a. Theoretical studies

Considerable theoretical research has addressed the dynamical effects of mountains on frontal systems. Both Egger and Hoinka (1992) and Blumen (1992) provide extensive reviews of past theoretical research on frontal interaction with orography; therefore, this section provides only a limited overview of more recent studies that can be related to COAST IOP5. Both two-dimensional (Blumen and Gross 1987a,b; Zehnder and Bannon 1988; Williams et al. 1992) and three-dimensional modeling results (Gross 1994) suggest that either reduced frontogenesis or frontolysis occurs as a front ascends a windward slope, while frontogenesis is favored in the lee. Frontolysis on the windward side is forced primarily by divergence associated with low-level flow acceleration over higher terrain, while frontogenesis is forced by convergence and stretching deformation as the front descends the lee slope (Gross 1994).

Frontal deformation and changes in frontal speed during passage across topography have also been noted in idealized modeling experiments. Fronts have been shown to experience substantial retardation and deformation if the Froude number [U/(Nh_m), where U is the flow normal to the barrier, N is the stability, and h_m is the height of the barrier] is ≪1 (Schumann 1987;Gross 1994). Therefore, a higher mountain, weaker flow normal to the barrier, and greater stability at low levels all favor greater windward deceleration and deformation of the front. Fronts characteristically experience an “anticyclonic distortion” (a clockwise distortion of the front around the barrier in the Northern Hemisphere) while crossing large three-dimensional barriers as a result of the circulations induced by the orographic pressure perturbations (Schumann 1987). Gross (1994) noted that if the Froude number is very low (Fr ∼ 0.1), a shallow layer of flow, similar to a gravity current, accelerates anticyclonically around the barrier. Most previous idealized studies considered large-scale barriers (>1000-km width); however, Li et al. (1996) and Schumann (1987) showed that large distortions of the front can also occur for smaller circular barriers (200–500-km width).

b. Observational studies

A growing number of studies have dealt with the interaction of fronts with the coastal orography of western North America. Overland and Bond (1993) described strong low-level coastal winds and pressure rises that propagated northward along the southeast Alaska coast following the landfall of a surface cold front. Steenburgh and Mass (1996) documented the interaction of an intense extratropical cyclone (the“Inauguration Day Cyclone”) with the topography of Washington State, in which postfrontal westerly flow resulted in troughing in the lee of the Olympics and a prolongation of damaging winds over Puget Sound. Doyle (1997) presented high-resolution simulations of a prefrontal coastal jet along the California coastal range, which slowed the eastward progression of a front in the coastal zone. Using NOAA P-3 Doppler wind observations, Braun et al. (1997) documented the terrain-enhanced prefrontal flow paralleling the Oregon coast during COAST IOP3. Chien et al. (1997) and Mass et al. (1986) described the onshore push of marine air that sometimes occurs in advance of a warm-season frontal passage along the Washington and Oregon coasts.

There have been several studies of the modulation of frontal precipitation by the coastal orography of western North America. Using data from the Cascade Project, Hobbs (1975) showed that the amount of cloud and precipitation ahead of occluded fronts decreased as they approached the Washington Cascades, while behind the front precipitation over the Cascades increased. Precipitation structures associated with fronts over the Pacific Northwest were also studied during the Cyclonic Extratropical Storms Project (CYCLES) (e.g., Houze et al. 1976; Hobbs 1978; Hobbs et al. 1980; Matejka et al. 1980; Hobbs and Persson 1982). Parsons and Hobbs (1983) used radar data from CYCLES to show the dissipation of warm-sector rainbands as they moved inland over the Washington terrain. Carbone (1982) used three land-based Doppler radars to describe the weakening of an intense narrow cold-frontal rainband as it approached the Sierra Nevada of California. Braun et al. (1997) showed that a narrow cold-frontal rainband appeared to weaken and become aligned along the Oregon coast as it neared landfall during COAST IOP3.

There have been a number of observational studies of the interaction of fronts with major mountain ranges in other parts of the world. The Alpine and German Front Experiments were designed to document frontal interaction with the European Alps (Egger and Hoinka 1992; Hoinka and Volkert 1992), including the anticyclonic deformation and retardation of fronts as they crossed the Alps. A prefrontal foehn was found to significantly increase the cross-frontal temperature gradients along the northern side of the Alps, resulting in fronts developing intense gravity current characteristics (Hoinka et al. 1990; Kurz 1990). In a statistical analysis of 6 years of frontal passages over the Appalachians, Schumacher et al. (1996) found that the Froude number was not useful in determining the degree of frontal retardation; they hypothesized instead that the strength of the associated upper-level potential vorticity anomaly primarily determines the frontal movement across the Appalachians. Trier et al. (1990) documented the strengthening of a shallow cold front in the vicinity of high terrain along the east coast of Taiwan during the TAMEX experiment. Orographically influenced cold fronts have also been investigated along New Zealand (Smith et al. 1991), eastern Australia (Baines 1980; Colquhoun et al. 1985), east of the Tibetan Plateau (Chang et al. 1983), and the Rocky Mountains (Hartjenstein and Bleck 1991; Colle and Mass 1995).

a. Structure and evolution of the cold front along the west side of the Olympics

At 2000 UTC 11 December 1993 (Fig. 6a), the surface cold front was making landfall along the Washington coast. The frontal passage at UIL, on the northwest coast, was associated with a 2°C 1-h drop in temperature and a rapid windshift from south-southeasterly (at ∼5 m s⁻¹) to weak northwesterly. A low center along the front was located approximately 35 km northwest of AST and was moving northeastward. An area of lower pressures was also observed to the northeast (the lee) of the Olympics.

