Observational analyses and numerical simulations are used to investigate the interaction between a warm-season frontal system/trough and the complex terrain of the western United States. Prior to frontal landfall, synoptically driven offshore flow was associated with the northward development of a thermal trough, and a coastal pressure ridge and associated southerly winds moved northward along the coasts of northern California, Oregon, Washington, and southwestern British Columbia. This coastal pressure ridge, coupled with the subsequent weakening of the offshore flow, resulted in the onshore push of cool marine air approximately 24 h before frontal passage. The onshore surge of cool marine air into warm, dry continental air associated with the thermal trough produced a mesofront that moved over the coastal and Cascade Mountains.
Diagnosis of a realistic simulation of the event shows that both solar heating and adiabatic warming from downslope flow over the Rockies and the Cascade Mountains are important in producing the inland thermal trough. Model experiments without surface fluxes and orography indicate that surface fluxes are important in the formation of the prefrontal onshore push and that damming on the coastal mountains helps produce the mesoscale coastal ridge and alongshore southerlies.
The meteorology of the west coast of the United States is greatly influenced by mesoscale circulations resulting from the interaction between the synoptic-scale flow and coastal orography. The mesoscale response can be divided into two seasonal regimes: the cool season with relatively strong winds and low stability (high Froude number),1 and the warm season with weaker winds and a relatively stable atmosphere (low Froude number). Some examples of mesoscale circulations produced by the interaction of synoptic flow with West Coast orography include the Catalina eddy (Bosart 1983; Mass and Albright 1989), the alongshore surge (Dorman 1987; Mass and Albright 1987; Mass 1995), the onshore push of marine air (Mass et al. 1986), and the Puget Sound convergence zone (Mass 1981).
The topography of the Pacific Northwest includes two barriers that parallel the coast: the coastal mountains and the higher Cascade Range to the east (Fig. 1). Farther inland lie the more elevated Rocky Mountains. During the summer, gaps in the coastal mountains often allow marine air to intrude into the coastal interior, while the Cascades usually separate cool, marine air to the west from hot, dry continental air over eastern Washington and Oregon.
During the warm season, the east Pacific high dominates the large-scale flow over the eastern Pacific, producing low-level northerly flow and large-scale subsiding motions over coastal western North America. The persistent surface northerlies result in coastal upwelling that reduces nearshore sea surface temperatures, while the subsiding motions aloft result in adiabatic warming and an inversion or stable layer that caps a shallow marine layer. This synoptic pattern is sometimes interrupted when the Pacific high extends inland, producing offshore flow over the western slopes of the Cascades. This subsiding offshore flow contributes to the intensification and northward extension of the thermal trough that is usually resident over the deserts of the southwest United States. The thermal trough is associated with above-normal temperatures over the coastal zone.
A major summertime weather phenomenon of coastal western North America, the onshore surge (Mass et al. 1986) or marine push (Cramer 1973; Johnson and O’Brien 1973; Schroeder et al. 1967), occurs when a synoptic trough and associated front approach the coast and inland high pressure begins to shift eastward. Usually preceded by a period of above normal temperatures and fair weather associated with enhanced offshore flow, the marine push typically occurs prior to frontal passage and is often accompanied by strong winds, sharp temperature drops, and an increase of low clouds. The most intense events are associated with winds of 20 m s−1 or more, and temperature drops of 5°–10°C within a few hours.
Mesoscale coastal ridging often precedes warm-season frontal landfall and marine push events. Such ridging has been interpreted as freely propagating Kelvin waves (Dorman 1985, 1988; Reason and Steyn 1992), topographically trapped density currents (Mass and Albright 1987; Dorman 1988), and mesoscale damming associated with the interaction of the synoptic-scale flow with coastal topography (Mass et al. 1986; Mass and Albright 1988). Coastal ridging also occurs in other places around the world, such as southeastern Australia (e.g., Colquhoun et al. 1985; Wilson and Steyn 1985; Holland and Leslie 1986; Howells and Kuo 1988) and southern Africa (e.g., Gill 1977; Reason and Jury 1990).
Although the prefrontal onshore push of marine air is the most significant warm-season forecasting problem for the U.S. northwestern coast, its three-dimensional structure, evolution, and underlying physics have not been fully documented, partially due to insufficient observations over the adjacent ocean. Realistic mesoscale model simulations can provide dynamically consistent high-resolution data both in time and space and thus greatly increase our ability to study this important coastal phenomenon.
This paper describes an observational and numerical study [using the Pennsylvania State University–National Center for Atmospheric Research (using the (PSU–NCAR) Mesoscale Model Version 5 (MM5)] of the interaction of a weak warm-season front/trough with coastal orography. The interaction resulted in a marine push event and attendant features such as coastal ridging, alongshore southerlies, and the extension of the thermal trough. Key issues to be addressed include the following. What is the three-dimensional structural and dynamical evolution as a weak synoptic cold front/trough approaches and interacts with the coastal mountain ranges? How do coastal orography and the land–water interface alter the mesoscale features associated with the front? What is the relative importance of various physical processes, such as orographic blocking and differential surface fluxes, in the formation of the inland thermal trough, coastal pressure ridges, and the prefrontal marine push? Section 2 presents a synoptic description of the event. Section 3 describes the model and experimental design. Analyses (including model validation) of the model results are presented in sections 4 and 5. A summary is presented in the final section.
2. Observational description of the event
During 24–26 May 1992, a weak frontal system approached and crossed the west coast of North America. Before the front made landfall, a coastal ridge formed along the coast and an onshore push developed.
a. Surface and upper-level observations
Figure 2 presents the National Centers for Environmental Prediction (NCEP) synoptic-scale surface analysis approximately 24 h before landfall; detailed mesoscale analyses for the entire event are found in Fig. 3. At 0000 UTC 25 May 1992, a weak Pacific front extended equatorward from a 982-mb low center located over the Gulf of Alaska (Fig. 2). High pressure was located over the northern Rockies and a thermal trough extended from California into western Washington (Figs. 2 and 3a). As will be discussed later, daytime solar heating coupled with subsiding offshore flow (cf. Fig. 4a) resulted in the northward development of the thermal trough into the Pacific Northwest. Relatively high pressure was apparent over the ocean between the offshore frontal trough and the inland thermal trough. At this time, the temperature gradient across the coast was very large (greater than 10°C within a few tens of kilometers), and mesoscale ridging and ageostrophic alongshore (southerly) flow were observed along the northern California coast.
