Abstract

Typical diurnal wind patterns and their relationship to transport of atmospheric aerosol in the Columbia River gorge of Oregon and Washington are addressed in this paper. The measurement program included measurements of light scattering by particles (bsp) with nephelometers, and wind speed and direction, temperature, and relative humidity at seven locations in the gorge. Winds are shown to respond to along-gorge pressure gradients, and five common patterns were identified: strong, moderate, and light westerly (west to east), light easterly, and winter easterly. The strong westerly and winter easterly patterns were the most common summer and winter patterns, respectively, and represented strong gap flow. The light westerly and light easterly patterns occurred most frequently in spring and autumn transition periods. Winter easterly had the highest light scattering and indicated sources east of the gorge mainly responsible for haze. During summer, as westerly winds increased diurnally, a pulse of hazy air from the Portland, Oregon, metropolitan area is transported eastward into the gorge, arriving later with distance into the gorge. During light easterly flow impacts to haze from the city of The Dalles, Oregon, are noted as the wind shifts direction diurnally.

1. Introduction

The Columbia River gorge is a narrow gap in the Cascade Mountains of Washington and Oregon. Cut by the flow of the Columbia River, the gorge is approximately 190 km in length, 5 km wide (at river level), and 1000 m deep and is generally oriented east–west. The elevations at river level are less than 100 m above mean sea level (MSL). The Portland, Oregon–Vancouver, Washington, metropolitan area, with a population of approximately 2 million, is located at the western terminus of the gorge. The western portion of the gorge is characterized by a maritime air mass while the eastern gorge is semiarid. Annual average precipitation at Troutdale, Oregon, at the western end of the gorge, is 1135 mm, while at The Dalles, Oregon, at the eastern end of the gorge, annual average precipitation averages 360 mm.

The Columbia River gorge has outstanding natural features including steep forested mountainsides with numerous high waterfalls. To help protect this area, the U.S. Congress passed legislation titled the “Columbia River Gorge National Scenic Area Act.” Among other reasons, the act was passed to protect the scenic and natural resources of the gorge. From 2003 to 2005 measurements were made to help understand the causes of haze in the Columbia River gorge. Results from the first portion of this study [The Columbia River Gorge Haze Gradient Study (Green et al. 2006a)] are reported upon here. The field study, described in more detail in the next section, included measurements of light scattering by particles (bsp) at eight sites along the gorge, and wind speed and direction, temperature, and relative humidity at seven of these sites. The measurements provided for a horizontal gradient in bsp and nearly vertical gradient in two locations.

Numerous urban and industrial sources in the region emit gaseous or particulate air pollution that may contribute to light scattering in the gorge. These include the Portland, Oregon–Vancouver, Washington, metropolitan area; numerous pulp and paper mills, mainly west of the gorge; coal-fired electric power plants (the largest regional source of sulfur dioxide is the Boardman plant east of the gorge); and small cities in the gorge such as Hood River and The Dalles, Oregon. Large livestock operations (dairies and feedlots) are located east of the gorge and emit large quantities of ammonia, which is significant in secondary aerosol formation, particularly in winter. The industrial sources emit substantial quantities of primary particulate matter, which scatters light, as well as significant quantities of sulfur dioxide and nitrogen oxide, which may react with other compounds to form light-scattering particles (sulfate and nitrate particles). Biogenic volatile organic compound (VOC) emissions from the heavily forested area also are likely to contribute to light scattering.

The measurements provided hourly averaged data for the period of July 2003 through February 2005. To summarize this large amount of data, we used cluster analysis to form groups of days with similar spatial and diurnal wind patterns. We computed averaged pressure fields for these groups of days to help understand the flow patterns and summarized the spatial and diurnal patterns in light scattering for each group of days. The analysis of transport of light-scattering aerosols was the first step in understanding source areas causing haze in the gorge. Additional aerosol measurements and data analysis, including receptor modeling, which are not discussed in this paper, were used to attribute haze in the gorge to source areas and source types (Green et al. 2006b; Green and Xu 2007).