Figure 6b shows the conventional observations at 2100 UTC 11 December 1993 as well as flight-level data from the P-3 at 200 m MSL between 2012 and 2050 UTC. The surface front at this time was located over the western foothills of the Olympics. The low center along the front was making landfall on the southwest Washington coast, with strong surface winds (15–20 m s⁻¹) located south of the low center. The flight-level winds veered from weak southerlies to strong west-northwesterlies across the front in the central Strait of Juan de Fuca, and the temperatures at 200 m decreased from 7° to 5°C.

NOAA P-3 reflectivities at 750 m MSL indicate that there was little or no precipitation to the northeast of the higher terrain of the Olympics around 2030 UTC 11 December (Fig. 7; see Fig. 6b for domain location). The 750-m P-3 Doppler winds over the eastern entrance of the Strait of Juan de Fuca had a northerly component as a result of a weak lee trough to the north of the Olympics. The precipitation associated with the front in the Strait was rather amorphous and stratiform, with the heaviest area (25–35 dBZ) straddling and behind the front.

Figures 8 and 9 show the frontal wind and precipitation structures at 1000 and 1750 m along the windward side of the Olympics (the dashed-boxed region in Fig. 6b). At 1000 m (Fig. 8a), the front was wrapped around the northwestern side of the Olympics at 2057 UTC. The wind shift and precipitation band was more intense across the foothills of the western Olympics than observed over the western strait. The strongest prefrontal southerlies (10 m s⁻¹) were located immediately upwind of the southwestern corner of the Olympics, with the flow veering rapidly to northwesterly behind the front. A line of relatively intense precipitation echoes (35–45 dBZ) containing two distinct elongated cores (C2, C3) was situated immediately west of this wind shift line. The precipitation band was broader to the north (precipitation area C1), where the front intersected the high terrain, resulting in a superposition of frontal and orographic precipitation processes.

At 2110 UTC (Fig. 8b), the prefrontal southerly flow at 1000 m was slowly being pinched off as westerly flow accelerated up the Hoh River Valley (point B). Precipitation areas C1 and C2 had weakened at this time. By 2135 UTC (Fig. 8c), the wind shift at 1000 m over the southwestern corner of the Olympics had advanced farther inland and weakened. Although the frontal precipitation band became much less organized (no well-defined precipitation cores were evident at this time), widespread orographic precipitation (30–35 dBZ) was located within the post-frontal northwesterlies along the lower windward slopes of the Olympics at 1000 m.

At 1750 m (Fig. 9a), the three precipitation cores (C1, C2, and C3) were evident at the 2057 UTC analysis time. These precipitation cores were situated in an area of strong horizontal shear and diffluent southwesterly flow at this level. The frontal wind shift at this level was located along the coast and was less sharp than at 1000 m. The prefrontal south-southwesterlies and precipitation areas at 1750 m weakened over the Olympics as the front approached the barrier between 2057 and 2135 UTC (Figs. 9a,b).

Cross section AA′ shows the vertical structure of the front as it impinged upon the Olympics using the NOAA P-3 dual-Doppler radar (Fig. 10).¹ At 2057 UTC (Fig. 10a), the front had gravity current characteristics, as seen in previous studies of intense frontal zones (e.g., Carbone 1982; Bond and Fleagle 1985; Shapiro et al. 1985). Strong low-level convergence at the leading edge of the front resulted in a narrow updraft core (2 m s⁻¹) extending upward to 2.5 km, and an area of heavy precipitation was located immediately behind the low-level wind shift. The front sloped upward to approximately 2 km MSL near the coast (1:25 frontal slope), consistent with UIL’s sounding at 2100 UTC, which also shows the transition from northwesterlies to southwesterlies around this level (800–750 mb) (Fig. 3b). The advection of precipitation particles to the rear of the frontal rainband and the gradual ascent above the frontal surface produced a broad area of stratiform precipitation behind the surface front (Fig. 10a). Meanwhile, prefrontal subsidence (∼1 m s⁻¹) was occurring above the lee (west side) of the western Olympic foothills.