Six hours later at 0600 UTC 25 May (Fig. 3b), a mesoscale coastal pressure ridge and associated southerlies extended northward to the northern Oregon coast, and marine air had begun to spread into western Oregon. As a result, the thermal trough dissipated over western Oregon and the lowest pressure shifted into eastern Oregon and Washington. At 1200 UTC 25 May (Fig. 3c), the synoptic front was located about 300 km off the Washington and Oregon coasts, while coastal southerlies had extended to the Washington border. Along the British Columbia coastline, strengthening alongshore pressure gradients associated with the approaching front produced enhanced southeasterly coastal winds. The thermal pressure trough had jumped into eastern Oregon and Washington at this time, with its northern portion still located along the Strait of Georgia.
Six hours later at 1800 UTC 25 May (Fig. 3d), the front was within approximately 100 km of the coast, resulting in the narrowing of the coastal pressure ridge. Cool marine air had inundated western Oregon and Washington, resulting in large pressure and temperature gradients across the Cascades. At 0000 UTC 26 May (Fig. 3e), the front had made landfall on Vancouver Island and was in the process of making landfall to the south. The narrow pressure ridge ahead of the front had nearly disappeared, and the leading edge of the onshore push, accompanied by intense pressure and temperature gradients, had moved eastward to the Cascades. Along the coast, the winds had switched from southerly to westerly.
At 0600 UTC 26 May, the NCEP surface analysis (not shown) abruptly jumped the synoptic front into eastern Washington. This is also done in the mesoscale analysis (Fig. 3f) to reflect the leading edge of the onshore push of marine air. Over the ocean, the northward extension of the east Pacific high dominated the offshore waters, with westerly to northwesterly winds along the entire Pacific Northwest coast. The winds were deflected into two branches by the Olympic Mountains of Washington State and converged on the lee side of the mountains, initiating a Puget Sound convergence zone (PSCZ) event that produced significant rainfall over the Puget Sound area. A detailed analysis of the PSCZ event is described in Chien and Mass (1997).
The NCEP 850-mb analysis indicates that a closed low and associated trough were located over the eastern Pacific Ocean at 1200 UTC 24 May 1992 (Fig. 4a). Over land, a weak ridge was centered over the Alberta Rockies, generating modest easterly/northeasterly flow over the northern Cascades and Rockies. As will be discussed later, this offshore flow contributed to the northward extension of the thermal trough over the coastal interior. Twenty-four hours later at 1200 UTC 25 May (Fig. 4b), the Pacific trough had moved toward the coast and the inland ridge had shifted eastward, resulting in a transition to onshore flow west of the Cascades. At 0000 UTC 26 May, just prior to landfall of the surface front, there had already been substantial cooling at 850 mb at coastal locations (Fig. 4c); for example, the temperature at Quillayute on the northwest Washington coast had dropped from 13° to 8°C during the 12-h period. By 1200 UTC 26 May (Fig. 4d), a ridge had built over the eastern Pacific and the winds had switched to the northwest over coastal Washington and Oregon.
At 500 mb, a trough was located over the eastern Pacific Ocean and an upper-level ridge dominated the west coast of North America at 1200 UTC 24 May (Fig. 4e). Twenty-four hours later at 1200 UTC 25 May, the trough moved eastward rapidly and the ridge line drifted inland (Fig. 4f). A forward-sloping frontal zone is implied at this time since 500-mb temperatures had begun to decrease at coastal locations as the upper trough, and its associated baroclinic zone, moved toward the coast (the surface front was still offshore). Temperatures aloft continued to fall through 0000 UTC 26 May (Fig. 4g), at which time the surface front was just making landfall. By 1200 UTC 26 May (Fig. 4h), the upper-level trough was well inland and coastal 500-mb winds had veered into the northwest.
b. Satellite imagery
Figure 5 presents GOES (Geostationary Operational Environmental Satellite) visible satellite imagery during the event. At 0001 UTC 25 May (Fig. 5a), the frontal cloud band was approximately 600 km offshore and low clouds were observed along the California coast. Clear skies over Washington and Oregon and adjacent coastal waters were indicative of dry offshore flow.
The next morning at 1601 UTC (Fig. 5b), the frontal band was still 100–200 km offshore, while marine air was entering western Oregon and Washington (cf. Fig. 3). In advance of the front, convection was developing over the high terrain of western Oregon and Washington. Eight hours later at 0001 UTC 26 May (Fig. 5c), the front, evinced by an amorphous line of low clouds, was making landfall on the Oregon and Washington coasts. Ahead of the front, considerable convection was apparent in a line stretching from western Washington southeastward to the eastern slopes of the Oregon Cascades and then over the Sierra Nevada of California and Nevada. This convective band, also apparent on the NCEP radar summary chart (not shown), was associated with tops exceeding 10 700 m (35 000 ft). The next morning (1501 UTC 26 May), low clouds covered western Oregon and Washington, while higher clouds and precipitation, associated with the upper short wave, had spread eastward over Idaho and Montana (Fig. 5d).
c. Surface time series
To illustrate the significant differences in evolution experienced over land and offshore as the trough/front passed eastward through the region, Fig. 6 presents sea level pressure, surface temperature, and wind at an offshore buoy near 130°W (46005), a buoy proximate to the coast (46050), and a land station (EUG) located approximately 80 km inland (see locations in Fig. 1).
Buoy 46005 is located about 500 km west of the Oregon and Washington coastlines. Frontal passage was clearly evident at this station, with a rapid wind shift from southerly to westerly/northwesterly and a sharp pressure rise occurring at approximately 0600 UTC 25 May. The temperature only dropped a few degrees after frontal passage at this location.