a. Gap winds in the Columbia River gorge

Gap winds are the result of a pressure gradient located along a break or gap in a mountain barrier. Confined by the terrain, gap winds are highly ageostrophic, blowing parallel to the gap from high to low pressure. Reed (1931) referred to winds in the Strait of Juan de Fuca as gap winds. Cameron (1931) described strong easterly gap winds in the Columbia River gorge. The climatology of gap winds in the gorge has recently been described in detail by Sharp and Mass (2004). In summer, northward migration of the Pacific high creates higher pressure offshore. Heating of the Columbia River basin creates a thermal low pressure east of the gorge. Thus a substantial pressure gradient exists through the gorge, causing consistently westerly flow through the gorge in summer that is regularly exploited by wind surfers near Hood River. In winter, conditions are more variable but easterly gap flow is common. With synoptic-scale high pressure areas over the interior western United States and Pacific low pressure systems typically to the west, east to west pressure gradients are common. The high pressure to the east is often enhanced by the development of “cold pools” in the Columbia Basin (Whiteman et al. 2001; Zhong et al. 2001). These cold pools formed by strong radiational cooling are associated with light winds, moisture, and a buildup of pollution in the Columbia Basin. In the gorge, air starts accelerating in response to the along-gorge pressure gradient and becomes generally stronger as it traverses the gorge from east to west. Most acceleration may occur in certain areas (Sharp and Mass 2002), but gorge wind data are not complete enough to well define the wind patterns.

2. Measurements

Table 1 lists the monitoring sites, their latitude, longitude, elevation (MSL), and elevation above the Columbia River. The sites are briefly described below:

  1. Steigerwald is a river-level site (10 m above river) at the mouth of the gorge.

  2. Mount Zion is a somewhat elevated site (about 210 m above river) near west end of the gorge and close to Steigerwald.

  3. Strunk Road is an elevated site (365 m above river) close to Mount Zion, horizontally farther from the gorge. Strunk Road, Mount Zion, and Steigerwald provide essentially a vertical profile at the west end of the gorge.

  4. Bonneville Dam is a river-level site (2 m above river) in the heart of the gorge.

  5. Memaloose State Park is a river-level site (8 m above river) between Hood River and The Dalles.

  6. Sevenmile Hill is an elevated site (540 m above river) west of The Dalles horizontally close to the river. It has good exposure to higher-level flows up and down the gorge.

  7. Wishram is a slightly elevated (125 m above river) site close to river near to and east of The Dalles.

  8. Towal Road is a near-river-level (70 m above) site east of Wishram.

Table 1.

Site name, latitude, longitude, elevation, and approximate elevation above the Columbia River for each site.

Site name, latitude, longitude, elevation, and approximate elevation above the Columbia River for each site.
Site name, latitude, longitude, elevation, and approximate elevation above the Columbia River for each site.

Figure 1 shows the location of the monitoring sites. All sites measured particle light scattering (bsp) using Radiance Research nephelometer model M903. All sites except Memaloose had surface meteorological measurements of wind speed and direction, temperature, and relative humidity (RH). Wind speed and direction were measured using R. M. Young model 09305 systems. The temperature probes were R. M. Young model 41342VF. Relative humidity was measured with Rotronic MP100H/MP400H probes.

Fig. 1.

Site locations. The Portland metropolitan area is located at the center-left edge of the figure.

Fig. 1.

Site locations. The Portland metropolitan area is located at the center-left edge of the figure.

At high relative humidity (over about 70%) particles containing nitrate and sulfate grow rapidly with increased humidity due to uptake of water. The particles scatter much more light due to the water uptake. Particles stay in solution (the deliquescence effect) until humidity is reduced to less than 50%. To help minimize the effects of varying RH levels among sites, the nephelometers were set to heat the sample air stream as needed to maintain an RH of not more than 50%. This allows for a good comparison of concentrations of light-scattering aerosols and determination of their horizontal and vertical gradients within the gorge. Actual light scattering (haze) is higher under high RH conditions than measured by the nephelometers.

3. Data analysis methodology

a. Cluster analysis to group days with similar wind patterns

To organize the approximately 600-day study period (1 July 2003 to 28 February 2005), a cluster analysis was done to obtain a small number of clusters for which common diurnal wind patterns were observed. We hypothesized that days with similar winds at each monitoring site, including their diurnal variation, should be similarly affected by transport from sources. That is, days grouped based on similarity of winds should have similarities in diurnal patterns of light scattering (bsp).