By 2135 UTC 11 December 1993 (Fig. 10b), the leading edge of the front was ascending the western slopes of the Olympics, and two areas of enhanced precipitation were evident. The first, associated with the leading edge of the front, had weakened compared to 2057 UTC, as had the associated vertical motions and surface convergence. A second area of intense stratiform precipitation with reflectivities greater than 35 dBZ was evident approximately 25 km to the west of the surface front above the lower foothills.

b. Structure and evolution of the cold front around the north, south, and east side of the Olympics

By 2300 UTC 11 December 1993, the southern section of the front had advanced past the Olympics and was moving toward southern Puget Sound (Fig. 11). Meanwhile, the portion of the front to the north of the Olympics had surged through the Strait of Juan De Fuca. The surface and flight-level (∼200 m) winds in the central strait had a significant cross-strait component, presumably because of frictional effects and low-level flow splitting around the northwest corner of the Olympics. Across the front the low-level flow veered rapidly from southwesterly to westerly.

Figure 12 shows reflectivities and winds within cross section BB′ taken from the western entrance of the Strait of Juan de Fuca to the northeast corner of the Olympic Peninsula. The vertical structure of the front changed considerably as it moved down the Strait of Juan de Fuca between 2030 and 2300 UTC 11 December. At 2030 UTC (Fig. 12a), the front sloped steeply to the west, with a broad area of mainly stratiform precipitation located behind the leading edge. Meanwhile, there was weak subsidence and a general lack of precipitation to the east of the front. By 2300 UTC (Fig. 12b), a shallow layer of westerly flow (<1500 m) had surged out ahead of the frontal surface aloft with a commensurate reduction in frontal slope. A large area of stratiform precipitation was located under the sloping frontal zone to the west. In contrast, there were weak convective cells immediately behind the low-level front to the east. The horizontal spacing of these bands was fairly uniform, and three of the four bands tilted slightly to the west with height as a result of the negative vertical shear at low levels.

Figure 13 shows a mesoscale analysis of the conventional synoptic observations at 0000 UTC 12 December 1993 as well as the Doppler winds and reflectivities at 500 m MSL from the NOAA P-3 between 2330 UTC 11 December and 0100 UTC 12 December. At this time, the northern section of the front was rounding the northeast corner of the Olympics. Associated with this part of the front was a rapid wind shift from modest southwesterlies to strong (∼20 m s⁻¹) northwesterlies. The highest reflectivities (30–35 dBZ) at 500 m were located immediately behind the wind shift. As observed by the lower-fuselage surveillance radar of the P-3 (not shown), this precipitation band did not propagate continuously around the northeast corner of the Olympics. Rather, it developed in situ approximately 20 km northwest of its 0000 UTC position as the front pivoted around the northeast of the Olympics. Over southern Puget Sound, the surface temperatures decreased approximately 2°C and the surface winds veered from southerly to southwesterly across the front; however, unlike the northern section of the front, there was no distinct banding of precipitation.

At 1000 m (Fig. 14a), a pronounced deformation zone extended northeast of the Olympics, with the wind shift displaced a few kilometers northwest of the wind transition at 500 m. The precipitation band at 1000 m was weaker within 10–15 km of the Olympics, presumably because of downslope effects. At 1750 m (Fig. 14b), the deformation axis was shifted 15–20 km farther to the northwest than at 1000 m, separating strong south-southwesterly (20–25 m s⁻¹) flow over the Puget Sound basin from westerly (10–15 m s⁻¹) flow exiting the strait over the northeast foothills of the Olympics. An area of weak winds extended 20 km downwind of the higher peaks of the Olympics at 1750 m. This wakelike structure is associated with significant anticyclonic/cyclonic relative vorticity at 1750 m (−1.2 × 10⁻⁴ s⁻¹ and 1.0 × 10⁻⁴ s⁻¹, respectively) and is reminiscent of the potential vorticity (PV) banners observed downwind of major barriers (Smith et al. 1997).

The transitions associated with the northern section of the front over Puget Sound can be seen more clearly by looking at the 1-s resolution flight-level temperature, horizontal wind, and vertical velocity from the P-3 aircraft along flight-track segment XY (see Fig. 13 for location) between 500 and 1000 m MSL (Fig. 15). When the aircraft was located downwind of the southeast corner of the Olympics (near point X), the winds were west-northwesterly around 20 m s⁻¹, and subsidence off the Olympics resulted in relatively low dewpoints. As the aircraft tracked northward to the east of the Olympics, the west-northwesterlies veered to southwesterly 10–15 km north of point X and the dewpoints increased rapidly (at ∼2350 UTC). The frontal transition to the northeast of the Olympics at 0005 UTC (see arrow in Fig. 15) was much more vigorous than observed over southern Puget Sound, where there was a gradual 1°–2°C temperature drop and a slow veering of the winds from southerly to westerly 10–20 km behind the front (not shown). In contrast, frontal passage to the northeast of the Olympics was associated with the southwesterlies veering rapidly to northwesterly and a temperature drop of 2°C within a few kilometers. Behind the leading edge of the cold air, the west-northwesterly winds increased and the temperatures continued to decrease. A significant region of upward and downward motion (±2 m s⁻¹) occurred immediately following frontal passage, which is reminiscent of the turbulent “wake” structures observed with other sharp cold frontal studies (Shapiro et al. 1985; Neiman et al. 1995).