The nearshore buoy 46050 had a considerably different wind evolution, with the winds initially from the north. Pressure rose abruptly at approximately 0200 UTC 25 May, followed by a windshift to southerly flow. This transition reflects the initiation of the onshore push and the northward movement of the coastal pressure ridge and associated ageostrophic southerlies. In addition, diurnal pressure variation (see Mass et al. 1991) and cooling aloft due to the forward-tilting baroclinic zone could contribute to the pressure rise. A second wind shift (to west-northwesterly) that occurred at 0000 UTC 26 May accompanied passage of the synoptic front.
Inland at Eugene (EUG), the winds veered from weak southerlies to northerlies on 24 May, with the maximum temperature attaining 31°C. Pressure reached a minimum at approximately 0200 UTC 25 May, after which the winds switched to southwesterly and increased with the arrival of marine air. Due to the onshore push of ocean air, the diurnal cycle was considerably suppressed on 25 May. The frontal passage at this Willamette Valley location was very subtle and was reflected in the rapid pressure rise at 0200 UTC 26 May. Later that day the winds turned northerly. The transition from continental to marine air was even more significant at Seattle, Washington, where the daily maximum temperature dropped from 31°C on 24 May to 22°C on 25 May (not shown).
In summary, while the pressure, wind, and temperature observations at the offshore buoy showed only a single frontal passage, the coastal and inland stations experienced two transitions: the first associated with the northward or inland intrusion of marine air and the second with the synoptic front. Offshore there was only a minor temperature drop with frontal passage, while inland the prefrontal onshore push of marine air resulted in a dramatic attenuation of the diurnal temperature variation.
d. Vertical soundings at Quillayute, Washington
Soundings at Quillayute, Washington (UIL), on the northwest Washington coast document the evolution of the lower troposphere during the event (Fig. 7). At 0000 UTC 25 May, the sounding was dry and warm, with a surface temperature of 27°C (Fig. 7a). Light surface winds from the northwest and north were surmounted by moderate southerly/southwesterly flow. Twelve hours later (1200 UTC 25 May), the sounding had moistened and a surface-based radiation inversion had formed. Onshore, southwesterly flow was now evident below 800 mb and temperatures in the midtroposphere (600–400 mb) had cooled by 2°–4°C. The sounding at 0000 UTC 26 May, just prior to surface frontal passage, was dramatically cooler and more moist throughout much of the lower and middle troposphere compared to 24 h before, with temperatures decreasing by about 11°C near the surface and 8°C at 500 mb (Fig. 7b). The postfrontal sounding at 1200 UTC 26 May shows a shift to northwesterly flow throughout the entire sounding. At this time, a saturated well-mixed marine layer below 900 mb was beneath a desiccated layer that extended to the top of the sounding.
3. Model description and experimental design
a. Model description
The multilayer Blackadar (1979) parameterization was used to represent planetary boundary layer processes including surface fluxes of heat, moisture, and momentum. The hydrological cycle includes the subgrid-scale convective parameterization of Grell (1993) and a grid-resolvable explicit moisture scheme, which includes prognostic equations for cloud water, rainwater, and ice (Dudhia 1993). In addition, the simulation utilized an upper radiation boundary condition, relaxation lateral boundary conditions, and a two-way interactive nested grid.
As shown in Chien and Mass (1994), the MM5 often predicts excessive low-level winds over mountainous regions by a factor of 2–5 because of the inability to properly represent the subgrid-scale effects of orography. To mitigate this problem, a simple procedure described in Chien and Mass (1994) was used to modify the boundary layer parameterization.2 These modifications produced significant improvement in the low-level winds.
b. Experimental design
Four experiments were conducted (Table 1) in this study. In each, the model was initialized at 1200 UTC 24 May 1992 and ran for 48 h. The domain configuration of the model simulation and the terrain within the inner domain (domain 2) are shown in Fig. 8. The outer and nested (domain 2) grids used in this study had horizontal resolutions of 45 and 15 km, respectively. Both grids had 23 sigma3 levels in the vertical. The time step for the coarse domain was 120 s.
The control/four-dimensional data assimilation (FDDA) (full physics) experiment included all the physical parameterizations discussed in the previous section. In addition, four-dimensional data assimilation4 was applied in the outer domain to provide realistic boundary conditions for the nested domain. Since one objective of the experiment was to examine the dynamical evolution in the inner domain, and any artificial forces created by FDDA would obscure such analysis, FDDA was not applied in the inner domain. The second simulation (experiment 2) was identical to experiment 1 except that FDDA was not applied in either domain so that experiment 2 could serve as a control run for comparison with the sensitivity experiments: experiments 3 and 4. Experiment 3 was designed to examine the influence of surface fluxes by removing both heat and moisture fluxes at the lower boundary in both domains. In experiment 4, the model terrain was set equal to zero everywhere in both domains; the atmospheric volume that was formerly occupied by mountains was initialized by interpolating the NCEP analyses to appropriate sigma levels. For these sensitivity experiments, FDDA could not be applied in the outer domain since FDDA implicitly would include the mechanisms that we were attempting to exclude.
4. Synoptic-scale and mesoscale evolution of the model simulation
a. Synoptic-scale surface evolution
Figure 9 presents the evolution of temperature and winds at the lowest sigma level (σ = 0.995) and sea level pressure in the nested domain (domain 2) of the control/FDDA experiment (experiment 1). At hour 6 (1800 UTC 24 May; Fig. 9a), high pressure was found over the far northeast portion of the domain, and easterly offshore flow was apparent along the western slopes of the Cascades and the Rockies. The offshore flow contributed to the northward extension of thermal troughs into western and eastern Washington. By hour 12 (0000 UTC 25 May; Fig. 9b), subsidence warming and diurnal heating had increased temperatures over land considerably, exceeding 30°C at some locations. As a result, the thermal trough intensified over western Oregon and Washington, producing a substantial pressure gradient and onshore flow along the Pacific Northwest coastline. Over the ocean, the synoptic front and accompanying trough have moved eastward into the domain. In general, the simulation compares quite well with observations (cf. Fig. 3a) except for the absence of the narrow southerly surge along the northern California border.
