Houston area measurements (empty diamonds) are the average of the peak 1 h values from all ozone measurements in the HGB CAMS network, covering an area of ∼48,000 km2. Note that Vmax is the absolute maximum speed of the jet, chosen from all altitudes of all the nocturnal profiles, and is not the meridional component in a constant altitude bin that was used in the comparisons of 15 min data shown in Figures 7 and 10. Although conditions that generate a strong LLJ tend to be synoptic in size, with pressure gradients extending from mesoscale to the synoptic scales, it was still possible for the nocturnal winds and conditions measured at the ship to be somewhat uncorrelated from peak ozone measurements at distant HGB CAMS locations (see Langford et al. [2009] for an explanation of this). Thus the average of all the 1 h averaged CAMS peaks for each day is used here. All 18 d during the study when the ship was near Houston at night, including 5 August and 15 August, are included.

Figure 11.

Peak 1 h-averaged afternoon ship ozone (ppbv) as measured from the ship (diamonds) and the average of all CAMS peak 1 h afternoon ozone (circles) versus maximum speed (m s−1) of the “jet” measured the preceding night. Ship-measured values may be lower or higher depending on ship location relative to the Houston plume but are often higher because the gray circles represent a regional average of the hourly peaks in the CAMS data. The highest ship-measured 1 h average value of 169 ppbv was observed on 1 September as the ship was crossing Galveston Bay, where no CAMS data are available.

[50] The peak ozone values also demonstrate a correlation with the height of the maximum speed of the jet Hmax (not shown). By multiplying these values to create a jet “area parameter,” Ajet = Vmax × Hmax, (m2 s−1), we form a more thorough description of the jet, in its form near Houston, that shows better correlation to peak Houston ozone measurements. The daytime peak ozone values demonstrate an inverse relationship to this area parameter as shown in Figure 12. This shows that the synoptic conditions that lead to formation of the strong nocturnal jet tended to have persistence, affecting ventilation and thus ozone concentrations the following afternoon. Stronger and deeper nocturnal jets correlate to lower peak ozone the following day. There are multiple, but not necessarily independent, reasons for this correlation that include synoptic conditions, background levels, insolation, and convection [Langford et al., 2010]. As previously discussed, however, the synoptically influenced strong LLJ is a mechanism by which air polluted with residual and nocturnally emitted ozone precursors was dispersed and removed during the night and replaced with relatively clean marine air. Given the common presence of the nocturnal jets, in their synoptically affected (weak or strong) form, they may be used as a diagnostic of the synoptic conditions and thus of the next day's ozone concentrations in Houston.

Figure 12.

Ship-measured peak 1 h average ozone (diamonds) and average of all CAMS peak 1 h ozone (circles) versus the inverse of the nocturnal jet area parameter (Ajet) measured the previous night.

[51] The data show a correlation of 0.97 between Ajet−1 and ship-based 1 h averaged peak ozone (measured the following afternoon) and a correlation of 0.95 between Ajet−1 and the average of all the peak 1 h CAMS ozone measurements in the HGB area, measured the afternoon after the jet. The ∼23 ppbv offset for the line fit for the ship-based data (dashed line in Figure 12) corresponds to measured background levels in the Gulf of Mexico (also see Figure 7 and Figure 10). Even if the peak ship and average peak CAMS values were low on a given day, however, ozone values at the edge or outside of the Houston network, downwind of Houston, may have been higher (as discussed by C. J. Senff et. al., Ozone flux and production downwind of Houston and Dallas, submitted to Journal of Geophysical Research, 2010).

[52] Banta et al. [2003] showed that the LLJ speed and shear information, along with stability values via potential temperature measurements, are necessary to accurately represent vertical turbulent mixing in numerical weather prediction (NWP) models. Likewise, Figure 12 demonstrates that LLJ parameters, including the height of the maximum speed of the jet, may also serve a predictive function for the Houston area when integrated with regular pollution forecasting methods.

7. Summary

[53] During TexAQS 2006, high-resolution (15 min temporal and 5–30 m vertical) profiles from the high-resolution Doppler lidar provided observations of mixing and transport under various flow regimes in the Houston coastal regions. This study demonstrated how the ground-relative speed and height of the nocturnal low-level jet, determined by the synoptic conditions and closely coupled to the smaller-scale boundary layer dynamics, related to pollution concentrations near Houston. With 24 h, 15 min updated boundary layer kinematics profiles that extend from the surface to ∼2 km agl altitude or more, HRDL observations demonstrate how the jets typically evolved from the afternoon southeasterly sea breeze and grew in altitude and wind speed as they veered toward the southwest throughout the night. Synoptic conditions determined whether the jet reached high wind speeds (9–12 m s−1) and altitudes (600–1000 m) and whether they advected relatively clean marine air from the Gulf of Mexico into the region to clear out Houston pollution overnight. Under strong southerly flow regimes observed during the study, the near-surface shear induced by the strong jets generated mechanical turbulence capable of mixing marine air down to the surface. Likewise, ozone and ozone precursors generated during the night (or left over from the previous day) were mixed up into the jet where they were dispersed by the high wind speeds, leaving less ozone and ozone precursors aloft to be mixed down the following daytime. In addition, persistence in synoptic conditions that lead to the strong jets meant stronger and more southerly horizontal winds in the afternoons, thus inhibiting buildup of ozone in the measurement area. Alternatively, northerly or easterly synoptic flow conditions, which typically carried higher background levels of ozone, resulted in relatively weaker and shallower nocturnal jets that veered through southerly to westerly and then northerly, and finally gave way to a shallow land breeze/offshore flow by early morning. Winds just above the shallow jet, but inside the previous afternoon's mixing height, during these periods were typically very low, indicating little to no transport of pollutants in that layer during the night. Strong shallow westerly jets that can correspond to higher ozone concentrations, like those observed during a high ozone episode in late August 2000 [Banta et al., 2005] were not observed using HRDL during this study period.

[54] Nocturnal horizontal mean wind speeds at the surface during the strong jets were similar to those during the weak jets, so that comparison of meridional winds at higher altitudes near the core of the stronger jets demonstrated a better dilution relationship with corresponding in situ surface ozone or Ox levels than did those near the surface. Turbulence profiles measured using HRDL during both strong and weak jet nights reinforced the relationships between wind speeds aloft and surface concentrations by showing the turbulent connections between the surface and the winds aloft inside the jet. Finally, the nocturnal LLJ as observed by HRDL was strongly correlated with peak afternoon ozone concentrations on the ship and in the Houston area the following afternoon via an inverse relationship to the “area parameter” of the jet, indicating that the strength and vertical extent of the nocturnal LLJ are good indicators of the larger-scale synoptic situation. The stronger jet nights were consistently followed by a low-ozone (“clean air”) day in the HGB area whereas weaker jet nights were typically followed by high ozone days in the HGB area. For nights when a weak jet was observed, the lack of ventilation must continue to be considered together with additional meteorological, emissions, and ozone production rate information to predict afternoon ozone concentrations throughout the region. Observation of a strong nocturnal jet however, may be used as an indicator of the strongly ventilating synoptic conditions important for air quality planning.


[55] Funding for this research was provided by the Texas Commission on Environmental Quality (TCEQ) and by NOAA's Health of Atmosphere program. The authors would like to thank Scott Sandberg for his help in preparing the HRDL instrument for the TexAQS 2006 experiment, Tim Bates for serving as the experiment's chief scientist, and the crew of the Ronald H. Brown for making the experiment possible. We are also grateful to Dan Wolfe and Ludovic Bariteau for the ship-based air-temperature data and to John Nielsen-Gammon for his valuable feedback and suggestions for this paper.