Figure 16 shows NOAA P-3 Doppler winds and reflectivities for cross section CC′ (see Fig. 14 for location) across this frontal transition to the northeast of the Olympics at 2356 UTC 11 December 1993. The frontal surface below 1000 m sloped back to the west more gradually than when the front was impinging on the windward side of the Olympics at 2057 UTC (cf. Fig. 10a). Upward motion above the forward portion of the front resulted in a wide precipitation core (>25 dBZ) extending to 4.0 km. A shallow and more intense precipitation core (35–40 dBZ) was located 5–10 km behind the surface wind shift. A brief episode of intense graupel was encountered by the P-3 at 1000 m MSL when it penetrated a portion of this convective line around 0015 UTC. The melting level (bright band) was approximately 1500 m ahead of the front and subsequently dropped to 1000 m behind the front.

By 0100 UTC 12 December 1993 (Fig. 17), the southward advancement of the wind shift to the east of the Olympics had slowed, with the boundary located over central Puget Sound. At this time pressure rises and temperature falls were evident over the Cascades (not shown), suggesting that aloft the front had moved east of the Puget Sound. A pressure trough was situated along the eastern slopes of the Olympics, and cold air had encircled the Olympics. Subsidence off the higher peaks of the barrier resulted in 1°–2°C warmer temperatures over the central Puget Sound basin than to the north or south. At 1000 m, the southern edge of heavier precipitation was coincident with the leading edge of weak northwesterlies. There was a lack of precipitation over southern Puget Sound as a result of downslope flow off the southern Olympics. The flight-level temperatures at 1000 m were generally around 0°C, except for a small area downwind of the higher terrain, where the temperature spiked to 2°C and dewpoints dropped from −2° to −5°C. The general northwesterly flow downwind of the Olympics at 1000 m is consistent with the low-level Froude number of approximately 0.8 (Fr ∼ U/(Nh_m), from the UIL sounding at 0000 UTC (U ∼ 12 m s⁻¹, moist N ∼ 0.01 s⁻¹, and h_m ∼ 1500 m), which is too high for the development of leeside vortices. Overall, the wind and precipitation structures were developing characteristics of a Puget Sound convergence zone (Mass 1981; Chien and Mass 1997), with low-level convergence between surface southerlies and northerlies over central Puget Sound and a band of precipitation developing over northern Puget Sound.

Since the NOAA P-3 completed its mission shortly after 0100 UTC 12 December, the remainder of the analysis is limited to conventional synoptic observations. By 0200 UTC 12 December (Fig. 18), light surface easterlies (<3 m s⁻¹) had developed over central Puget Sound, with weak northwesterlies (2–5 m s⁻¹) and stronger southwesterlies (5–10 m s⁻¹) to the north and south of this convergence zone, respectively. Some stations in central Puget Sound were reporting light rain; therefore, a band of precipitation associated with low-level convergence was still present to the east of the Olympics. At this time, the postfrontal air mass had entirely engulfed central Puget Sound (temperatures were approximately 2°C lower than at 0100 UTC).

a. Model description

The PSU–NCAR MM5 was used in nonhydrostatic mode to simulate this case and to provide additional data for diagnosing the frontal interaction with the Olympics and the subsequent development of the Puget Sound convergence zone. The simulation used the explicit moisture scheme of Hsie et al. (1984), with improvements to allow for mixed liquid-ice phase below 0°C and graupel microphysics (Grell et al. 1994), and the Kain–Fritsch cumulus parameterization (Kain and Fritsch 1990) was applied, except for the inner domain where convective processes could be resolved explicitly. The planetary boundary layer was parameterized using the scheme of Zhang and Anthes (1982). Klemp and Durran’s (1983) upper-radiative boundary condition was applied in order to prevent gravity waves from being reflected off the model top.

For this simulation, stationary 3- and 9-km domains were nested within a 27-km domain using one-way interfaces (Fig. 19a). The model top was set at 100 mb. Thirty-eight unevenly spaced full-sigma levels were used in the vertical, with the maximum resolution in the boundary layer.² Five-minute averaged terrain data were analyzed to the 27- and 9-km model grids using a Cressman analysis scheme and filtered by a two-pass smoother/desmoother. For the 3-km domain, a 30-s topography dataset was interpolated to the grid in order to better resolve the Olympic Mountains (Fig. 20). A 10-min land use dataset was used to initialize the various surface categories for all the domains; however, the coastlines for the 9- and 3-km domains were enhanced by setting grid points to either coniferous forest or water when the 30-s topography at those grid points is greater than or equal to zero, respectively. Initial atmospheric conditions and sea surface temperatures were generated at 0000 UTC 11 December 1993 for the 27-, 9-, and 3-km domains by first interpolating the National Centers for Environmental Prediction (NCEP) global analyses (2.5° lat × 2.5° long resolution) to the model grid. These analyses were improved by incorporating surface and upper-air observations using a Cressman-type analysis scheme (Benjamin and Seaman 1985). Additional analyses generated in the same manner every 12 h were linearly interpolated in time in order to provide the evolving lateral boundary conditions for the 27-km domain.