A potentially important factor not considered in the clustering was the occurrence of precipitation that could cause washout of particles from the atmosphere. Precipitation was considered when averaging light scattering for days with similar winds (i.e., by removing rainy days), as will be described later. Also not considered explicitly were the differences that may result due to reactivity rates, dust entrainment, etc. that can vary seasonally under the same wind patterns. As it turns out wind patterns, and thus the wind clusters, are strongly seasonally dependent, as will be demonstrated shortly.

The K-means method of cluster analysis was used with five clusters chosen to represent the main wind patterns. It is acknowledged that grouping all days into five patterns is not a thorough classification that includes all flow patterns occurring in the gorge. The objective here is to represent the major flow patterns and interpret the average spatial and diurnal patterns of light scattering under the different average wind conditions. Mainly we wish to determine from the wind and light-scattering patterns if visibility impairment is mainly due to sources west of the gorge, east of the gorge, or within the gorge. Individual days may differ significantly from the cluster mean in terms of diurnal patterns. We argue that the results show that the clusters are meaningful.

Ideally, we would have used wind data from all seven nephelometer sites with meteorological data for the cluster analysis. If any sites used in the cluster analysis have one or more hours of data missing for a day, that day cannot be clustered. Using only the three sites with most complete data (Mount Zion and Strunk Road in the western gorge, and Wishram in the eastern gorge), we were able to cluster 563 out of 609 possible days. Using all seven sites with meteorological and light-scattering data, we would have been able to cluster only 332 days. A sensitivity analysis was done to compare clusters using all sites versus the three most complete sites. Qualitatively the results were similar, so results are presented for all sites with the clustering based on the three most complete sites. This allows us to include nearly all days for comparing typical diurnal patterns.

Each of the monitoring sites showed a bimodal wind distribution oriented along the local axis of the gorge (basically westerly or easterly) (Fig. 2, sites ordered from west to east). The direction of each mode varies somewhat from site to site depending upon the local direction of the river path cut through the Cascade Range. For each mode of the bimodal distribution at each site, the direction of peak frequency was determined. For each hour of the day the component of the wind in the up-gorge (westerly, upriver) or down-gorge (easterly, downriver) direction (whichever the wind direction was closest to) was computed and was assigned a positive sign for up-gorge flow and a negative sign for down-gorge flow. This simplifies the analysis by turning the vector wind into a scalar representation of it. Table 2 shows the frequency of westerly or easterly winds at each site averaged over the 20-month study period.

Fig. 2.

Frequency distribution of wind direction by site. The x axis is direction from which wind is blowing (meteorological convention); y axis is number of hours with wind from each 1° increment in direction. Period of record is 1 Jul 2003–28 Feb 2005.

Fig. 2.

Frequency distribution of wind direction by site. The x axis is direction from which wind is blowing (meteorological convention); y axis is number of hours with wind from each 1° increment in direction. Period of record is 1 Jul 2003–28 Feb 2005.

Table 2.

Percentage of hours with wind direction westerly and easterly by site.

Percentage of hours with wind direction westerly and easterly by site.
Percentage of hours with wind direction westerly and easterly by site.

The input data to the cluster analysis were the hourly westerly wind component for each site used (Strunk Road and Mount Zion in the western gorge, and Wishram in the eastern gorge) for each hour for each day between July 2003 and February 2005. Each row in the input array is for one day and thus has 72 columns (24 h × 3 sites). The cluster analysis was carried out using the STATISTICA software. The K-means cluster analysis used here requires the number of desired clusters (k) to be specified. The program will start with k random clusters and then will move objects between those clusters with the goal to 1) minimize variability within clusters and 2) maximize variability between clusters. In other words, the similarity rules will apply maximally to the members of one cluster and minimally to members belonging to the rest of the clusters. The more clusters that are specified, the less the variation within each cluster is. However, more clusters will also result in less difference between clusters and will be more effort to characterize.

We tried five and seven clusters and interpreted the five cluster groups. The cluster analysis software computes typical diurnal wind patterns for each cluster for each site. While not used in the cluster analysis, typical winds for days in each cluster were generated for the four other nephelometer sites with meteorological data and will be presented and discussed. The typical wind patterns were reviewed to understand the wind characteristics of each pattern. Example diurnal wind patterns are shown in section 4.

b. Computation of pressure field patterns and their diurnal variation by wind pattern

To help to understand the spatial and diurnal variation associated with each wind pattern, pressure adjusted to sea level was gathered and averaged by hour for each site and pattern combination. Hourly pressure data were obtained for the period of July 2003–February 2005. Data were obtained for sites west and east of the gorge, within the gorge, and to the north and south.

