1. Introduction
A significant challenge today is the prediction of orographic precipitation in regions of complex terrain. Improvements in numerical weather prediction models have allowed for significantly higher-resolution mesoscale forecasts, facilitating the explicit treatment of cloud processes and precipitation, and capturing important small-scale terrain features. These improvements should allow for more realistic forecasts of precipitation in regions of complex terrain. In particular, terrain geometry can have important influences on the spatial distribution and intensity of orographic precipitation. However, our understanding of the behavior of orographic flows in complex terrain and their influence on orographic precipitation is incomplete. This study examines some sensitivities of orographic precipitation to changes in terrain geometry and how those sensitivities are compounded by small variations in upstream conditions.
Orographic precipitation can form via a number of mechanisms as determined by the airflow velocity and direction, the thermodynamic conditions, and the shape and dimensions of the orography [for a review, see, e.g., Smith (1979), Roe (2005), and Smith (2006)]. Despite a wealth of documentation (e.g., Baines 1995; Smith 1989) and study (e.g., Pierrehumbert and Wyman 1985; Baines 1987; Smith 1989; Rotunno and Ferretti 2001; Colle 2004), the uninvestigated parameter space for orographic precipitation remains vast. Recent efforts have attempted to bridge the gap between often oversimplified idealized studies and more complicated case studies. For example, Rotunno and Ferretti (2001) recreated the mechanisms responsible for the Piedmont flood of 1994 using a semi-idealized arc-shaped ridge that represented the general shape of the Italian Alps. A similar experiment was undertaken by Schneidereit and Schär (2000). Zängl (2008) used a large plateau-like mountain ridge to generate a seeder cloud via large-scale lifting, and then superimposed atop the large ridge a smaller hill to generate orographic feeder clouds. Jiang (2006) performed a systematic study on the influence of a concave ridge on precipitation intensity and distribution, motivated by the observed funneling effect in the European Alps where a series of concave and convex ridges modifies moist flows from the Mediterranean Sea.
These studies have motivated the following questions: How does terrain geometry influence orographic precipitation, and how do these terrain features influence the sensitivity of orographic precipitation to variations in the upstream conditions?
A major impediment for predicting orographic flow modifications is the effect of latent heating. Latent heating only affects flow dynamics when the air becomes saturated; however, it is impossible to state a priori where saturation will occur (Rotunno and Ferretti 2001). The theory of orographic flow modification is therefore often considered in the absence of latent heating (i.e., dry dynamics). Without friction and rotation, the nonlinearity of a two-dimensional flow can be represented by the nondimensional mountain height 
In three dimensions, the addition of the spanwise-to-streamwise horizontal aspect ratio of the orography β can be of great importance. Smith (1989, see his Fig. 5) constructed a flow regime diagram to represent the relationship among 
Experimental and numerical studies (e.g., Pierrehumbert and Wyman 1985; Smolarkiewicz and Rotunno 1989; Durran 1990; Baines 1995; Colle 2004; Hughes et al. 2009) have shown that for 
The shape of the mountain can also have a significant influence on the intensity and distribution of precipitation. In the case of a concave ridge, the forward-reaching ridge arms inhibit flow diffluence upstream of the ridge and intensify the high pressure perturbation that develops on the windward slope (Jiang 2006). This effectively extends the influence of the barrier farther upstream and strengthens flow deceleration between the ridge arms. Jiang (2006) performed a systematic study on the influence of a concave ridge with flows of varying 
The impact on orographic precipitation of varying 
For precipitation to form, water vapor must be present in the atmosphere. Thus, to determine the behavior of an orographically modified flow with moisture, the effect of latent heat release through saturation must be considered. Durran and Klemp (1982) and others have shown that when an atmosphere becomes saturated, the atmospheric static stability is best described by the moist Brunt–Väisälä frequency Nm [as defined by Lalas and Einaudi (1974)]. This can be incorporated into 
One motivation of this study is to understand the dynamics underlying the important orographic precipitation events in southeastern Australia, a region that has received relatively little attention. The Australian Alps are the highest section of Australia’s Great Dividing Range, comprising the only peaks on the continent that exceed 2000 m. The Australian Alps have significant influence on the weather across the highly populated southeastern seaboard, and the complex network of V-shaped valleys and elevated plateaus generate significant orographic precipitation events throughout the year. Prefrontal northwest (NW) flows play an important role in these events. As described by Sturman and Tapper (1996, 217–219), prefrontal NW flows and NW Australian cloud bands (Tapp and Barrel 1984) often cause heavy, persistent rainfall over southeastern Australia. The prefrontal flow impinges upon the Australian Alps and is orographically modified, resulting in the initiation or enhancement of orographic precipitation on the windward slopes. When a preexisting cloud band is present, it is usually stratiform. Moreover, these frontal events have been shown to be responsible for a large proportion of the precipitation in this region (e.g., Wright 1989; Landvogt et al. 2008).
Idealized studies on orographic precipitation typically adopt an atmosphere that is initially saturated (or nearly saturated) in the lower levels (e.g., Colle 2004; Miglietta and Rotunno 2005; Jiang 2006; Galewsky 2008). This helps remove uncertainty about where saturation will occur, if at all. However, this approach is not necessarily representative of reality, where the upstream flow is often unsaturated and only becomes saturated through forced ascent. While this study examines idealized terrain shapes, the upstream environment is based on an observed prefrontal Australian event that is moist but initially unsaturated. In one way this is more realistic, but it complicates the scenario because of the mixture of both dry and moist dynamics [see, e.g., Reeves and Rotunno (2008) for discussion on the influence of upstream humidity variations on the orographic flow response].
The aims of this study are to investigate the following: How do specific terrain features influence orographic precipitation, how sensitive is orographic precipitation to variations in the upstream wind direction, and how do specific terrain features modify this sensitivity? Section 2 describes the model characteristics and numerical setup of the experiments reported. Section 3 describes the dynamics of a moist flow impinging upon different alpine-scale terrain shapes, presenting the impact of changing the ridge length and vertex angle. Section 4 presents results on the influence of small variations in the upstream wind direction, including results from a small ensemble, and the sensitivity of precipitation to changes in the upstream thermodynamic profile. Section 5 summarizes these results.
2. Model configuration
a. The numerical model
Version 3.1.1 of the Advanced Research Weather Research and Forecasting (ARW-WRF) Model (Skamarock et al. 2008) was used to perform several sets of idealized simulations. The WRF Model is nonhydrostatic, nonlinear, and fully compressible. The simulations have a single domain with 400 grid points in the x and y directions and 70 grid points in the z direction. The 2.5-km horizontal grid spacing establishes a relatively large 1000 km × 1000 km domain, necessary to prevent any spurious boundary reflections that could interfere with flow near the terrain. The model top is prescribed at 25 km, with a vertical grid spacing that is approximately 250 m in the troposphere, stretching to 800 m at the model top. The uppermost 5 km features a Rayleigh damping layer to prevent reflections, and the lateral boundary conditions are open. Microphysics is parameterized using the Thompson scheme (Thompson et al. 2004)1 and subgrid mixing is parameterized using a prognostic (1.5-order) turbulence kinetic energy closure. The horizontal and vertical advection schemes are fifth order and third order, respectively. For simplicity there is no surface friction, radiation, or Coriolis effects.