The simulation of this IOP proved to be very difficult. Figure 19a shows the sea-level pressure, 950-mb temperature, and surface winds³ for the 27 km (D1) domain 12 h into the simulation at 1200 UTC 11 December 1993 (12 h). All the NCEP operational models and our first attempts at a control run had the low center making landfall along the northern Oregon coast by 1200 UTC 11 December (landfall was approximately 150 km too far south and 8 h too fast), and the frontal boundary to the north of the cyclone developed inland to the east of the Olympics rather than slightly offshore as observed (Fig. 4). Because of the lack of data offshore around the cyclone west of the Oregon and Washington coast, applying four-dimensional data assimilation on the 27-km domain during the first 12 h of the integration made the simulation worse (only a 998-mb trough developed along coast). Overall, this case illustrates the initialization problem over the Pacific Ocean, which greatly affects short-term numerical forecasts along the West Coast. Since it was imperative for this study of front–terrain interactions that development of the north–south frontal boundary occur just offshore rather over the Olympics, the revised “control” simulation shown below used initial conditions in which the 0000 UTC 11 December 1993 surface and upper-air analyses on pressure levels were shifted slightly westward (135 km or 5 grid points for the outer domain) before interpolating to sigma levels. Though admittedly ad hoc in nature, this slight modification had the positive effect of rendering the simulation more realistic without introducing any unwanted perturbations.

b. Simulation of the front interacting with the Olympics

Figure 19b shows sea level pressure, 950-mb temperature, and surface winds at 1200 UTC 11 December 1993 (12 h) for the 27-km simulation with the altered initialization. As observed (Fig. 2b), a developing low pressure center (994 mb) was located approximately 150 km west of the northern Oregon coast at this time, with a trailing surface trough extending southward. A low-level weak baroclinic zone and associated pressure trough were advancing northward along the Washington coast. Between 1500 and 1800 UTC this trough became situated just offshore of the Washington coast, and subsequently propagated eastward as a weak cold front.

Even given the aforementioned modification of initial conditions, the simulated progression of the cold front toward the Olympics was 1–2 h too fast; therefore, in order to compare the simulated structures to observations, the displayed model output time will be applicable to the simulation time plus 1.5 h (i.e., hour 22.5 of the simulation corresponds to 0000 UTC 12 December 1993). At 1800 UTC 11 December 1993 on the inner D3 grid (Fig. 21a), there was no distinct coastal low center in the simulation. Rather, an elongated trough extended southward from the northwest tip of Washington, with sea level pressures 2–3 mb lower than observed over much of the domain. Most of the heavy precipitation in the model was falling over the higher terrain rather than along the front.

The front neared the foothills of the western Olympics by 2100 UTC 11 December 1993 (Fig. 21b), which corresponds to the time the NOAA P-3 began executing the first Doppler leg along the western side of the barrier (cf. Fig. 8a). Although the observed frontal wave (cf. Fig. 6b) was not simulated along the southwestern Washington coast, the pressure gradients and surface winds were fairly realistic around the Olympics. South-southeasterly surface winds ahead of the front veered rapidly to northwesterly behind the front. The temperature gradient at 500 m had intensified during the previous few hours as the front moved into the foothills, and a significant line of precipitation had developed along the front near the foothills. Meanwhile, the simulated precipitation was more amorphous and less intense over the western Strait of Juan De Fuca, consistent with the P-3 Doppler observations (Fig. 7).

At 2200 UTC 11 December 1993 (Fig. 21c), the front was moving over the high terrain of the Olympics, and the surface winds immediately ahead of the front had become more southwesterly. As a result of weakened surface convergence, the frontal precipitation decreased as the front moved over the barrier, a trend that was also observed by the NOAA P-3 (cf. Figs. 8a–c). Meanwhile, a surge of low-level westerlies developed to the north of the Olympics along the Strait of Juan De Fuca, where the frontal temperature and pressure gradients were enhanced as the front approached the area of lee troughing over the eastern strait. A band of precipitation developed immediately behind the leading edge of the front over the strait.

By 2300 UTC 11 December 1993 (Fig. 21d), low-level westerlies had surged through the strait, with a sharp wind shift from southerly to westerly along the front’s leading edge; little precipitation was located in this area of strong low-level convergence, perhaps due to leeside subsidence. The temperature gradient associated with the front had also weakened to the northeast of the Olympics. Lee troughing was beginning to extend southward along the eastern slopes of the Olympics as the incident lower-tropospheric flow veered to the northwest. As observed, the surface front to the south of the Olympics was relatively weak as it approached southern Puget Sound.