The following sites were used: Astoria, Portland Hillsboro, Portland International Airport, Troutdale, The Dalles, Pendleton, Pasco, Boise, Seattle, Salem, and Eugene. A summary of the site locations is given below:

  • Astoria is a Pacific Coast site at the mouth of Columbia River,

  • Hillsboro, Portland International, and Troutdale are sites spanning the Portland area (Troutdale is near gorge exit),

  • The Dalles is an in-gorge location toward the eastern end of the gorge,

  • Pasco and Pendleton are near the east end of gorge, Pasco along Columbia River and Pendleton some distance away, and

  • Boise, Seattle, Salem, and Eugene are sites north, south, and east of the gorge.

Because of the tendency of flow from high to low pressure through the gorge, we are mainly interested in along-river pressure gradients. The pressure field patterns were used to help interpret the wind patterns.

c. Relationship of wind patterns to the particle light-scattering coefficient (bsp)

For each wind pattern, the nephelometer data were used to compute hourly average particle light-scattering coefficients (bsp) for each nephelometer site. These diurnal patterns by site were interpreted in light of the wind and pressure patterns. This interpretation provided insight into the roles of source regions in affecting light scattering. We stratified the days in each wind pattern as to whether there was precipitation in the area or not, using data from Portland International Airport (PDX) and The Dalles. Days with no precipitation were defined as those days with 0.01 in. of precipitation or less at both stations. We computed pattern average bsp at each nephelometer site for 1) all days; 2) days without precipitation; 3) days with >0.01 in. (1 in. ≈ 2.54 cm) of precipitation at PDX; and 4) days with >0.01 in. of precipitation at The Dalles. Note that some days would have precipitation at both PDX and The Dalles. Pattern average bsp are shown in section 5 along with some example diurnal patterns of bsp. The effect of precipitation on bsp is also briefly presented.

4. Wind field patterns

The cluster analysis, being purely statistical, gives no physical explanation of the results. The software gives results as clusters 1–5 with average wind components by hour for each site. To keep track of the clusters, it is helpful to assign names to them based on their characteristics. The five wind patterns were named as follows: strong westerly; moderate westerly; light westerly; light easterly; and winter easterly. Winter easterly was named as such because it occurred mainly in winter and never between May and September. Figure 3 gives the hourly westerly wind component by hour for the three sites used for the cluster analysis—this is the quantitative definition of the wind pattern—and shows the daily average westerly wind component at each site for the five wind patterns.

Fig. 3.

The hourly westerly wind component by hour for the three sites used for the cluster analysis—this is the quantitative definition of the wind pattern: (top left) Mount Zion, (top right) Strunk Road, and (bottom left) Wishram. (bottom right) The daily average westerly wind component at each site for the five wind patterns.

Fig. 3.

The hourly westerly wind component by hour for the three sites used for the cluster analysis—this is the quantitative definition of the wind pattern: (top left) Mount Zion, (top right) Strunk Road, and (bottom left) Wishram. (bottom right) The daily average westerly wind component at each site for the five wind patterns.

All sites show a consistent decrease in average westerly wind speed from strong westerly to winter easterly. Strong westerly, moderate westerly, and light westerly increase in speed with distance eastward into the gorge until Towal Road, which has lower speeds than Wishram. Winter easterly is light at the east end of the gorge and generally increases in speed toward the western gorge. At the easternmost sites, light easterly had weak net westerly flow with periods of easterly flow and a diurnal change in flow direction. At western and central gorge sites weak easterly had easterly flow all hours of the day. Table 3 shows the frequency of each wind pattern by month. March though June had data only from one year (2004); all other months had data from two years.

Table 3.

Percentage of occurrence of each wind pattern by month.

Percentage of occurrence of each wind pattern by month.
Percentage of occurrence of each wind pattern by month.

Strong westerly occurs mainly in summer, peaking in frequency in July. Moderate westerly peaks in the late summer to early fall (August–October). Light westerly occurs throughout the year, with a peak in the transition months April and October. Light easterly is another transition pattern, peaking in April and November. Winter easterly is a wintertime pattern and never occurred from May to September.