At hour 18 (0600 UTC 25 May), the simulation produced a coastal pressure ridge, the intrusion of cool marine air and higher pressure into western Oregon, and a strengthened alongshore pressure gradient (Fig. 9c). This alongshore pressure gradient, coupled with a weakening cross-shore (east to west) pressure gradient, resulted in an attenuation of the coastal northerlies and a reversal to southerlies along portions of the Oregon coast. Over eastern Washington, the thermal trough was maintained by the strong easterly/northeasterly winds descending the western slopes of the Rockies. A similar pressure distribution is suggested by the observations (cf. Fig. 3b).
At hour 24 (1200 UTC 25 May), the synoptic trough/front had progressed eastward, and the coastal pressure ridge had narrowed and extended northward to the central British Columbia coast (Fig. 9d). The southerly winds along the Oregon coast were directed down gradient, with relatively geostrophic southwesterlies offshore. Southwesterly marine flow had begun to move inland over western Washington.
Six hours later (1800 UTC 25 May), the front was approximately 150 km offshore and the coastal pressure ridge had become extremely narrow (Fig. 9e). The winds along the coast were generally alongshore, reaching approximately 10 m s−1 near the coast of Vancouver Island and southwestern British Columbia, where the alongshore pressure gradient was large. Cool marine air had flooded the coastal areas west of the Cascades and downslope motion on the Rockies had produced subsidence warming and lee troughing in the interior of British Columbia.
At hour 36 (0000 UTC 26 May), as the synoptic front/trough was making landfall (Fig. 9f), marine air covered the coastal area to the west of the Cascades with temperatures about 10°C cooler than those of the previous day. To the east, the thermal trough reintensified due to diurnal heating. Large pressure and thermal gradients, accompanied by moderate westerly winds, were present over the Cascades, corresponding to the leading edge of the marine air.
By hour 42 (0600 UTC 26 May), the synoptic front/trough had made landfall and its structure had been greatly modified and attenuated by the coastal orography (Fig. 9g); behind the front, the coastal winds had veered into the northwest. Marine air had pushed well into eastern Oregon and was just entering eastern Washington, accompanied by strong northwesterly and southwesterly winds. As noted earlier, the NCEP subjective surface analysis at this time placed the landfalling synoptic cold front at the leading edge of the marine air mesofront (see Fig. 3f), which caused the analyzed frontal movement to appear discontinuous after landfall. Over the Puget Sound area, a trough formed in the lee of the Olympic Mountains.
At hour 48 (1200 UTC 26 May; Fig. 9h), the lee trough to the east of the Olympic Mountains of Washington State had strengthened; responding to this troughing, south-southwest winds to the south of Puget Sound converged with north-northwest flow channelled from the Strait of Juan de Fuca, producing a Puget Sound convergence zone over central Puget Sound [the development of this convergence zone is described in detail in the accompanying paper, Chien and Mass (1997)]. Marine air had swept into eastern Washington at this time, resulting in the attenuation of the thermal trough.
To illustrate how the temporal evolution varied in the cross-shore direction, the simulated wind and temperature at the lowest sigma level and sea level pressure are shown in Fig. 10 for several points along an east–west line (the locations of these points are shown in Fig. 8b). Several hundred kilometers west of the coast at point a, there was a frontal wind shift at approximately 0500 UTC 25 May from southeasterly/southerly to westerly. The initiation of this wind shift followed a pressure rise starting several hours before and was followed by a slow temperature decline. Comparisons with observations at nearby buoy 46005 (cf. Fig. 6) show that the windshift and pressure rise associated with frontal passage are quite consistent with observations except that the model winds were less northerly after frontal passage. At point b, approximately 200 km offshore, the surface evolution was significantly different. Southerly flow increased at 0300 UTC 25 May, followed by a rapid transition to the northwest at approximately 1400 UTC 25 May during frontal passage. Frontal passage was also accompanied by a transition to a larger rate of increase in surface pressure.
At point c, approximately 60 km west off the coast, the winds became northwesterly on 24 May as the onshore “sea breeze” circulation developed. With the development of the coastal pressure ridge and the approach of the synoptic trough/front, the winds turned southerly/southwesterly at 0600 UTC 25 May. As at the other locations, pressure began to rise at 0300 UTC 25 May after falling the previous 12 h. Frontal passage, which occurred at approximately 2200 UTC 25 May, was associated with a rapid shift to northwesterly flow, with little pressure signal. Temperature shows little modulation during the period except for a slight decline after 0600 UTC 25 May. In general, this simulated time series agrees very well with the observations at a nearby buoy (buoy 46050; Fig. 6), which experienced similarly timed transitions from northwesterly to southerly to northwesterly flows.
Over the Willamette Valley of Oregon (point d), winds were weak and generally from the northwest on 24 May, but increased and backed into the southwest on 25 May as marine air pushed into western Oregon. The exact time of frontal passage on 26 May is difficult to determine from the wind field but probably occurred during the shift from southwesterly to northwesterly flow at 0300 UTC 26 May. In contrast to the ocean locations, there was a considerable diurnal temperature signal at point d during the first 24 h; such diurnal variability was considerably attenuated on 25 May due to the intrusion of the low-level marine air and clouds. Pressure increased rapidly at around 0300 UTC 25 May, with the intrusion of cool, marine air at low levels. The simulated results at this point show good agreement with the observations at a nearby station, Eugene (EUG; Fig. 6). The final point, e, located over the high plateau of eastern Oregon, experienced the arrival of the marine air intrusion at approximately 0000 UTC 26 May, as suggested by the increase in westerly flow, a sharp pressure rise, and a substantial drop in temperature.