To document the structural evolution of the simulated front as it approached the Olympics, Fig. 22 shows a cross section (DD′) across the windward side of the barrier. At 1800 UTC 11 December (Fig. 22a), the front had just crossed the coast, with little if any organized precipitation along its leading edge. The prefrontal flow below 900 mb had an offshore component. By 2100 UTC 11 December (Fig. 22b), the front was along the western Olympic foothills. The temperature gradient and precipitation rates had increased, and the flow had become more westerly behind the front at low levels (stronger flow in the cross section). Even though there was a well-defined band of precipitation and wind shift associated with the front in the simulation, these features are not as intense as the density-current-like structures seen in the observations (cf. Figs. 8a, 10a). At 2200 UTC 11 December (Fig. 22c), as the front ascended the western slopes of the Olympics, the horizontal temperature gradient at the leading edge of the front weakened, and the frontal precipitation was less organized. The heaviest precipitation had shifted over the peaks of the barrier and to the west where postfrontal upslope flow resulted in a large area of orographic precipitation.

Cross section EE′ shows the evolution of the front as it advanced through the Strait of Juan De Fuca between 2100 and 2300 UTC (Fig. 23). At 2100 UTC (Fig. 22a), the low-level temperature gradient associated with the front over the western strait was greater than over the western slopes of the Olympics at this time (Fig. 22b). Two relatively weak precipitation bands were present near the leading edge of the front. The significant depression in the isentropes and weak flow below 850 mb over the central and western strait were indicative of the subsidence wake off the Olympics. The subsidence warming did not reach the surface, as evinced by the stable layer in the lowest 25 mb. By 2200 UTC (Fig. 23b), the temperature gradient, depth of the cold air, and the precipitation intensity at the leading edge of the frontal transition had increased significantly as the front advanced into the region that had been warmed by downslope subsidence. The front accelerated eastward and became substantially more shallow by 2300 UTC 11 December 1993 (Fig. 23c), and the precipitation weakened near the leading edge. The sinking motion behind the front shown here and in the P-3 observations (cf. Fig. 12b) was likely enhanced by low-level acceleration and diffluence over the eastern strait. Overall, the shallow headlike structure of the front is similar to that of a gravity current as seen in laboratory experiments (Simpson 1969). Both the shallow nature of the front and the general lack of widespread precipitation over the eastern strait agree well with the NOAA P-3 Doppler cross section (Fig. 12b).

As observed (Fig. 13), low-level blocking split the front around the Olympics by 0000 UTC 12 December 1993 (Fig. 24a). The northern segment of the front, which was advancing around the northeast corner of the Olympics, had a much stronger wind shift and temperature gradient than the southern section over southern Puget Sound. An intense line of precipitation had developed along the northern section as the low-level convergence increased, while west-northwesterly flow at crest-level favored rain shadowing to the south of this band. Weaknesses of the simulation at this time include excessive rain shadowing to the east of the Olympics and simulated 500-m temperatures approximately 2°C too warm over much of Puget Sound compared to the NOAA P-3 temperature trace (Fig. 15).

During the next 2 h surface northerlies and precipitation pushed southward toward central Puget Sound while the southern section of the front moved into the Cascades (not shown). As observed at 1000 m (cf. Fig. 17), the leading edge of the precipitation band at 0100 UTC 12 December 1993 was situated within a mountain wake that extended to the east of the Olympics. By 0200 UTC 12 December 1993 (Fig. 24b), the precipitation band narrowed and stalled over central Puget Sound. Surface easterlies had developed over Puget Sound from the western Cascade foothills eastward to the Olympics. A pressure ridge extended northward along the west side of the Cascades, while lee troughing along the east side of the Olympics decreased in amplitude (by 2–3 mb) as a result of a diminishing mountain wave (not shown). By 0300 UTC 12 December 1993 (not shown), the wind shift associated with the convergence zone was weakening (winds becoming more southeasterly over central and northern Puget Sound) and the precipitation band broadened and drifted toward the Cascades. This weakening convergence in central Puget Sound occurred even though the upstream surface winds at the coast changed little.

a. Frontal evolution along the windward and north side of the Olympics

There have been a number of recent theoretical studies examining the interaction of fronts with topography (Williams et al. 1992; Gross 1994; Li et. al 1996). Overall, the consensus from these studies has been that fronts weaken as they move up windward slopes and strengthen on the leeward side of barriers. A primary objective of this COAST IOP5 study was to determine whether the frontal interaction with the Olympics had characteristics similar to previous theoretical studies and to use a very detailed dataset to evaluate previously discussed mechanisms.

Even though the thermal contrasts across the front were not directly observed over the Olympics, the NOAA P-3 Doppler analyses suggest that the front weakened as it moved up the windward slopes between 2100 and 2200 UTC (Fig. 10). Since detailed observations were not available before the front made landfall, the high-resolution simulation was used to fill in some of the details of the frontal interaction with the Olympics. Using the model data at 15-min time intervals, the various terms that can change the magnitude of the horizontal gradient of the potential temperature following an air parcel were evaluated on pressure levels using a simplified form of the Miller frontogenesis equation (Miller 1948):

View Expanded

In (1), the three expressions on the right side are the stretching deformation (A), shearing deformation (B), and tilting terms (C), respectively. Diabatic contributions to frontogenesis could be significant, but they are neglected since only direct topographic effects on the wind field are considered in this discussion. To assess low-level frontogenesis, the individual terms were averaged over the lowest 6 half-sigma levels (approximately 800 m layer above the surface) between 1800 and 2345 UTC 11 December 1993 as the front made landfall along the Washington coast and moved up the windward slopes of the Olympics. Figure 25 shows the traces of the various terms as well as total frontogenesis for a 6 × 6 gridpoint box (18 km × 18 km) surrounding the front (see box in Fig. 21a). Figure 25 also shows a trace of the average topography and the average magnitude of the horizontal potential temperature gradient within this moving box.