Each model simulation is integrated for 20 h and for most cases a steady state is achieved after 6–10 h; however, some cases feature nonlinear flow attributes (such as shedding vortices) isolated to the lee. Model snapshots for all simulations are examined at 16 h and the average hourly accumulated precipitation is measured between 14 and 20 h.
b. The idealized terrain
Figure 1 presents the idealized ridge, comprising a center section at the vertex Y1 connected on either side to a pair of straight ridges Y2 that are dampened at the ridge ends. The center section is either straight (α = 0) or parabolic (α ≠ 0), and the pair of straight ridges are parallel to the y axis (α = 0), tilt forward (α > 0), or tilt backward (α < 0). Other parameters that determine ridge geometry are hm, the maximum ridge height; b, the ridge half-length measured in the y direction; and a, the ridge half-width measured in the x direction. Table 1 presents the suite of terrain (and other experimental) variables used in this study. For all cases, a maximum ridge height of 1650 m is used, consistent with the profile of the Australian Alps.
The idealized ridge used in this study as described by the equation in section 2b. (top) Two ridges: a concave ridge with α = 45° and a straight ridge with α = 0° for a flow originating from the negative x direction. (bottom) Cross section along the centerline of the concave ridge (identical for all ridges). See Table 1 for a list of terrain shapes.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
Summary of parameters used in the idealized experiments. The experimental constants are maximum ridge height, hm = 1650 m; ridge half-width, a = 25 km; upstream wind velocity = 20 m s−1; and upstream stability, N = 0.0121 s−1.
Three values of α are initially employed: α = 0°, α = 45°, and α = −45°. When α = 0° the ridge is straight (ST) and b is the distance along the ridge top from the center of the ridge to the 
c. Atmospheric sounding
The model is initialized with the sounding shown in Fig. 2. This control sounding is based on an observed sounding from Wagga Wagga, New South Wales (35.2°S, 147.7°E) at 1000 local time (LT) 3 August 2005, approximately 50 km northwest of the Australian Alps. This was during the passage of a prefrontal NW flow that produced heavy precipitation in the Australian Alps (see Landvogt et al. 2008); in some regions the accumulated peak was more than 70 mm in 24 h (Falls Creek, 36.9°S, 147.1°E).
The atmospheric sounding (skew T–logp) used in all control simulations. It is representative of a real sounding recorded in Wagga Wagga during the passage of a northwestern Australian cloud band (see section 2c for details).
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
Modifications were made to the original sounding to simplify the interpretation. The wind profile was modified to be unidirectional (directed along the x axis) with a uniform velocity of 20 m s−1. This profile is comparable to the original Wagga Wagga sounding that measured a maximum wind speed below ridge height of 18 m s−1, although it removes the influence of vertical shear, which can be important (see, e.g., Colle 2004). The atmosphere is stable to parcel ascent and the lifting condensation level (LCL) for a surface parcel is approximately 750 m and approximately 2000 m for an upstream parcel at ridge height (1650 m). A surface layer temperature inversion, presumably a remnant of nocturnal cooling, was removed. The low-level static stability N is nonuniform, which complicates the evaluation of 
The initial sounding, which represents the upstream conditions through all simulations, is unsaturated. As the moist flow impinges upon the ridge, saturation occurs over the windward slope. In these saturated regions, Nm is approximately half the value of N, making 
Plan view of ST-LR with flow originating from the negative x direction showing surface streamlines (solid contours) and perturbation x velocity [dotted contours, contour interval (CI) = 2 m s−1] at 16 h for (a) dry and (b) moist atmospheres. Terrain contours are at an interval of hm/4.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
3. Sensitivity to ridge geometry
a. Long ridges (β = 4.0)
1) Straight long ridge (ST-LR)
The results from ST-LR (with moisture) will serve as preliminary findings to inform results from the other terrain shapes. Figure 4a shows the surface streamlines and the x component wind velocity u perturbation for ST-LR. As seen in the previous section, the impinging moist flow ascends the ridge without significant lateral deflection, and windward flow deceleration is relatively weak; the minimum perturbation u is approximately −5 m s−1 and the region of decelerated flow has an upstream extent of around 125 km (measured from the ridge crest to the −2 m s−1 perturbation u contour). The windward surface streamlines indicate movement away from the centerline, consistent with a widespread region of flow diffluence upstream of the ridge (Fig. 4b). Here, flow diffluence is defined as ∂υ/∂y and is positive in regions where streamlines separate. Equivalently, flow confluence corresponds to regions where streamlines move toward each other (∂υ/∂y < 0) and is present near the ridge ends in Fig. 4b (hatched).
Selected fields for ST-LR. (a) Plan view of surface streamlines (solid contours) and perturbation u (dotted contours, CI = 2 m s−1); (b) plan view of average hourly accumulated precipitation (shaded, CI = 1 mm from 0.5 mm) and confluent flow (hatched), where Pmax is the maximum average precipitation (mm) and Ptot is the total hourly precipitation mass (109 kg); and (c) vertical cross section along the centerline of potential temperature (solid contours, CI = 2 K), positive vertical velocity (shaded, CI = 0.1 m s−1, black represents > 1 m s−1), perturbation u (dotted contours, CI = 2 m s−1), and cloud water mixing ratio (dashed contours, CI = 0.2 g kg−1). Terrain contours in (a) and (b) are at an interval of hm/4. Model snapshots for all simulations are examined at 16 h and average precipitation is measured between 14 and 20 h.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
Flow deceleration and vertical motion are two factors that can facilitate the amount and distribution of orographic precipitation by 1) determining the depth of orographic cloud; 2) changing the time scale for a parcel to move across the windward slope, altering the precipitation efficiency; and 3) changing the parcel trajectory (Jiang and Smith 2003). The weak flow deceleration and negligible lateral deflection in ST-LR effectuates a relatively uniform distribution of precipitation across the upper ridge slope (Fig. 4b). Cloud water extends to approximately 3-km altitude over the windward slope (Fig. 4c), generating a maximum average hourly accumulated precipitation at the center of the ridge of 3.6 mm. A small region of snow and ice develops downstream of the ridge above 4 km; however, it is not associated with any surface precipitation. The potential temperature θ contours in Fig. 4c reveal that mountain waves are generated by ST-LR that steepen in the lee, consistent with 
2) Concave long ridge (CC-LR)
Figure 5 presents results from CC-LR with α = 45°. Windward flow deceleration is significantly stronger in CC-LR, with a minimum perturbation u of −16 m s−1 and an upstream extent of 250 km from the ridge vertex to the −2 m s−1 perturbation u contour (Fig. 5a). This is consistent with the hypothesis that the forward-reaching arms of a concave ridge inhibit flow diffluence upstream of the ridge centerline, which effectively expands and strengthens windward flow deceleration. Far upstream of the ridge is weak, widespread diffluent flow and between the two ridge arms is a zone of confluent flow, indicated by the hatched region in Fig. 5b. This flow confluence is indicative of and fundamental to the precipitation-enhancing “funneling mechanism” discussed by Jiang (2006).