The basic wind patterns, including their relative diurnal variations, were similar for the following groups of sites: west—Strunk Road, Mount Zion, and Steigerwald; east—Wishram, Towal Road, and Sevenmile Hill. Bonneville was unique in its pattern. Figure 4 shows the diurnal wind variation for representative near-river-level sites at the west gorge (Steigerwald), central gorge (Bonneville), and eastern gorge (Wishram).

Fig. 4.

Diurnal variation in average westerly wind component by wind pattern at Steigerwald, Bonneville, and Wishram. Hour of day is Pacific standard time (PST).

Fig. 4.

Diurnal variation in average westerly wind component by wind pattern at Steigerwald, Bonneville, and Wishram. Hour of day is Pacific standard time (PST).

a. Discussion of main winter and summer patterns

Strong westerly and winter easterly represent classical gorge summer and winter patterns (strong westerly also occurred frequently in March). They also are the patterns most consistent with the gap wind description of strong flows accelerating with distance into the gorge due to east–west pressure gradients. Strong westerly shows persistent westerly flow, increasing in speed toward the east end of the gorge (strongest at Sevenmile Hill and Wishram). Winter easterly shows persistent easterly flow, increasing in speed from Towal Road to Mount Zion (with the exception of Strunk Road, which is farther from the river than Mount Zion and likely out of the main easterly flow). Both of these flows are expected to result from an acceleration along the gorge due to a pressure gradient across the gorge, as described in Sharp and Mass (2004) and elsewhere. Hourly averaged pressure for strong westerly days for selected stations is shown in Fig. 5. There is a decrease in pressure from Portland though the gorge to Pasco, on the Columbia River northeast of the gorge. Hourly averaged pressure for winter easterly days for selected stations is shown in Fig. 6. For winter easterly, there is

  1. essentially equal pressure all day at Pasco and The Dalles, and

  2. a lower, but nearly equal, pressure at Hillsboro, Portland International, and Troutdale.

Fig. 5.

Diurnal pressure pattern at Portland, Troutdale, The Dalles, and Pasco for the strong westerly wind pattern. Hour is PST.

Fig. 5.

Diurnal pressure pattern at Portland, Troutdale, The Dalles, and Pasco for the strong westerly wind pattern. Hour is PST.

Fig. 6.

Diurnal pressure pattern at Astoria, Portland, Troutdale, The Dalles, and Pasco for the winter easterly wind pattern. Hour is PST.

Fig. 6.

Diurnal pressure pattern at Astoria, Portland, Troutdale, The Dalles, and Pasco for the winter easterly wind pattern. Hour is PST.

Thus, the main pressure gradient through the gorge is entirely between The Dalles and Troutdale. These two items noted above can explain

  1. light winds at Towal Road, east of The Dalles, and

  2. strong easterly winds at Bonneville, Mount Zion, and Steigerwald.

The pressure data do not provide details regarding the pressure gradient such as where between The Dalles and Troutdale the gradient is greatest.

In winter the gradient is often reversed from summer with higher pressure east of the gorge due to a cold, synoptic-scale high pressure area and lower pressure to the west of the gorge, often due to Pacific lows.

b. Discussion of other patterns

The other wind patterns were less strongly forced by east–west pressure gradients than strong westerly and winter easterly. Light westerly and light easterly occurred mostly in spring and fall and had relatively small pressure gradients. Light easterly had light easterly flow at the western and central gorge sites (Steigerwald, Mount Zion, Strunk Road, and Bonneville) and diurnally fluctuating (easterly in afternoon, westerly all other hours) flow at the three easternmost sites (Sevenmile Hill, Wishram, and Towal Road). The apparent divergence of air in the eastern gorge for light easterly demonstrates that air must be entering the gorge from aloft or from the sides of the gorge in order to maintain mass balance. Divergence is also suggested for the westerly wind regimes as wind speed increases with distance into the gorge. Air entering the gorge from above or through the sides of the gorge could affect the bsp levels in the gorge in ways that cannot be determined from available data. The pressure gradient though the gorge (from Troutdale to The Dalles) decreases and changes sign during light easterly days. Moderate westerly is the most frequent pattern for the months August to October. It has a stronger diurnal variation than the other westerly patterns with an increase in wind speed in the late morning of about 2 m s−1 in the western gorge, 4 m s−1 in the central gorge (Bonneville), and continued increase until late afternoon (6 m s−1 increase) in the eastern gorge (Fig. 4).