All the above locations experienced a nearly simultaneous pressure rise beginning at approximately 0300 UTC 25 May. This feature might be the result of the superposition of the large diurnal pressure rise that typically occurs at this time (Mass et al. 1991) and pressure increases accompanying the considerable cooling aloft that preceded the surface front (cf. Fig. 4).
b. Three-dimensional evolution
To illustrate the three-dimensional structural evolution during this event, two east–west vertical cross sections (locations shown in Fig. 8b) are presented in this section. Figure 11 presents the evolution of potential temperature, winds, and relative humidity along cross section AA′, which extends from 1000 km out into the Pacific Ocean, eastward through the Willamette Valley of Oregon, and over the high plains of eastern Oregon. At 0000 UTC 25 May (Fig. 11a), the surface front, associated with a weak windshift at low levels and only minor cooling, is approximately 700 km west of the coast. Middle- and upper-tropospheric baroclinity is found over and east of the surface front. Above and ahead of the surface front there is also a large area of moist air, associated with moderate upward motion (not shown). At this time, surface heating over the interior has produced a deep dry adiabatic layer, especially over eastern Washington, where it extends to about 1600 m above the surface. Over the coastal zone, a shallow layer of cool marine air has intruded eastward to the coastal mountains.
By 1200 UTC 25 May (Fig. 11b), the surface synoptic front has moved eastward to about 300 km offshore, while the leading edge of the marine air has reached the Cascades. The low-level winds back from southwesterly just offshore to nearly terrain parallel (southerly) near the coast. The forward tilt of the tropospheric baroclinity has increased at this time and dry air has pushed aloft over the surface cold front. The forward-tilting baroclinity and midtropospheric dry air intrusion of this section are reminiscent of similar structures described by Browning and Monk (1982), Browning (1990), McBean and Stewart (1991), and Mass and Schultz (1993).
Twelve hours later at 0000 UTC 26 May (Fig. 11c), the surface front is near landfall, while aloft the leading edge of the dry air has swept east of the front. The midtropospheric baroclinity is now over land, well in advance of the surface front. As on the previous day, surface heating produces a well-mixed boundary layer over eastern Washington. In contrast, the heating is suppressed to the west of the Cascades by the marine air intrusion. As a result, a sharp temperature contrast or mesofront develops, which represents the leading edge of the marine air crossing the Cascades. The combination of this low-level mesofront and synoptic-scale upward motion results in convection over eastern Oregon (cf. Fig. 5c). By 1200 UTC 26 May (Fig. 11d), the surface front has made landfall and attenuated, and the low-level marine air had moved to the eastern boundary of the domain.
To further illustrate the development of the thermal trough and the subsequent onshore push of marine air, Fig. 12 presents vertical cross sections (BB′ in Fig. 8b) that cut across the Washington coast, the Cascades, and eastern Washington, and end along the western slopes of the Rockies. At hour 12 (0000 UTC 25 May; Fig. 12a), during the time of thermal trough development over Washington State, deep, well-mixed boundary layers are found over both western (1–1.5 km) and eastern (∼2 km) Washington. Easterly flow descends the western slopes of the Rockies into eastern Washington, with some subsiding flow west of the Cascades. There is a cooler, more stable air to the west of the coastal mountains, with some suggestion of an onshore, sea breeze circulation.
By hour 24 (1200 UTC 25 May; Fig. 12b), marine air and onshore flow have extended to the western foothills of the Cascades, while strengthening easterly subsiding flow over eastern Washington helps maintain the thermal troughing apparent in Fig. 9d. The synoptic front was approximately 200 km offshore at this time. Twelve hours later at 0000 UTC 26 May (Fig. 12c), as the front/trough made landfall, marine air has pushed across the Cascade crest. At the leading edge of the marine air, there is a plume of strong upward motion and moist air that corresponds to the convection seen in the satellite imagery over Washington and Oregon Cascades. With a deep adiabatic layer over eastern Washington, this leading edge of marine air is similar to a mesoscale cold front and is associated with a large horizontal temperature gradient. This mesofront moves into eastern Washington by 0600 UTC 26 May (Fig. 12d), accompanied by strong downslope winds and higher humidity. At the same time, the synoptic front/trough has made landfall and dissipated to the west of the Cascades (cf. Fig. 9g).
c. The formation of the mesoscale coastal ridge and alongshore southerlies
To further examine the mesoscale orographically induced response to the approaching synoptic system, including the formation of the narrow coastal pressure ridge and the alongshore southerlies, sea level pressure, and winds over the coastal zone are displayed in Fig. 13. At hour 15 (0300 UTC 25 May; Fig. 13a), a large-scale coastal pressure ridge is evident between the synoptic front/trough (at the western boundary of the domain) and the thermal trough over land. Marine air has pushed across southwest Oregon coast, resulting in an eastward extension of the coastal ridge into the southern Willamette Valley of Oregon. Pressure ridging with a smaller scale is apparent on the windward (northwest) side of the Siskiyou Mountains along the Oregon–California border. Coastal winds vary from strong northwesterlies on the northern Oregon coast to near calm along the California border.
By 0900 UTC 25 May (Fig. 13b), the synoptic front (and associated trough) has entered the western boundary of the domain, with strong southerly/southwesterly flow ahead of it. A narrow coastal ridge extends northward to the northern Washington coast, and the leading edge of the alongshore southerlies reaches the northwest coast of Oregon. In addition, a mesoscale pressure ridge was evident over the western slopes of the Oregon Cascades. Strong southerly flow accelerates over the northern Willamette Valley toward the remnant heat trough over the Puget Sound region.
Six hours later at 1500 UTC 25 May (Fig. 13c), the synoptic front has moved closer to the coast and winds along the entire coast have become southerly. The mesoscale coastal ridge along the coast has an amplitude of approximately 0.25 mb and a half-width of 60 km. At this time, the pressure ridge in the coastal zone extends northward along the western slopes of the Oregon and Washington Cascades.