Between 1800 and 1900 UTC 11 December 1993, the front was slowly weakening as it made landfall along the Washington coast. Most of this weakening was the result of the tilting term, which increased as a result of greater upward motion on the warm side of the baroclinic zone. During the next hour, the front strengthened as it approached the foothills. The frontogenesis calculations suggest that the intensification of the front primarily came from two sources. First, flow splitting around the Olympics resulted in increased horizontal stretching deformation (cf. Fig. 21a). Second, the flow deflection around the Olympics resulted in weak prefrontal subsidence above the lee (west side) of the western Olympic foothills, which reduced the negative contribution by the tilting term while the front approached the foothills. As the front climbed the windward slope, its temperature gradient weakened primarily due to the tilting term. As seen in Fig. 21c above, by 2200 UTC 11 December the flow ahead of the front turned more southwesterly, which favored upslope flow (and adiabatic cooling) ahead of the front and reduced stretching deformation and a more negative tilting term. When the front approached the crest (around 2300 UTC) the temperature gradients began to increase again as the downward motion associated with the mountain wave resulted in a more positive tilting term.

Overall, these results suggest that the prefrontal flow splitting and orographically induced vertical motions around the Olympics have an important role in modulating the frontal intensity along the windward side. For COAST IOP5, flow splitting of the south-southwesterly flow ahead of the front was favored since the Froude number U/(Nh_m) was less than 1; the average Froude number between the surface and 900 mb for the southerly flow ahead of the front was approximately 0.67, where U ∼ 10 m s⁻¹, h_m ∼ 1500 m, and N ∼ 0.01 s⁻¹.

To demonstrate that enhanced frontogenesis in the foothills depends strongly on the amount of flow splitting ahead of the front, another simulation (HOLYM experiment) was completed in which the height of the Olympics was reduced by half (h_m ∼ 750 m). The Froude number for this case was 1.34, which resulted in less flow splitting (not shown), and reduced adiabatic cooling/warming induced by the barrier. Figure 26 shows the traces of terrain height, temperature gradient, and frontogenesis for a box moving with the front in the HOLYM simulation. As in the control simulation, the front weakened as it made landfall as a result of the tilting term. However, unlike the control run, there was much weaker frontogenesis as the front approached the foothills since the flow splitting and blocking ahead of the front was reduced. The HOLYM experiment also demonstrated that the front in the control simulation was retarded slightly by the Olympics; the front moved up the barrier approximately 10% faster in the HOLYM simulation than in the control run (the front reached the crest at 2230 UTC in the HOLYM run 2300 UTC in the control simulation).

An important feature of the cold front evolution was its interaction with the prefrontal downslope warming to the north of the Olympics. Both the P-3 observations and MM5 simulation indicated that this was a highly frontogenetical location. As shown by the simulations, the prefrontal downslope flow provides a warm anomaly and light winds, which enhances horizontal confluence and the temperature gradient. Chien (1997) also noted this prefrontal warming and frontogenesis to the north of the Olympics during COAST IOP2. In addition, a similar evolution has been observed around much larger barriers such as the Rockies and Alps. For example, Colle and Mass (1995) noted that northerly cold surges along the Rockies are frequently preceded by lee-troughing and downslope warming that enhances the temperature contrast across the transition. Similarly, the prefrontal foehn to the north of the Alps can increase the frontal temperature gradient, which can lead to the front having density current characteristics (Hoinka et al. 1990).

To determine whether the front intensified and accelerated down the Strait of Juan de Fuca because of gap flow effects, a simulation was completed in which Vancouver Island was removed and replaced by water. Compared to the control run, there was little difference in the structure, intensity, and speed of the front to the north of the Olympics (not shown). However, when the water to the north of the Olympics was replaced by flat land, the frontal temperature gradient decreased by 30%–40% below 925 mb (not shown); therefore, the reduction in surface drag over the water to the north of the Olympics (Strait of Juan de Fuca) and not the existence of a gap was evidently important for maintaining and enhancing the front.