5. Aerosol and wind pattern relationships

In this section, we compare the spatial patterns of light scattering by particles (bsp) with wind patterns. This provides some insight into likely source regions of aerosols in the gorge. Figure 7 gives daily average bsp at each site for each wind pattern. While there is a lot of information to process from Fig. 7, the following facts may be easily noted:

  1. For all sites except Steigerwald, winter easterly has the highest average bsp of all clusters. As the highest bsp levels are at the easternmost site, it suggests that sources to the east of the gorge are the main sources responsible for the haziest days due to wintertime inversions in the Columbia Basin. Figure 8 shows an example of intense haze on a winter easterly day in comparison with pristine conditions. For all sites, strong westerly has the lowest average bsp. Thus, the most typical summer pattern and most typical winter pattern have the lowest and highest bsp, respectively, for nearly all sites.

  2. The eastern sites (from Memaloose east) have much larger variations in average bsp between clusters than do the other sites (Steigerwald, Mount Zion, Strunk, and Bonneville).

  3. For the four eastern gorge sites, average bsp is inversely proportional to the westerly flow component.

Fig. 8.

View looking west into the gorge from Wishram for (a) a winter down-gorge (1200 PST 10 Nov 2004) day and (b) pristine conditions. The episode was characterized by daily average bsp over 200 Mm−1, high RH, and high concentrations of nitrate, sulfate, and organic carbon aerosols (see Green et al. 2006b).

Fig. 8.

View looking west into the gorge from Wishram for (a) a winter down-gorge (1200 PST 10 Nov 2004) day and (b) pristine conditions. The episode was characterized by daily average bsp over 200 Mm−1, high RH, and high concentrations of nitrate, sulfate, and organic carbon aerosols (see Green et al. 2006b).

Fig. 7.

Cluster average light scattering by particles (bsp), by site.

Fig. 7.

Cluster average light scattering by particles (bsp), by site.

a. Diurnal variation in bsp

Figure 9 shows the diurnal variation in bsp for five of the eight sites. For clarity the other three sites (Strunk Road, Sevenmile Hill, and Wishram) were left off the plots. Strunk Road diurnal patterns were very similar to Mount Zion’s and Wishram’s were nearly identical to Towal Road. Sevenmile Hill was similar to Memaloose, but differed some because of its elevation high above the river. For light westerly and moderate westerly, bsp increases first at Steigerwald and then progressively later eastward to Mount Zion and then Bonneville. This is presumed to be due to transport of pollutants from the Portland urban area. The farther east sites see a gradual rise in bsp late in the day for light westerly, possibly because of continued transport of the urban area pollutants. The eastern sites see a decrease as the day proceeds for moderate westerly and strong westerly, presumably because of greater dispersion with the stronger winds and deeper mixing as these patterns are most common in summer. The most pronounced diurnal variation for light easterly is at Steigerwald where bsp rises substantially in late morning. Winter easterly has increasing bsp during late morning at Memaloose along with transport from the east where bsp is higher. In the afternoon bsp then decreases at Memaloose possibly because of increased dispersion.

Fig. 9.

Diurnal patterns in bsp for selected sites by wind pattern: strong westerly, moderate westerly, light westerly, light easterly, and winter easterly.

Fig. 9.

Diurnal patterns in bsp for selected sites by wind pattern: strong westerly, moderate westerly, light westerly, light easterly, and winter easterly.

b. Effect of precipitation on bsp

Overall 61% of the days had 0.01 in. of precipitation or less at both PDX and The Dalles. Of the 39% of days with precipitation > 0.01 in. at one or both sites, about one-half had precipitation at both sites, about one-half had precipitation at Portland only, and a small percentage (3%) of days had precipitation at The Dalles only. Moderate westerly had no days with precipitation at The Dalles but not at PDX. Winter easterly was about equally likely to have precipitation only at The Dalles as only at PDX.