Figure 14 displays an east–west vertical cross section (DD′; see Fig. 13c for location) across the coastal pressure ridge at hour 27 (1500 UTC 25 May). Extending approximately 60 km to the west of the coastal range, there is topographic damming and trapped low-level southerly flow. This damming of cooler air results in the mesoscale coastal pressure ridge along the coast (as will be shown in the no terrain simulation in the next section). A similar mesoscale damming configuration occurs over the western slopes of the Cascades.
5. Diagnosis and sensitivity studies
a. Force balances
To determine how force balances associated with the coastal pressure ridge and the alongshore southerlies evolved in time, and to evaluate coastal dynamical adjustment, the magnitudes of the terms in the momentum equations [Eq. (2.2.1) and (2.2.2) in Grell et al. (1994), with p*5 being decoupled] were diagnosed using model data. For simplicity, the nonhydrostatic equations are summarized as
where d/dt is the total derivative, which includes local time derivative and 3D advection terms; PGF represents the pressure gradient force; and the residual term includes diffusion, friction, and numerical errors. Time differences were approximated using centered time differencing (Δt = 30 min) from model output data.
Figure 15 presents force balances in the cross-shore (x) and alongshore (y) directions along cross section CC′ (see Fig. 13a for location) at the lowest sigma level (σ = 0.995). At hour 15 (0300 UTC 25 May), northwesterly flow dominates the coastal portion of this cross section, with the winds backing to southerly offshore. The force balances in the cross-shore (x) direction (Fig. 15a) show that far offshore the zonal wind component is nearly geostrophic. Approaching the coast, the signs of the Coriolis (C curve) and the pressure gradient (P curve) forces reverse, maintaining an approximate geostrophic balance over the near-coastal waters. Over land, the primary force balances are between the strong pressure gradient (between the offshore ridge and the inland thermal trough) and residual terms (R curve; friction is probably the key component). In the alongshore (y) direction (Fig. 15b), the flow is roughly geostrophic between approximately 130 and 300 km of the coast. Within 130 km of the coast, the pressure gradient force increases significantly, while the Coriolis force remains near constant, resulting in downgradient (northward) acceleration (T curve) that would tend to weaken and reverse the coastal northerlies observed at this time.
Six hours later at 0900 UTC 25 May, when the mesoscale coastal ridge and associated alongshore southerlies extend northward up the coast (cf. Fig. 13b), there are notable changes in the force balances within the cross section. In the zonal direction, the cross-shore pressure gradient force (P curve) has reversed within 130 km of the coast (Fig. 15c) compared to the earlier section (Fig. 15a). The Coriolis force (C curve) also reverses to onshore-directed at this time because of the formation of the alongshore southerlies. However, it is too small to balance the large offshore-directed pressure gradient force, resulting in deceleration (T curve) of the onshore flow near the coast. In the alongshore direction within 100 km of the coastline, there is a downgradient (northward) acceleration within the mesoscale coastal ridge (Fig. 15d) since the pressure gradient force exceeds the sum of the Coriolis and residual (friction) forces.
As discussed in section 4c, there are two coastal ridges with different scales forming offshore (cf. Fig. 13): a large-scale coastal ridge with a half-width of about 130 km between the offshore synoptic front and the inland thermal trough and an orographically induced mesoscale coastal ridge with a half-width of 60 km embedded in the eastern portion of the large-scale ridge. This dual-ridge structure is also evident in the force balances (Fig. 15). For example, in Fig. 15c the coastal orographically induced pressure ridge is associated with a narrow (about 60 km offshore) zone of enhanced negative (toward the west) zonal pressure gradient force and a resulting large zonal deceleration. This offshore scale is very close to the Rossby radius,6 which is approximately 70 km calculated from the model data along this cross section at 0900 UTC 25 May. The scale of the larger pressure ridge is simply dependent on the distance between the approaching offshore trough and the thermal trough over the coastal lowlands.
b. Thermal balances
The thermal balances during this event were computed by diagnosing the thermodynamic energy equation. It was hoped that such a diagnosis could answer several important questions, such as the physical mechanisms responsible for the coastal thermal trough and the temperature changes associated with the marine air intrusion. The thermodynamic equation can be written as
where the local temperature change is on the left-hand side, and the horizontal advection of temperature, the temperature change due to vertical motion, and the residual terms (including diabatic heating, diffusion, and numerical error) are on the right. Local change was approximated using the same method applied for the calculation of the force balances.
Figure 16 presents the thermal balances of the above terms averaged over the lowest five sigma layers in the model (approximately the lowest 80 mb above the ground). At hour 3 (1500 UTC 24 May), easterly subsiding flow produced adiabatic warming over the coastal zone and the western slopes of the Cascades and the Rockies. This warming was partly compensated by horizontal advection over the western slopes of the Cascades and Rockies. The coastal zone, where temperature advection, adiabatic warming, and diabatic heating were all occurring, was the focus of the largest warming.
At hour 12 (0000 UTC 25 May), when the coastal thermal trough reached its maximum amplitude, temperature was falling rapidly over western Oregon as a result of the onshore flow of cool, marine air and adiabatic cooling along the western slopes of the Oregon Cascades. Diabatic heating of 20 × 10−5 K s−1 was found over most of the domain. Six hours later at hour 18 (0600 UTC 25 May), horizontal advection and adiabatic cooling produced cooling centered over northwest Oregon, while downslope flow was still producing adiabatic warming over the western slopes of the Washington Cascades and the Rocky Mountains. At hour 36 (0000 UTC 26 May), as the marine air pushed eastward across both the Oregon and Washington Cascades, adiabatic cooling over the western slopes and warming over the eastern slopes were opposed by horizontal cold advection.