As in Fig. 4, but for CC-LR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
Despite the presence of the funneling mechanism, precipitation at the vertex of the terrain in CC-LR is not enhanced with respect to precipitation on the centerline in ST-LR; rather, precipitation is enhanced on each ridge arm, flanking a local precipitation minimum at the vertex (Fig. 5b). These dual-precipitation maxima form because flow funneled toward the vertex impinges upon an area of strongly decelerated flow. As shown in previous studies (e.g., Marwitz 1980; Smolarkiewicz and Rotunno 1989; Hughes et al. 2009), decelerated flow can effectively behave like an extension of the terrain and, in the present case, the pocket of decelerated air at the vertex forces the approaching flow to separate and flow over the ridge arms. This dual-precipitation maxima distribution was not observed in Jiang’s (2006) simulations of moist flow in concave terrain; the reasons for these differences will be discussed later.
According to Galewsky (2008), if windward flow deceleration is strong enough, a secondary vertically propagating gravity wave can be launched by air flowing over the region of decelerated flow. In CC-LR, the windward extent of the decelerated flow can be identified by the upward kink in the θ contours, approximately 80 km upstream of the vertex (Fig. 5c). The positive vertical velocity coinciding with this feature is the lowest updraft of the secondary gravity wave, and it promotes the development of cloud and precipitation approximately 40 km farther upstream than in ST-LR. The location of the lowest updraft of the secondary gravity wave coincides with a perturbation u magnitude of around 6–8 m s−1, suggesting that, with the present thermodynamic profile at least, the steady-state flow must be decelerated by around 6–8 m s−1 before such a feature will develop. Notably, there is no secondary vertically propagating gravity wave in ST-LR where the largest windward perturbation u is 5 m s−1 in magnitude.
CC-LR also exhibits steepening and overturning θ contours over the lee slope, indicative of wave breaking (Fig. 5c). Although leeside wave breaking does not seem to have any direct implications for precipitation over the windward slope, it is indicative of a flow-regime transition induced by terrain shape, induced in this case by strengthened flow deceleration due to the upstream extent of the ridge arms. Wave breaking can also be responsible for the development of strong downslope winds via a hydraulic jump–like mechanism (Long 1953; see Durran 1990 for more details). Downslope winds, or foehn winds, commonly bring dry, high-θ air down to lower levels in the lee, which can inhibit condensation and prevent precipitation from being advected over the ridge crest. Figure 5c reveals an abrupt downstream edge of orographic cloud in CC-LR, aligned with the ridge crest. This is in contrast to the orographic cloud in ST-LR that spills into the lee (Fig. 4c), presumably induced by the transition to wave breaking and the subsequent development of strong downslope winds.
Overall, the average precipitation generated by CC-LR is enhanced by approximately 40% with respect to ST-LR. Two (related) flow features are responsible for this enhancement: 1) the funneling mechanism between the ridge arms that enhances precipitation near the vertex, albeit displaced from the vertex comprising two distinct maxima; and 2) the initiation of upward motion farther upstream of the ridge crest that results in precipitation at lower terrain heights. When considered together, these two processes seem to be counteractive in enhancing precipitation: the upward motion coincident with the lowest updraft of the secondary gravity wave requires the deceleration of flow to be initiated; however, this deceleration restricts the amount of flow funneled between the ridge arms.
3) Convex long ridge (CV-LR)
Figure 6 presents results from CV-LR with α = −45°. The surface streamlines (Fig. 6a) indicate more lateral deflection of the flow compared to ST-LR. This is driven by the backward reaching arms that, in contrast to CC-LR, allow more efficient diffluence of the impeded flow at the vertex. Consequently, the approaching flow follows the ridge arms more closely, and only flow that was initially within a narrow (approximately 50 km) lateral range of the vertex flows over the ridge crest. This flow pattern is consistent with the diffluence upstream of the vertex and confluence upstream of the ridge ends (Fig. 6b). This manifests as an along-barrier u gradient, with a minimum at the vertex and maxima near the ridge ends. The direct relationship among upslope wind velocity, cloud depth, and precipitation facilitates a precipitation distribution with a local minimum at the vertex and maxima toward the ridge ends. The average precipitation generated by CV-LR is almost 50% less than ST-LR, presumably because more efficient diffluence reduces the flux of mass across the ridge. The cross-section at the vertex (Fig. 6c) reveals the flow to be in a linear regime.
As in Fig. 4, but for CV-LR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
b. Shorter ridges (β = 2.0)
The previous section demonstrated that for a fixed ridge length (b = 100 km; β = 4.0), variations in terrain geometry could induce a flow regime transition, variations in upstream flow deceleration, and changes in the spatial pattern of orographic precipitation. This section presents results from simulations where the spanwise length of each ridge is halved to b = 50 km (β = 2.0) to determine how the flow response and orographic precipitation for each terrain geometry are impacted by a reduced β.
1) Straight short ridge (ST-SR)
Based on Smith (1989) and others, a reduced ridge length is expected to reduce the upstream extent of windward flow deceleration as the spanwise cross section poses less of an impediment to the incident flow. However, the flow pattern is expected to remain essentially the same as β remains above unity (Baines 1995, p. 420). The results of ST-SR are consistent with these expectations. The upstream extent of the flow deceleration is 30 km closer to the ridge crest than when β = 4.0 (Fig. 7a) and, consistent with Fig. 5 in Smith (1989), the flow regime is linear. Closer to the ridge, flow near the centerline experiences greater deflection away from the centerline, leading to an overall larger change in u. This reduced cross-barrier flow velocity hinders the strength of vertical motions and the subsequent depth of orographic cloud over the windward slope (Fig. 7c). Accordingly, precipitation along the centerline is 40% weaker with respect to ST-LR. Like ST-LR, the horizontal spatial distribution of precipitation is confined to the uppermost terrain contours (Fig. 7b).
As in Fig. 4, but for ST-SR. Note the different axes in (a) and (b) compared to Fig. 4 (chosen to account for altered ridge length).
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
2) Concave short ridge (CC-SR)
For the concave ridge, halving the ridge length reduces both the strength and upstream extent of windward flow deceleration (Fig. 8a). The minimum perturbation u of −9 m s−1 is located on the lower-windward slope of CC-SR, in contrast to the minimum perturbation of −16 m s−1 on the upper-windward slope in CC-LR. Accompanying this reduced deceleration in CC-SR is a larger flow-confluence zone (relative to β) that extends closer to the ridge crest compared to CC-LR (Fig. 8b). This implies that the funneling mechanism continues to influence flow as it ascends the ridge, consistent with decreasing perturbation u magnitude up the windward slope.