As the front rounded the northeast corner of the Olympics, the associated three-dimensional airflows were highly complex. Figure 27 shows 6-h backward trajectories terminating at points within cross section EE′ (Fig. 23c) using 15-min model output (interpolated to 5 min) for the 3-km domain. Trajectories 1 and 2, which terminated behind the front below 950 mb, were associated with low-level westerly flow through the Strait of Juan de Fuca. These trajectories sank approximately 20 mb to the north of the Olympics as the postfrontal flow became more shallow. Trajectory 3 (ending at ∼910 mb) began in low-level northwesterly flow and was subsequently deflected toward the strait by the windward pressure ridge. Trajectory 4 (starting at approximately 850 mb), ending above the low-level westerly flow, had its origins to the south of the Olympics, as prefrontal southerly flow that was deflected around the west side of the barrier. Trajectory 5, ending at approximately 875 mb immediately above the front, also originated to the south of the Olympics before passing directly over the barrier. As depicted by trajectories 5 and 6, which ended immediately above the front, the warm tongue and general lack of organized precipitation near the front was the result of air that descended off the northeast corner of the Olympics. Low-level trajectories 7 and 8 ended ahead of the front and were associated with the prefrontal southerly flow that passed to the east of the Olympics. Overall, there was substantial confluence between low-level trajectories 1, 2, and 3 originating near the western entrance of the strait and trajectories 6 and 7 originating to the south of the Olympics.

b. Frontal evolution to the east of the Olympics

With westerly flow and significant static stability near crest level (Fig. 22c), mountain wave amplification and lee troughing were observed along the eastern slopes of the Olympics (Fig. 24a). As a result, the low-level portion of the front that passed over the barrier was destroyed by leeside subsidence. To show that the front would not split as much around the Olympics when the Froude number doubled, Fig. 28 shows the HOLYM simulation at 0030 UTC 12 December 1993 (Froude number of 1.34). Compared to the control simulation (Fig. 24a), the front in the HOLYM experiment maintained more integrity as it passed over the Olympics. Since downslope warming ahead of the front to the north of the Olympics still favored some frontogenesis (not shown), the northern portion of the front remained stronger than the section of the front to the southeast of the barrier.

The Olympics clearly had a profound influence upon the southward movement of the front and associated precipitation into central Puget Sound; however, how important were diabatic effects in maintaining the frontal structure as it pushed southward? Several investigators have illustrated that latent cooling processes located behind a front (sublimation and melting of ice and evaporation of rainwater) can be important in maintaining low-level frontal intensity (e.g., Huang and Emanuel 1991; Parker and Thorpe 1995; Barth and Parsons 1996).

To address this question, a simulation was completed in which the precipitation was allowed to occur but the diabatic cooling processes associated with precipitation were turned off in the model. This NLATC simulation was completed by restarting the control (CTL) simulation at 2200 UTC 11 December (20.5 h into the run) with the diabatic cooling turned off. Figure 29a,b shows cross section FF′ for the CTL and NLATC simulations, respectively, at 2330 UTC 11 December 1993. Compared to the CTL run (Fig. 29a), the front and associated precipitation band were much weaker in the NOLATC simulation (Fig. 29b). The temperatures behind the front below 800 mb were 1°–2°C warmer in the NLATC simulation, with no evidence of a local minimum in the near-surface potential temperature field. In both the CTL and observations, the precipitation band and front weakened as it pushed southward into central Puget Sound (cf. Figs. 24a); however, diabatic cooling provided by the precipitation processes delayed this weakening by maintaining the cold pool.

c. Puget Sound convergence zone evolution

As seen in the observations (Fig. 18) and the 3-km model simulation (Fig. 24b), the wind shift associated with the weakening front became stationary in central Puget Sound, which resulted in a more canonical PSCZ structure by 0200 UTC 12 December 1993, with surface easterlies and an east–west precipitation band across Puget Sound.

There are some subtle and important differences between this PSCZ and a late spring event described in Chien and Mass (1997). In their case (as well as many other PSCZ events), the development of the PSCZ occurred 3–9 h after frontal passage since it took a while for the winds at crest level and along the Washington coast to veer to west-northwesterly behind the frontal zone. For the IOP5 case, the front developed PSCZ characteristics as it progressed southward (postfrontal flow splitting around the Olympics resulting in an east–west band of precipitation across central Puget Sound). The Froude number behind the front for this case was approximately 0.8, compared to around 0.4 for the event documented in Chien and Mass (1997). As a result, unlike the PSCZ in Chien and Mass, there were no leeside vorticies between 0.5 and 1-km mean sea level (MSL) in either the observations or the 3-km simulation.

In both the observations and model simulation for this case, the PSCZ did not last more than a few hours over central Puget Sound even though the upstream flow at crest level and along the coast did not change appreciably. However, an aircraft sounding taken at around 2300 UTC approximately 100 km upstream of the Olympics (see Overland and Bond 1995, Fig. 8) showed that the upstream static stability decreased substantially as the cold air deepend (moist N ∼ 0.007 s⁻¹ in the observed sounding and 0.008 s⁻¹ in our simulation). Therefore, the Froude number over the Olympics likely increased from 0.8 during and shortly after frontal passage to approximately 1.1 a few hours later. As a result, the PSCZ weakened since the potential for flow splitting around the Olympics was reduced. Furthermore, the higher Froude number flow resulted in much less wave amplification over the Olympics (not shown), which reduced the surface troughing and associated low-level convergence in the lee of the Olympics.