With some exceptions, days without precipitation have considerably higher bsp than days with precipitation at one or both sites. The average diurnal wind patterns for each wind type were computed and examined for both rainy days and dry days. Some differences were apparent. For the westerly wind patterns, smaller diurnal ranges in wind speed occurred on the rainy days for some sites, notably Bonneville and Sevenmile Hill. This is reasonable as cloud cover on rainy days would reduce the east to west temperature gradient that is a significant factor in these flows. Wind speed and direction differences would not account for the lower light scattering on rainy days compared to dry days.

In most cases, days with precipitation at The Dalles have the lowest bsp. Most days that had precipitation at The Dalles also had precipitation at PDX; this implies widespread precipitation resulting in scavenging of aerosol by precipitation throughout the gorge. An example of the relationship between precipitation and bsp is shown in Fig. 10 for the moderate westerly pattern. As Fig. 10 shows, moderate westerly days with precipitation at The Dalles had bsp levels at all but the two easternmost sites less than one-half as high as for days with no precipitation.

Fig. 10.

Average bsp at each site for the moderate westerly wind pattern for all days, days without >0.01 in. precipitation at either Portland or The Dalles (no rain), days with >0.01 in. at Portland and The Dalles.

Fig. 10.

Average bsp at each site for the moderate westerly wind pattern for all days, days without >0.01 in. precipitation at either Portland or The Dalles (no rain), days with >0.01 in. at Portland and The Dalles.

c. Evidence of local impact to bsp

Figure 11 shows the average bsp by time of day for four sites in the western and central gorge and one site in the eastern gorge for June through August. Winds in June through August are nearly always westerly, typically increasing in speed during midmorning to early afternoon in the western and central gorge. As winds increase from the Portland–Vancouver urban area, bsp increases first at the westernmost gorge site of Steigerwald and then at Mount Zion and Strunk Road, and finally at Bonneville. This pattern suggests transport of a pulse of hazy air from the urban area through the gorge. Later in the day as winds and vertical mixing increase and the air spends less time over the urban area bsp levels decrease. During June–August, the eastern sites have minimum bsp in midafternoon, then see an increase in bsp into the nighttime hours. The increase of bsp later in the day as distance eastward into the gorge increases suggests transport times of the urban emissions of about 2 h at Mount Zion and Strunk Road, 4 h at Bonneville, and 10–16 h for the eastern sites.

Fig. 11.

Diurnal variation in bsp for four western and central gorge sites and one eastern gorge site (Steigerwald eastward to Towal Road), summer (July–August). Note the eastern propagation of higher bsp levels late morning to early afternoon at the western and central gorge sites.

Fig. 11.

Diurnal variation in bsp for four western and central gorge sites and one eastern gorge site (Steigerwald eastward to Towal Road), summer (July–August). Note the eastern propagation of higher bsp levels late morning to early afternoon at the western and central gorge sites.

All sites from Memaloose east have their second highest average bsp for light easterly (highest is winter easterly), with the Sevenmile Hill and Memaloose sites having higher average bsp than the other eastern sites. Bonneville shows diurnally consistent down gorge flow for light easterly, while Sevenmile Hill, Wishram, and Towal Road showed diurnal variation in flow direction (westerly and easterly both) during light easterly days. Figure 12 shows the diurnal variation in westerly wind component and bsp at Sevenmile Hill for light easterly days. At 7 a.m., the wind direction changed from westerly to easterly and average bsp increased over the next few hours from about 29 to 38 Mm−1. Sevenmile Hill is located nearly directly above The Dalles. As the wind changes direction and comes from The Dalles, bsp increases. This increase suggests contributions to haze from local sources in the area of The Dalles.

Fig. 12.

Sevenmile Hill up-gorge wind component and bsp for the light easterly pattern.

Fig. 12.

Sevenmile Hill up-gorge wind component and bsp for the light easterly pattern.

6. Summary

The field portion of the Columbia River Gorge Haze Gradient Study was conducted from July 2003 through February 2005. Measurements included particle light scattering (bsp) and meteorological measurements at locations throughout the length of the gorge. The objectives of the study were to characterize horizontal, vertical, and temporal patterns in haze and to gain insight into possible source regions contributing to haze in the gorge. Cluster analysis was used to group days with similar wind patterns. Summaries of wind, pressure, and particle light scattering (bsp) were computed for each group of similar days (each cluster). Five wind patterns were constructed:

  1. strong westerly flow,

  2. moderate westerly flow,

  3. light westerly flow,

  4. light easterly flow (diurnal reversal at eastern sites), and

  5. winter easterly flow (light at east end, strong at west end).