As in Fig. 7, but for CC-SR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
According to Jiang (2006), flow deceleration plays an important role in determining the strength of precipitation enhancement: the funneling mechanism is more efficient when more flow is allowed to converge between the ridge arms, generating additional updrafts that enhance precipitation. CC-SR has a total precipitation enhancement of 61% with respect to ST-SR, which is larger than the total precipitation enhancement of CC-LR with respect to ST-LR (42%); the extent of the flow confluence zone up the windward slope appears fundamental to this process. In CC-LR, the increased deceleration inhibits flow confluence up the windward slope and results in dual precipitation maxima away from the vertex. In contrast, the precipitation maximum in CC-SR forms at the vertex, suggesting a concave ridge with weaker flow deceleration has a more efficient funneling mechanism.
As with CC-LR, the −6 to −8 m s−1 perturbation u contour is coincident with the lowest updraft of a secondary gravity wave (Fig. 8c). The smaller upstream extent of flow deceleration (150 km) dictates that the secondary vertically propagating gravity wave develops farther downstream than in CC-LR, and precipitation is restricted to within 50 km of the vertex peak. Despite the weakened flow deceleration, like CC-LR, the flow regime is still nonlinear and there is wave breaking over the lee slope, a downslope wind over the vertex lee, and an associated abrupt cloud edge aligned with the ridge crest.
3) Convex short ridge (CV-SR)
While the flow dynamics in CV-SR are very similar to CV-LR, the precipitation distribution is qualitatively and quantitatively different. CV-SR has a relatively uniform precipitation distribution along the upper windward slope (Fig. 9b) and therefore lacks the precipitation maxima located at the ridge ends in CV-LR (Fig. 6b). Although average precipitation at the vertex of CV-SR is comparable to CV-LR, average precipitation is significantly less overall in CV-SR; the shortened ridge length allows a greater proportion of the deflected flow to pass around the ridge ends, whereas in CV-LR a larger proportion of the deflected flow ultimately ascends the ridge. The flow regime remains linear (Fig. 9c) and is virtually indistinguishable from CV-LR.
As in Fig. 7, but for CV-SR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
c. Sensitivity to concave angle
The previous sections showed that for a short concave ridge with β = 2.0 and α = 45°, precipitation is enhanced at the vertex of the terrain via the funneling mechanism explored by Jiang (2006). Jiang (2006) also showed that this mechanism could be modified by the concavity of the terrain, with increased α resulting in greater precipitation enhancement at the vertex. However, the previous sections also showed that the precipitation enhancement by a long concave ridge (with β = 4.0 and α = 45°) manifests as distinct precipitation maxima located on the ridge arms. This is a different result to the findings of Jiang (2006), related primarily to different upstream conditions (see section 4b). Since a long straight ridge possesses maximum precipitation at the center of the ridge, the concavity of the terrain may play an important role in modulating the location of the precipitation maxima. In this section, precipitation sensitivities to α are explored.
Figure 10 (left column) presents results from a simulation with β = 4.0 and α = 30°. Flow deceleration is weaker when α = 30° and the minimum perturbation u of −6.5 m s−1 develops on the lower windward slope (Fig. 10a), compared to a minimum perturbation u of −16 m s−1 for CC-LR where α = 45°. Along the centerline, flow confluence extends from beyond the reaches of the ridge ends to the upper windward slope (Fig. 10c). Precipitation is enhanced at the vertex resulting in a single precipitation maximum (i.e., it does not exhibit the dual precipitation maxima of CC-LR). In fact, the distribution of flow confluence and average precipitation more closely resembles CC-SR with α = 45°. In addition, a secondary vertically propagating gravity wave is generated 60 km upstream of the vertex peak (Fig. 10e), causing small rainfall accumulations farther upstream than for ST-LR.
As in Fig. 4, but for CC-LR with (a),(c),(e) α = 30° and (b),(d),(f) α = 60°.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
When α = 60° and β = 4.0 (Fig. 10, right column), the flow response is more complicated. The increased distance in the x direction between the ridge ends and the vertex allows a large pool of stagnant air to develop between the ridge arms, extending more than 120 km upstream of the vertex peak (Fig. 10b). This effectively becomes an extension of the terrain, preventing a majority of the approaching flow from entering the stagnant region. Flow confluence between the ridge arms essentially vanishes (Fig. 10d) and the surface streamlines indicate that the approaching flow separates when it impinges upon the stagnant air, flowing over the ridge arms approximately 90 km upstream of the vertex peak. This facilitates the development of distinct precipitation maxima on each ridge arm, approximately 100 km apart. The ascent over the stagnant air forms the lowest updraft of the secondary gravity wave more than 125 km upstream of the vertex peak (Fig. 10f), creating a band of precipitation that stretches between the two precipitation maxima. With respect to ST-LR, there is no enhancement of average precipitation. The increased strength and extent of upstream flow deceleration, and the associated absence of flow confluence, has made the funneling mechanism obsolete.
The cross-stream precipitation distribution for different concave angles between α = 0° and 60° (Fig. 11) helps elucidate the relationship between concave angle and the funneling mechanism. When β = 4.0, the total and maximum precipitation increases from α = 0° to 30°, and at these angles a single precipitation maximum is present at the vertex. From α = 30° to 45°, flow deceleration becomes stronger (without stagnating) and the impinging flow preferentially flows over the ridge arms creating dual-precipitation maxima. This coincides with the development of a secondary vertically propagating gravity wave that generates additional precipitation upstream of the ridge. Total precipitation increases from α = 30° to 45° despite precipitation at the vertex decreasing along with the maximum precipitation. From α = 45° to 60°, the flow stagnates between the ridge arms and the strongly decelerated flow extends far upstream. The dual-precipitation maxima migrate along the ridge arms toward the ridge ends, and total precipitation declines toward that of α = 0°. At α = 60°, the funneling mechanism at the vertex is effectively obsolete, yet at this angle the precipitation on the ridge arms is substantially larger than ST-LR. When β = 2.0, a similar precipitation response occurs, although because flow deceleration is generally weaker, a single precipitation maximum exists at the vertex for all but α = 60°.
Cross-stream precipitation distribution (mm) as α increases from 0° to 60°: (a) CC-LR with β = 4.0 and (b) CC-SR with β = 2.0. The thick black line represents the maximum cross-stream terrain height.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
These results demonstrate that flow deceleration plays a key role in determining the nature of precipitation enhancement in concave terrain, controlled in this case by modifications to the vertex angle. With relatively weak flow deceleration, the flow confluence zone extends up the windward slope, enhancing precipitation at the vertex. On the other hand, stronger deceleration leads to dual precipitation maxima forming on the ridge arms as the incident flow is deflected away from the vertex by a pool of decelerated air. This decelerated air also induces flow ascent upstream of the vertex, initiating a secondary vertically propagating gravity wave that further contributes to precipitation away from the vertex.