In summary, adiabatic warming, diabatic heating, and, to a minor degree, temperature advection contributed to the intensification of the thermal trough over the coastal area on 25 May. Subsequently, the onshore push of cool, marine air and upslope adiabatic cooling resulted in a decrease of temperature over western Oregon and Washington. Over eastern Washington, diabatic heating as well as subsidence warming that resulted from easterly flow off the Rockies contributed to the maintenance of the inland thermal trough through 26 May.
c. Trajectory analyses
At the most basic level, the onshore push phenomenon represents the shift from offshore to onshore flow over the coastal zone. To document this trajectory transition, Fig. 17a presents 24-h backward trajectories of air parcels (calculated from 15-min model output data) that were released at σ = 0.99 at 1200 UTC 25 May along a line extending through the offshore front, the coastal pressure ridge, and the inland thermal trough. The synoptic front at this time was located between the ending locations of air parcels 2 and 3. Parcels 1 and 2 approached the front from the west, while ahead of the front trajectories 3 and 4 moved to the north. Air parcels farther to the east, such as 6–9, first experienced offshore flow and then turned eastward as marine air began to move onshore. A distinct discontinuity occurred between trajectories 9 and 10, which were located on opposite sides of the advancing onshore push; trajectory 9 indicated onshore marine flow, while trajectory 10 and those to the east moved westward and were of continental origin. Several of these offshore-directed trajectories experienced downslope motion, in contrast to the marine trajectories (1–8), which were quasi-horizontal.
At hour 36 (0000 UTC 26 May) the synoptic front was on the coast, while the mesofront accompanying the leading edge of the cool, marine air was crossing the Oregon and southern Washington Cascades (Fig. 9f). The 30-h trajectories ending at this time (Fig. 17b) show that air parcels ending behind the synoptic front/trough (e.g., 1–5) originated far to the west. In contrast, trajectories ending on the coast (7–10) in advance of the front started from the southwest and moved northeastward. The air parcels ending on the cold air side of the advancing onshore push/mesofront (12–15) moved eastward and then rose rapidly on the western slopes of the Cascades, while air parcels terminating on the warm side of the push had continental origins and were generally descending from the north during much of the period.
d. Effects of surface fluxes and terrain
Since it is evident that the complex terrain of the Pacific Northwest alters the lower-tropospheric evolution accompanying the landfalling synoptic front—producing complex mesoscale phenomena such as the coastal pressure ridge, coastal ageostrophic southerlies, and the prefrontal onshore push—several sensitivity experiments were performed in which the orography and surface characteristics were varied. As explained in section 3, these sensitivity experiments were compared with experiment 2 to examine the impact of the removed physics or terrain. It was found that the fields produced in experiment 2 (the control/no-FDDA experiment) are quite similar to those found in experiment 1 (the control/FDDA experiment) except that the synoptic-scale front is a little weaker and sea level pressure is generally about 1 mb higher. For the sake of brevity, the results of experiment 2 are not shown in this paper, and those of experiment 1 are used as references for comparisons.
Figures 18a,b show the results of the simulation without surface heat and moisture fluxes (experiment 3). At hour 12 (0000 UTC 25 May; Fig. 18a), surface temperature is much cooler over land and the thermal trough is considerably weaker compared to the control experiment (cf. Fig. 9b). As a result, in the no-flux simulation the coastal pressure gradients are attenuated and only weak onshore winds are evident along the Oregon coast. Along the Washington coast, the winds are reversed (offshore) compared to those of the control experiment. At hour 36 (0000 UTC 26 May), as the front is making landfall on the coast, the inland thermal trough and the strong pressure/thermal gradients over the Cascades do not develop without surface heat fluxes (Fig. 18b), in contrast to the control simulation (Fig. 9f). There is also little evidence of a prefrontal onshore push of marine air.
Figures 18c,d present vertical cross sections along line BB′ (location is shown in Fig. 8b) at 12 and 36 h into the no-flux simulation. At hour 12 (0000 UTC 25 May), there are dramatic differences between the control (Fig. 12a) and no-flux (Fig. 18c) cross sections; without surface fluxes, temperatures are far lower over land and the deep adiabatic layers found in the control simulation are absent. As a result, there is little thermal gradient across the coast, and low-level onshore flow is not present; instead, strong easterly offshore flow is evident along the entire section. Subsidence over the western slopes of the Cascades contributes to the formation of the weak trough (cf. Fig. 18a). At hour 36 (0000 UTC 26 May; Fig. 18d), the deep mixed layer east of the Cascades and the strong mesofront over the Cascade crest found in the control experiment (cf. Fig. 12c) do not develop in this section without surface heat fluxes.
In another sensitivity experiment (experiment 4), the terrain in both domains was set to zero and surface fluxes were retained. At hour 12 (0000 UTC 25 May; Fig. 19a), temperatures have warmed over land, producing a far broader thermal trough than in the control simulation (cf. Fig. 9b). Without topography, the coastal sea breeze and the inland offshore flow are far stronger, but the coastal troughing is less defined because of the absence of subsidence warming. Although the synoptic front/trough position 18 h later at 1800 UTC 25 May (Fig. 19b) is similar to the control (Fig. 9e), the mesoscale wind, temperature, and pressure structures over land are greatly altered. Without the regional mountains, the narrow coastal pressure ridge and associated ageostrophic southerlies are absent, the sea-breeze flow penetrates much farther inland, and strong mesoscale pressure and thermal gradients do not develop. By hour 36 (Fig. 19c), the synoptic front/trough has made landfall on the Pacific Northwest coast. Without coastal mountains, the front moves inland faster than in the control simulation (cf. Fig. 9f). It is interesting to note that without topography a mesoscale front/onshore push still develops in advance of the synoptic front, as a strong onshore flow intrudes on the diurnally heated land.
Similar to the control run (see Fig. 12a), cross section BB′ of the no terrain simulation shows initially a thermal contrast between land and ocean, resulting in a shallow sea-breeze front between the cool, marine air and the easterly, heated flow over land (not shown). By hour 36 (0000 UTC 26 May; Fig. 19d), the leading edge of the marine air has moved farther to the east than in the control run (cf. Fig. 12c). However, without terrain in the model, the thermal contrast at the leading edge of the onshore push and the associated convection aloft are weaker than in the control simulation.