Strong westerly was the predominant pattern in midsummer; winter easterly was the most frequent winter pattern. Light westerly and light easterly were most frequent in fall and spring transition months; moderate westerly was most frequent in late summer to early fall.

Winter easterly had the highest average bsp. The transport and bsp gradient pattern suggests that for winter easterly sources east of the gorge cause much of the haze.

Summertime showed diurnal patterns of increasing bsp progressing easterly to the Bonneville site during the day, suggesting the Portland–Vancouver urban area as a significant contributor to aerosol in the gorge in summer. Light easterly and winter easterly showed an increase in bsp from Wishram to Sevenmile Hill and Memaloose, suggesting impact from The Dalles area.

More detailed data analysis has been done for the Causes of Haze in the Gorge (CoHaGo) study (Green et al. 2006b). This includes additional analysis and receptor modeling using the nephelometer and surface meteorology data from the Haze Gradient Study and aerosol composition data collected for CoHaGo [filter samples, high time resolved sulfate, nitrate, elemental carbon/organic carbon (EC/OC), etc.]. Source modeling of a winter and summer episode using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research Mesoscale Model (MM5) and the Community Multiscale Air Quality (CMAQ) Modeling System has been completed and is reported elsewhere (Emery et al. 2007).

REFERENCES

REFERENCES
Cameron
,
D. C.
,
1931
:
Easterly gales in the Columbia River gorge during the winter of 1930–1931—Some of their causes and effects.
Mon. Wea. Rev.
,
59
,
411
413
.
Emery
,
C.
,
E.
Tai
,
U.
Nopmongcol
,
J.
Johnson
, and
R.
Morris
,
2007
:
Modeling analyses conducted for the Columbia River Gorge Air Quality Study. Final Rep., Southwest Clean Air Agency, Vancouver, WA, 310 pp. [Available online at http://www.swcleanair.org/gorgedata/Columbia_River_Gorge_ Modeling_Final_Report_082807.pdf.]
.
Green
,
M.
, and
J.
Xu
,
2007
:
Causes of haze in the Columbia River gorge.
J. Air Waste Manage. Assoc.
,
57
,
947
958
.
Green
,
M.
,
N.
Adhikari
,
J.
Xu
, and
G.
Nikolich
,
2006a
:
Columbia River gorge haze gradient study. Columbia River Gorge Air Quality Rep., Southwest Clean Air Agency, Vancouver, WA, 57 pp. [Available online at http://www.swcleanair.org/reports. html.]
.
Green
,
M.
,
J.
Xu
,
N.
Adhikari
, and
G.
Nikolich
,
2006b
:
Causes of haze in the gorge (CoHaGo). Columbia River Gorge Air Quality Rep., Southwest Clean Air Agency, Vancouver, WA, 106 pp. [Available online at http://www.swcleanair.org/reports.html.]
.
Reed
,
T. R.
,
1931
:
Gap winds of the Strait of Juan de Fuca.
Mon. Wea. Rev.
,
59
,
373
376
.
Sharp
,
J.
,
2002
:
Columbia gorge gap flow: Insights from observational analysis and ultra-high-resolution simulation.
Bull. Amer. Meteor. Soc.
,
83
,
1757
1762
.
Sharp
,
J.
, and
C. F.
Mass
,
2004
:
Columbia gorge gap winds: Their climatological influence and synoptic evolution.
Wea. Forecasting
,
19
,
970
992
.
Whiteman
,
C. D.
,
S.
Zhong
,
W. J.
Shaw
,
J. M.
Hubbe
,
X.
Bian
, and
J.
Mittelstadt
,
2001
:
Cold pools in the Columbia Basin.
Wea. Forecasting
,
16
,
432
447
.
Zhong
,
S.
,
C. D.
Whiteman
,
X.
Bian
,
W. J.
Shaw
, and
J. M.
Hubbe
,
2001
:
Meteorological processes affecting the evolution of a wintertime cold air pool in the Columbia Basin.
Mon. Wea. Rev.
,
129
,
2600
2613
.

Footnotes

Corresponding author address: Mark Green, Desert Research Institute, 755 E. Flamingo Road, Las Vegas, NV 89119. Email: green@dri.edu