4. Sensitivity to upstream conditions
a. Sensitivity to changing wind direction
The previous section showed how terrain shape can have a significant impact on the distribution of orographic precipitation. An additional aspect that is worth considering is how sensitive orographic precipitation is to small variations in upstream conditions. For example, Nuss and Miller (2001) showed that changes to the terrain orientation can significantly impact the distribution and amount of precipitation. This is certainly important for forecasting because observations and forecasts of upstream conditions have inherent uncertainties. This section examines the influence of terrain shape on the sensitivity of orographic precipitation to changes in the upstream wind direction. Conceptually, there is a connection between terrain shape and the wind direction close to the ridge, with the terrain controlling the wind direction regardless of small variations in its upstream direction. For example, the ridge arms of a concave ridge funnel the approaching flow toward the vertex; thus, it might be expected that small variations in the upstream wind direction would have limited influence on precipitation near the vertex. However, this section shows that this connection is not entirely straightforward.
The six terrain shapes investigated in sections 3a and 3b are again adopted here: a straight, concave, and convex ridge with β = 4.0 and 2.0. In an attempt to systematically explore the changes in precipitation with upstream wind direction, a small ensemble of upstream wind direction changes is constructed; this ensemble can also be thought of as providing a simple representation of the uncertainty that might be expected in observations or forecasts of upstream wind conditions. The small ensemble is composed of modifications to the wind direction everywhere of up to ±10° either side of 270° (i.e., a westerly flow aligned with the x axis). The directions have an approximately normal distribution: the mean is 270° and the standard deviation is 5° (see Table 1 for more details). The upstream wind velocity remains constant at 20 m s−1; thus, a wind direction change of 10° reduces u by only 0.3 m s−1 (and by design has a negligible influence on 
1) Straight ridge
When the upstream wind direction is from 260° for a straight ridge, the upstream flow diffluence effectively straightens the approaching flow in the south while further reducing the incident angle of approaching flow in the north (e.g., Fig. 12a). An along-barrier u gradient develops, with the strongest cross-barrier wind velocity located on the southern windward slope where the approaching flow has effectively been straightened. Precipitation is advected farther downstream where u is stronger (e.g., Fig. 12b). The cross-stream precipitation distribution for ST-LR shows a progressive increase in precipitation on the southern ridge arm and decrease on the northern ridge arm as the wind direction is progressively modified from 270° to 260° (Fig. 13a), although these changes are relatively small. These findings are similar for ST-SR (Fig. 13b) despite a weaker along-barrier u velocity gradient (not shown).
Selected fields for ST-LR with upstream wind direction of 260°. Each figure shows (a) plan view of surface streamlines (solid contours) and perturbation u (dotted contours, CI = 2 m s−1); and (b) plan view of average precipitation (shaded, CI = 1 mm from 0.5 mm) and confluent flow (hatched). Terrain contours shown at an interval of hm/4.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
Cross-stream precipitation distribution (mm) for upstream wind direction changes from 270° to 260° (different symbols; see inset): (a) ST-LR with β = 4.0, (b) ST-SR with β = 2.0, (c) CC-LR with β = 4.0, (d) CC-SR with β = 2.0, (e) CV-LR with β = 4.0, and (f) CV-SR with β = 2.0. The thick black line represents the maximum cross-stream terrain height.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
A flow-regime transition is evident in ST-LR when the change in wind direction is larger than 5° (i.e., less than 265° or greater than 275°). Figure 14a indicates steepening and overturning of θ contours over the lee slope 50 km north of the centerline when the wind direction is 260°; this nonlinear feature is absent 50 km south of the centerline (Fig. 14b). The steepening of the θ contours is consistent with the reduced u impinging upon the northern ridge arm, effectively increasing 
Vertical cross section of ST-LR with upstream wind direction of 260° along (a) y = 550 km and (b) y = 450 km: potential temperature (solid contours, CI = 2 K), positive vertical velocity (shaded, CI = 0.1 m s−1, black represents > 1 m s−1), perturbation u (dotted contours, CI = 2 m s−1), and cloud water mixing ratio (dashed contours, CI = 0.2 g kg−1).
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
The precipitation difference between the northern and southern ridge in ST-LR is subtle, which makes it difficult to determine the responsible mechanisms. There is some evidence of a secondary vertically propagating gravity wave on the northern windward slope where the flow suffers greater deceleration (Fig. 14a), creating an updraft upstream of the ridge crest that may enhance precipitation. It is also possible that the stronger deceleration allows for a longer advective time scale that increases the microphysical growth of precipitation.
2) Concave ridge
For both CC-LR and CC-SR, when the wind direction is from 260°, the northern arm of the concave ridge becomes closer to normal to the incident wind direction and the southern arm becomes more closely aligned with the wind direction (e.g., Fig. 15a). Accordingly, the strength and depth of vertical motions over the windward slope increases in the north and decreases in the south, with subsequent strengthening and weakening of precipitation, respectively (e.g., Fig. 15b).
As in Fig. 12, but for CC-LR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
As the wind direction is progressively modified from 270° to 260°, the cross-stream precipitation distribution shows the magnitude of precipitation change on each ridge arm is different in CC-LR and CC-SR, with CC-LR exhibiting a stronger response (Figs. 13c,d). Additionally, there is a reduction in the average precipitation at the vertex in CC-LR, whereas it remains relatively constant in CC-SR. This is coincident with a deterioration of flow confluence between the ridge arms of CC-LR (Fig. 15b). While difficult to unambiguously reconcile the mechanisms responsible for this reduction in precipitation in CC-LR, it appears that the region of near-stagnant flow between the ridge arms contributes to this process by further deflecting the perturbed flow.
In contrast, the weaker flow deceleration in CC-SR allows the single precipitation maximum to strengthen and migrate northward away from the vertex for the same wind direction changes. This enhancement in precipitation is offset by a reduction in precipitation to the south, ensuring there is no discernible change in total precipitation (or the size of the flow confluence zone; not shown).
Although not shown here, additional simulations of the concave ridges were conducted with Coriolis effects active on flow perturbations. These simulations showed that rotation caused the incident flow to be deflected to the left (for f > 0), showing qualitative similarity to those simulations with a southerly perturbation in the upstream wind direction (without rotation).
3) Convex ridge
For CV-LR and CV-SR, when the wind direction is from 260°, the southern ridge arm becomes closer to normal to the impinging flow and the northern ridge arm becomes more closely aligned with the wind direction (e.g., Fig. 16a). Precipitation subsequently strengthens on the southern ridge arm and weakens on the northern ridge arm (e.g., Fig. 16b). While the total change of this precipitation strengthening/weakening is significantly larger for CV-LR, the relative change—approximately 50% on each ridge arm—is comparable between the two convex ridges (Figs. 13e,f). The modified wind direction does not alter the strength of upstream flow diffluence, nor does there appear to be a change in the amount of flow passing over or around CV-LR or CV-SR. Therefore, the flow response to changes in the upstream wind direction does not appear heavily dependent on the convex ridge length.