In summary, sensitivity experiments in which either surface fluxes or orography were removed suggest that surface fluxes are important in producing the inland thermal trough, and that the contrast in surface properties between land and water plays a crucial role in producing a prefrontal intrusion of marine air into the Pacific Northwest. Although secondary in importance, orography modifies and intensifies the structure of the thermal trough through subsidence warming, is crucial for the development of the narrow coastal pressure ridge and associated ageostrophic coastal southerlies, and greatly strengthens the thermal transition associated with the onshore push.
6. Discussion and conclusions
An observational and numerical study of a warm-season Pacific front that made landfall on the complex terrain of the Pacific Northwest is presented above. Interaction between the synoptic-scale flow and the complex terrain of the region greatly modified the low-level evolution associated with the front, producing a range of mesoscale phenomena such as the prefrontal coastal pressure ridge, the prefrontal onshore push of marine air and mesofront, thermal troughing, and ageostrophic alongshore southerlies.
Even before it approached the coast, the Pacific frontal system of 24–25 May 1992 had a structure different from that of the classic Norwegian cyclone model. As seen in both observations (Figs. 3 and 4) and the model simulation (Fig. 11), the baroclinity associated with the front tipped forward with height, as did the moisture and precipitation features. This tilted structure is similar to those found by Browning (1990), Locatelli et al. (1995), and others; such a structure develops when the temperature gradient associated with the upper-level short-wave trough moves ahead of the surface front as the short wave aloft propagates eastward more rapidly than the surface trough. Because of its long overwater trajectory, the synoptic front was associated with a very weak temperature gradient at low levels.
Prior to frontal landfall, there was a northward extension of a tongue of warm temperatures and associated thermal troughing into the Pacific Northwest. This thermal trough was most intense over lowland areas, such as Oregon’s Willamette Valley and the basin of eastern Washington, where adiabatic warming due to downslope flow was strongest. Offshore flow over the Pacific Northwest was an important factor in the thermal trough development because of its association with subsidence warming on the regional topography and, more significantly, because it allows the diurnal heating to be unattenuated by cool onshore flow. A weak pressure ridge developed over the ocean between the thermal trough and the trough accompanying the synoptic front. With a large onshore pressure gradient developing between the coastal pressure ridge and the inland thermal trough, as well as a weakening of the offshore flow, marine air began to push into western Oregon and Washington during the late evening hours of 24 May. This onshore push was well in advance of the synoptic front, which was still hundreds of kilometers offshore. The coastal mountains partially blocked the influx of marine air, resulting in amplification of mesoscale coastal pressure ridging and ageostrophic southerly flow near the coast (within about 60 km of the coast). This offshore scale is very close to the Rossby radius, which approximately represents the offshore distance of orographic influence of the coastal mountains. As the synoptic front approached the coast on 25 May, the leading edge of the cool, marine air surged over the Cascade Mountains as an intense mesofront, with strong winds, rising pressure, and plummeting temperatures. The synoptic front weakened rapidly upon making landfall. As a result, NCEP frontal analyses jumped the frontal position eastward to the location of the mesofront.
Sensitivity experiments that removed surface fluxes or the regional orography revealed that surface fluxes (i.e., surface-based diurnal heating) played a dominant role in producing the inland thermal trough and the prefrontal onshore push of marine air. Implicitly, the existence of offshore flow is significant in trough development because it prevents the intrusion of cool, marine air at low levels. Orography played a secondary role, modifying the structure of the thermal trough and intensifying the mesofront as it passed over and to the lee of the Cascades. In addition, coastal orography slowed down the front, enhanced the mesoscale coastal pressure ridge, and forced the alongshore southerlies.
This paper has dealt primarily with the synoptic and mesoscale features preceding and immediately following the passage of a front across the Pacific Northwest coast. In the accompanying paper, the subsequent development of an important local mesoscale phenomenon, the Puget Sound convergence zone, is described.
This research was supported by the National Science Foundation (Grants ATM-9111011 and ATM-9416866) and the ONR Coastal Meteorology Accelerated Research Initiative (Grant NH45543-4454-44). Use of the MM5 was made possible by the Microscale and Mesoscale Meteorological Division of the National Center for Atmospheric Research. The mesoscale model was run at the Scientific Computing Division of NCAR. We wish to thank Ernie Recker for acquiring the observational data, Mark Albright for his suggestions regarding the surface analyses, and three anonymous reviewers for useful recommendations.
Corresponding author address: Dr. Clifford Mass, Department of Atmospheric Sciences, University of Washington, P.O. Box 351640, Seattle, WA 98195-1640.
The Froude number is defined as Fr = U/(hmN), where U is the wind component approaching the barrier, hm is the height of the obstacle, and N is the buoyancy frequency [(g/θ)∂θ/∂z]1/2.
Specifically, higher-resolution terrain data were used to calculate the subgrid-scale terrain roughness surrounding the grid points. Then, the friction velocity at each grid point was increased by a factor proportional to the subgrid-scale roughness.
The vertical coordinate σ is defined as (p − pt)(ps − pt)−1, where p is pressure, ps is surface pressure, and pt is a constant pressure at the top of the model (50 mb). Here σ = 0.995, 0.985, 0.97, 0.945, 0.91, 0.87, 0.825, 0.775, 0.725, 0.675, 0.625, 0.575, 0.525, 0.475, 0.425, 0.375, 0.325, 0.275, 0.225, 0.175, 0.125, 0.075, and 0.025.
Four-dimensional data assimilation is a technique for dynamically relaxing or “nudging” the model solution toward the observed data. Two kinds of nudging (analysis and observation nudging) can be utilized in the MM5 (Stauffer and Seaman 1990; Stauffer et al. 1991), but only analysis nudging was used in the simulation. The analysis data are obtained from NCEP grids that were modified using surface and upper-level observations.
Here, p = ps − pt, where ps is surface pressure and pt is a constant pressure at the top of the model (50 mb).
The Rossby radius that approximately represents the offshore distance of orographic influence is defined as lR =Nhm/f, where N is the Brunt–Väisälä frequency, hm is the mountain height, and f is the Coriolis parameter.