As in Fig. 12, but for CV-LR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
4) Spatial patterns of precipitation sensitivity to wind direction
In the previous subsections it was shown that small variations in upstream wind direction could lead to significant changes in the location and intensity of precipitation; however, these variations were different depending on the terrain geometry. Such sensitivities have important implications for the predictability of precipitation in complex terrain. Bearing in mind the funneling mechanism, it might seem reasonable to assume that precipitation near the vertex of a concave ridge would be relatively insensitive to upstream wind direction (and hence more predictable) as the ridge arms ultimately control the wind direction near the vertex. Conversely, the vertex of a convex ridge might suffer large precipitation sensitivity, hence leading to low predictability in this region. However, as shown earlier, the dynamics of the funneling mechanism is not necessarily straightforward because its behavior changes with ridge geometry. The implications of these changes on precipitation sensitivities are examined here.
Precipitation from each set of ensemble simulations are analyzed to determine the spatial patterns of the standard deviation σ and the standard deviation normalized by the mean σn, also known as the coefficient of variation. Note that σ can be interpreted as a measure of how sensitive precipitation is to changes in the upstream wind direction, with high σ representing high sensitivity; σn allows σ to be interpreted in the context of the sensitivity relative to the precipitation strength. Spatial maps of σ for each of the six ridge geometries are presented in Figs. 17 and 18.
Plan view of the standard deviation of precipitation (shaded, CI = 0.2 mm from 0.1 mm) from the upstream wind direction ensemble detailed in Table 1. Figures are of ridge geometries: (a) ST-LR, (b) CC-LR, and (c) CV-LR. Terrain contours shown at an interval of hm/4.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
As in Fig. 17, but for (a) ST-SR, (b) CC-SR, and (c) CV-SR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
For the straight ridges, as the wind direction changes from 270° each precipitation maximum progressively moves away from the centerline across the upper windward slope. This creates σ maxima near the ridge ends and a local σ minimum on the centerline where the precipitation change is relatively negligible (Figs. 17a, 18a). The influence of ridge length is illustrated through ST-LR exhibiting a band of higher σ spanning across the windward slope—absent in ST-SR—that arises from the along-barrier u gradient, which is more pronounced in ST-LR, advecting precipitation farther downstream where u is stronger.
The two convex ridges share similar distributions of σ, with each showing a distinct local σ minimum at the vertex and σ maxima located near the ridge ends (Figs. 17c, 18c). The ridge ends of CV-LR and CV-SR are especially sensitive to wind direction changes; the σn distribution (which is spatially similar to σ) shows large σn values, approaching unity, at the ridge ends (not shown). As mentioned above, it was expected that precipitation at the vertex of CV-LR and CV-SR would be sensitive to wind direction changes. Somewhat surprisingly, the σ and σn distributions reveal that the sensitivity of precipitation at the vertex is of comparable size to the other terrain shapes.
Although the straight and convex ridges show limited sensitivity to ridge length, the σ distributions for CC-LR and CC-SR are different from each other: CC-LR exhibits four distinct local σ maxima compared to two σ maxima for CC-SR (Figs. 17b, 18b). The strong flow deceleration in CC-LR (which is associated with the dual-precipitation maxima) seems to be the primary cause of this difference, inducing a greater difference in precipitation on each ridge arm with respect to CC-SR. Strong flow diffluence near the vertex peak also induces a greater difference in precipitation either side of the vertex on the lee slope. Furthermore, the strong flow deceleration in CC-LR shifts the location of the local σ minima near the vertex farther upstream with respect to CC-SR.
Despite the initial hypothesis that precipitation at the vertex of the concave ridge would be the least sensitive to upstream wind direction changes, all six ridges display local σ and σn minima of comparable magnitude near the vertex/on the centerline. The inference from this result is that for the prescribed conditions and upstream wind direction changes, precipitation near the vertex of all six ridges has similar predictability (neglecting other sensitivities, of course). The low σ at the vertex of the convex ridge is especially surprising. However, a possible explanation is that flow over the convex ridges is more linear with respect to the other ridge geometries, and flow deceleration therefore plays a less important role. This postulation is supported by the larger σ values for CC-LR compared to CC-SR, coincident with more nonlinear flow attributes.
Despite the similar sensitivities near the vertex, the ridge geometry is an influential factor in moderating precipitation sensitivities to changes in the wind direction. A comparison of σ between the respective long and short ridges shows that the precipitation change for longer ridges is of a greater magnitude than for shorter ridges (Figs. 17, 18). Moreover, for a given ridge length, the straight ridges show the least overall sensitivity in terms of σ and σn (inferred from Fig. 13), and the concave ridges show the greatest absolute sensitivity (in terms of σ). However, when this sensitivity is normalized by the mean precipitation (i.e., σn), the convex ridges show the greatest sensitivity, with σn approaching unity at the ridge ends. These sensitivities have obvious implications for the predictability of orographic precipitation in complex terrain.
b. Changes to the atmospheric sounding
The dual precipitation maxima in CC-LR illustrate a different response to that shown by Jiang (2006), demonstrating additional complexity to the funneling mechanism. This difference is primarily related to differences in the upstream atmospheric conditions: Jiang (2006) adopted a saturated atmosphere upstream of the topography in contrast to the initially unsaturated atmosphere adopted in the present study. It was shown in section 3c that the existence, strength, and location of the dual precipitation maxima are strongly influenced by the strength of flow deceleration between the ridge arms. This section examines how modifications to the atmospheric stability profile alter the strength of flow deceleration and, more specifically, the existence, strength, and location of the dual precipitation maxima.
Additional simulations with three different upstream thermodynamic profiles are conducted for CC-LR and ST-LR. These profiles, presented in Fig. 19, are 1) Moist Lower—increased water vapor mixing ratio below ridge height, which increases the low-level relative humidity from 75% to 82%; 2) Moist Upper—increased water vapor mixing ratio above ridge height, which increases the upper-level relative humidity and the vertical equivalent potential temperature θe gradient; and 3) Warm Upper—increased ambient potential temperature above ridge height, which increases the upper level (dry and moist) stability. As with the control sounding (Fig. 2), the modified soundings are initially unsaturated and stable to parcel ascent.
The atmospheric soundings (skew T–logp) of the three modified thermodynamic profiles: (a) Moist Lower—increased water vapor mixing ratio below ridge height; (b) Moist Upper—increased water vapor mixing ratio above ridge height; and (c) Warm Upper—increased ambient potential temperature above ridge height.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
When the atmosphere is almost saturated below ridge height in Moist Lower (Fig. 19a), 
As in Fig. 4, but for the modified upstream thermodynamic profile Moist Lower with CC-LR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
The reduced flow deceleration and enhanced zone of flow confluence in Moist Lower (CC-LR) suggest that the increased low-level moisture facilitates a more efficient funneling mechanism and increases the precipitation enhancement accordingly; however, this is only partially true. To quantify the precipitation enhancement, the precipitation distributions of CC-LR and ST-LR (with the same upstream thermodynamic profiles) are compared. It is found that enhancement of the precipitation maxima (from ST-LR to CC-LR) increases from 16% with the control sounding to 22% with Moist Lower; whereas enhancement of total precipitation (from ST-LR to CC-LR) decreases from 40% with the control sounding to 22% with Moist Lower. The enhanced precipitation maxima can be attributed to stronger low-level flow confluence and stronger upslope flow contributing to stronger updrafts and precipitation near the vertex peak; the total precipitation decreases (relatively) because flow deceleration is not strong enough to initiate a precipitation-inducing secondary vertically propagating gravity wave upstream (Fig. 20c).
To better consider the role of moisture in Moist Lower, the drying ratio (DR) is calculated (as well as for the other simulations with an adjusted thermodynamic profile; see Table 2). The DR is defined as DR = P/WV, where P is the mass of accumulated precipitation over a given area and WV is the water vapor influx far upstream of the ridge (Smith et al. 2003); WV is calculated using 
Summary of computations for four experiments with different upstream thermodynamic soundings: Control, Moist Lower, Moist Upper, and Warm Upper (see section 4b). The low-level relative humidity (RH, average from surface to ridge height); the total hourly precipitation mass P (for ST-LR, CC-LR); the accumulated hourly water vapor (WV) influx; and the drying ratio (DR, for ST-LR, CC-LR).
The DR in Moist Lower increases for both ST-LR and CC-LR with respect to the DR for the control sounding, reflecting precipitation increases that are larger than the corresponding WV influx increase. Previous studies have shown that the precipitation efficiency of an orographically modified flow can greatly improve when the low-level humidity rises (e.g., Browning et al. 1975; Jiang and Smith 2003), suggesting that an increased DR for both ridges with Moist Lower is within expectations. However, the DR enhancement by CC-LR compared to ST-LR with Moist Lower is smaller than the enhancement with the control sounding because the increased low-level humidity reduces the stability, which strengthens the cross-barrier flow velocity and reduces the influence of windward flow deceleration.
When moisture is increased above ridge height in Moist Upper, 
As in Fig. 4, but for the modified upstream thermodynamic profile Moist Upper with CC-LR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
Computation of the DR for ST-LR and CC-LR with Moist Upper reveals contrasting responses to the increased moisture content above ridge height: for the straight ridge, the DR changes very little with respect to the control sounding whereas the DR increases substantially for the concave ridge. This contrast arises because CC-LR sources comparatively more precipitation from Moist Upper via deeper and more widespread updrafts above the windward slope, driven by the funneling mechanism. As the additional influx of WV provided by Moist Upper is restricted to flow above ridge height, precipitation can only be sourced from the moisture-enriched flow if the updrafts generated are sufficiently deep. While the flow response of ST-LR is sufficient to increase precipitation by an amount comparable to the increase in WV influx, the flow response of CC-LR is far more effective at sourcing precipitation from higher altitude. This highlights an important precipitation sensitivity to the vertical distribution of moisture through the atmosphere and the terrain geometry.
The third modified sounding is Warm Upper, in which none of 
As in Fig. 4, but for the modified upstream thermodynamic profile Warm Upper with CC-LR.
Citation: Journal of the Atmospheric Sciences 69, 4; 10.1175/JAS-D-11-0198.1
In addition to being sensitive to terrain geometry and upstream wind direction, orographic precipitation is highly sensitive to the upstream thermodynamic profile. This section has shown that relatively small changes to the water vapor mixing ratio and the ambient potential temperature can substantially modify the distribution of precipitation, from having maxima on the ridge arms to having a maximum at the vertex. Furthermore, precipitation is very sensitive to the vertical structure of the upstream thermodynamic profile. Of course, there are infinite sensitivity tests that could be conducted and only a select few have been shown here.
5. Summary and conclusions
This study has examined how variations in relatively simple terrain geometries influence orographic precipitation and its spatial patterns of sensitivity to changes in upstream conditions. An idealized model was used to simulate a moist, precipitating flow impinging upon three alpine-scale terrain shapes: a straight ridge, a concave ridge, and a convex ridge. Two different ridge lengths were used and the environment was based on an observed atmospheric sounding, representative of a northwestern Australian cloud band. Inspired by the flow regime diagram of Fig. 5 in Smith (1989), the scale of the terrain was chosen to ensure that the orographically modified flow was near bifurcation between linear and nonlinear flow regimes; the nondimensional mountain height 
It was found for the straight and convex ridges that the flow response was generally linear and passed over the obstacles with little lateral deflection. The concave ridge, however, exhibited strengthened flow deceleration (especially when the horizontal aspect ratio of the terrain β was larger) and wave breaking in the lee. The concave ridge also generated substantially more precipitation than the other two ridge geometries, attributed to the forward-reaching ridge arms that funnel approaching flow toward the vertex and limit upstream flow diffluence. This process, termed the “funneling mechanism,” created additional low-level flow confluence between the ridge arms and enhanced both vertical motions and precipitation over the windward slope.
Jiang (2006) found that the efficiency of the funneling mechanism is sensitive to 
The vertex angle of the concave ridge was found to moderate the strength of windward flow deceleration. This provided an opportunity to discern a relationship between precipitation enhancement and flow deceleration. Simulations were carried out with ridges of varying concavity, from α = 0° (i.e., a straight ridge) to α = 60°, and it was found that total precipitation continued to increase with increasing concavity despite windward flow deceleration strengthening. However, when the flow stagnated (at α = 60°, β = 4.0), total precipitation reduced to be similar to that of the straight ridge. Additionally, from α = 30° to 45° the single-precipitation maximum split into dual-precipitation maxima. These results suggest that a balance exists between the contribution to precipitation enhancement from lateral flow convergence and from the development of an upstream source of vertical motion. Jiang (2006) defined the funneling mechanism as the contribution of precipitation from low-level lateral flow convergence; however, it has been shown that precipitation in concave terrain can also be enhanced from strengthened flow deceleration. It seems that when 
It was originally hypothesized that precipitation in concave terrain would be less sensitive to variations in the upstream wind direction because approaching flow would be funneled toward the vertex. A small ensemble was used to explore the sensitivity of precipitation to small variations in upstream wind direction. Somewhat unexpectedly, all three ridge geometries showed local minima in precipitation sensitivity near the vertex of comparable magnitude. For a given ridge length, straight ridges showed the least sensitivity to variations in upstream wind direction as both the concave and convex ridges showed areas of substantial sensitivity. In some cases, the standard deviation of precipitation among the ensemble members was as large as the mean, a result that implies low predictability of orographic precipitation in this regime.
The dual precipitation maxima were absent from Jiang’s (2006) study most likely because of the chosen atmospheric thermodynamic profile. Additional simulations were carried out with three modified soundings to test the role of moisture and stability in moderating the dual precipitation maxima, and the flow response for a concave ridge. It was found that when water vapor was increased below ridge height, flow deceleration weakened and the dual precipitation maxima were close to merging to a single precipitation maximum. Presumably, further increases of low-level moisture would create a precipitation distribution that more closely resembled Jiang’s (2006) result. When water vapor was increased above ridge height, flow deceleration remained as strong as in the control case, although the increased upper-level relative humidity promoted cloud formation near the center of the ridge and created a single precipitation maximum. When the ambient temperature was increased above ridge height, vertical motions were suppressed and the impinging flow preferentially flowed around the region of decelerated air rather than over. Although flow confluence was absent between the ridge arms, precipitation was enhanced by 77% with respect to the straight ridge when the ambient temperature was increased above ridge height, implying that a concave ridge can still significantly enhance precipitation in the presence of strong and widespread flow deceleration.
This study has demonstrated that changes in terrain geometry have important influences on the orographic flow response and the resultant distribution of precipitation. These changes not only affect the pattern and amount of precipitation but also influence its sensitivity to small variations in upstream conditions. Among other things, these results have important implications for predictability of precipitation in complex terrain. This is an especially important topic because modern numerical weather predication models are now capable of resolving realistic terrain features with complicated geometries. However, we are mindful that the parameter space is infinite and continued research in this area is necessary.
Acknowledgments
C. Watson acknowledges support from a Melbourne Research Scholarship, and T. Lane was funded by an Australian Research Council Future Fellowship (FT0990892). Computing facilities were provided by the Victorian Partnership for Advanced Computing. We would also like to thank Simon Caine for assistance with the numerical model, Rich Rotunno for useful discussion in the early stages of the study, and three anonymous reviewers for their comments on an earlier version of the paper.
REFERENCES
Baines, P. G., 1987: Upstream blocking and airflow over mountains. Annu. Rev. Fluid Mech., 19, 75–95.
Baines, P. G., 1995: Topographic Effects in Stratified Flow. Cambridge University Press, 482 pp.
Browning, K. A., C. Pardoe, and F. F. Hill, 1975: The nature of orographic rain at wintertime cold fronts. Quart. J. Roy. Meteor. Soc., 101, 333–352.
Colle, B. A., 2004: Sensitivity of orographic precipitation to changing ambient conditions and terrain geometries: An idealized modeling perspective. J. Atmos. Sci., 61, 588–606.
Durran, D. R., 1990: Mountain waves and downslope winds. Atmospheric Processes over Complex Terrain, Meteor. Monogr., No. 45, Amer. Meteor. Soc., 59–81.
Durran, D. R., and J. B. Klemp, 1982: On the effects of moisture on the Brunt–Väisälä frequency. J. Atmos. Sci., 39, 2152–2158.
Galewsky, J., 2008: Orographic clouds in terrain-blocked flows: An idealized modeling study. J. Atmos. Sci., 65, 3460–3478.
Hughes, M., A. Hall, and R. G. Fovell, 2009: Blocking in areas of complex topography, and its influence on rainfall distribution. J. Atmos. Sci., 66, 508–518.
Jiang, Q., 2003: Moist dynamics and orographic precipitation. Tellus, 55A, 301–316.
Jiang, Q., 2006: Precipitation over concave terrain. J. Atmos. Sci., 63, 2269–2288.
Jiang, Q., and R. B. Smith, 2003: Cloud timescales and orographic precipitation. J. Atmos. Sci., 60, 1534–1559.
Lalas, D. P., and F. Einaudi, 1974: On the correct use of the wet adiabatic lapse rate in stability criteria of a saturated atmosphere. J. Appl. Meteor., 13, 318–324.
Landvogt, P. K., J. A. T. Bye, and T. P. Lane, 2008: An investigation of recent orographic precipitation events in northeast Victoria. Aust. Meteor. Mag., 57, 235–247.
Long, R. R., 1953: Some aspects of the flow of stratified fluids. I. A theoretical investigation. Tellus, 5, 42–58.
Marwitz, J. D., 1980: Winter storms over the San Juan Mountains. Part I: Dynamical processes. J. Appl. Meteor., 19, 913–926.
Miglietta, M., and A. Buzzi, 2001: A numerical study of moist stratified flows over isolated topography. Tellus, 53A, 481–499.
Miglietta, M., and R. Rotunno, 2005: Simulations of moist nearly neutral flow over a ridge. J. Atmos. Sci., 62, 1410–1427.
Nuss, W. A., and D. K. Miller, 2001: Mesoscale predictability under various synoptic regimes. Nonlinear Processes Geophys., 8, 429–438.
Pierrehumbert, R. T., and B. Wyman, 1985: Upstream effects of mesoscale mountains. J. Atmos. Sci., 42, 977–1003.
Reeves, H. D., and R. Rotunno, 2008: Orographic flow response to variations in upstream humidity. J. Atmos. Sci., 65, 3557–3570.
Reinecke, P. A., and D. R. Durran, 2008: Estimating topographic blocking using a Froude number when the static stability is nonuniform. J. Atmos. Sci., 65, 1035–1048.
Roe, G. H., 2005: Orographic precipitation. Annu. Rev. Earth Planet. Sci., 33, 645–671.
Rotunno, R., and R. Ferretti, 2001: Mechanisms of intense alpine rainfall. J. Atmos. Sci., 58, 1732–1749.
Schneidereit, M., and C. Schär, 2000: Idealised numerical experiments of alpine flow regimes and southside precipitation events. Meteor. Atmos. Phys., 72, 233–250.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp.
Smith, R. B., 1979: The influence of mountains on the atmosphere. Advances in Geophysics, Vol. 21, Academic Press, 87–230.
Smith, R. B., 1989: Hydrostatic airflow over mountains. Advances in Geophysics, Vol. 31, Academic Press, 1–41.
Smith, R. B., 2006: Progress on the theory of orographic precipitation. Special Paper 398: Tectonics, Climate, and Landscape Evolution, S. D. Willet et al., Eds., Geological Society of America, 433–454.
Smith, R. B., Q. Jiang, M. G. Fearon, P. Tabary, M. Dorninger, J. D. Doyle, and R. Benoit, 2003: Orographic precipitation and air mass transformation: An Alpine example. Quart. J. Roy. Meteor. Soc., 129, 433–454.
Smolarkiewicz, P. K., and R. Rotunno, 1989: Low Froude number flow past three-dimensional obstacle. Part I: Baroclinically generated lee vortices. J. Atmos. Sci., 46, 1154–1164.
Sturman, A. P., and N. J. Tapper, 1996: The Weather and Climate of Australia and New Zealand. Oxford University Press, 476 pp.
Tapp, R. G., and S. L. Barrel, 1984: The northwest Australian cloud band: Climatology, characteristics and factors associated with development. J. Climate, 4, 411–424.
Thompson, G., R. M. Rasmussen, and K. Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132, 519–542.
Wright, W. J., 1989: A synoptic climatological classification of winter precipitation in Victoria. Aust. Meteor. Mag., 37, 217–229.
Zängl, G., 2008: The temperature dependence of small-scale orographic precipitation enhancement. Quart. J. Roy. Meteor. Soc., 134, 1167–1181.
This microphysics parameterization scheme includes the generation of ice as well as warm rain processes. However, the vast majority of precipitation is in liquid form and although there are some cases of snow and ice forming downstream of the ridge, it is restricted above 4 km and does not result in any precipitation accumulation. Furthermore, the formation of ice does not have a noticeable impact on the flow response.
