Using data from the Hawaiian Rainband Project (HaRP), Richard Carbone, in collaboration with John Tuttle, William Cooper (joint appointment with ASP), Wen-Chau Lee (ATD), and Vanda Grubisic (ASP), completed a study of Hawaiian rainfall that supports both sides of the argument about the relative roles of dynamic (flow blocking) and land-seabreeze effects on Hawaiian rainfall. They found that total rainfall in the windward Hawaiian domain is principally controlled by the island Froude (FR) number U/Nh, where h is the characteristic height of the island, N is the Brunt-Vaisala frequency, and U is the windspeed. However, the distribution of rainfall within this domain is principally controlled by thermal forcing and mostly associated with radiative flux divergence and evaporative cooling. The island seems to have no significant effect on rainfall production beyond 50 km upstream from the windward shore. This is about a factor of two smaller than anticipated.
There are several secondary influences. Large U alone is associated with greater-than-normal rainfall everywhere, including the oceanic background precipitation. Increased static stability is associated with reduced rainfall over the ocean but rainfall fully consistent with Fr over the the windward island. Anomalously deep tradewind layers can produce heavy rain almost anywhere, with large increases typically experienced offshore.
These findings, while specific to Hawaii, are generally applicable to vast regions of the tropics and subtropics where there are mountainous islands, and the Froude number concepts are applicable to orographic precipitation with small conditional instability worldwide.
Carbone, Grubisic, and Piotr Smolarkiewicz used a new Smolarkiewicz model to examine the effects of various thermal forcings on rainband initiation and location. The initial simulations were at low (5 km) resolution and seek to reproduce observed mesoscale flow and better document the effects of radiation and evaporation through improved physical representations. Early findings suggest that the diurnal radiative effects exert a high level of control over rainband formation and location, and that evaporative effects are often important as well. These findings differ significantly from Smolarkiewicz and colleagues' earlier simulations, where physical influences were less apparent due to properties of the model used and the way experiments were conducted. Higher resolution simulations will be conducted in 1997 to more fully represent convective and evaporative processes on the cloud scale. It is expected that Carbone and colleagues' (1995) findings about the evaporative processes, will be confirmed as a key evening transition forcing of, and flow reversal in, nocturnal "landband" formation.
Grubisic and Smolarkiewicz accomplished a study of the gravity-wave/critical-level interaction in stratified 3D orographic flows. The effect of a (single) critical level on airflow past an isolated axially-symmetric obstacle was investigated in the small-amplitude, inviscid, hydrostatic limit for mean flows with linear negative shear at Richardson numbers (Ri) greater than 1/4. The problem was examined using linear theory (with the characteristic singularity of solutions at critical levels) and the semi-Lagrangian/Eulerian numerical model. Below the critical level, the linear theory predicts the 3D wave pattern with wave envelopes that widen quickly with height as the critical level is approached. Above the critical level, the waves are strongly attenuated, with the attenuation factor that depends on the horizontal wavenumbers. Drag imposed on the mountain by this wave field decreases with Ri but does not lead to a zero-drag limit as Ri --> 1/4. The numerical study showed that the well-defined linear regime (where steady-state analytic and numerical results agree uniformly everywhere except in the vicinity of the critical level) exists but was confined to increasingly smaller (infinitesimal) mountain heights as Ri --> 1/4. Even well within the linear regime, flow in the vicinity of the critical level was dissipative in nature, as evidenced by the development of a potential vorticity doublet below the critical level.
Over the last year, Terry Clark, Janice Coen, William Hall, Hsiao-Ming Hsu (visitor, Woods Hole Oceanography Institute), and Teddie Keller continued their studies of flows over Lantau Island as part of furthering the understanding of terrain-induced turbulence at the new Hong Kong airport. The main effort during FY96 was on finalizing analyses and writing results as journal articles.
The 7 June 1994 Tropical Storm Russ case study of mechanical turbulence over Lantau Island was analyzed and compared with observations. Overall, the high horizontal resolution simulation compared well with both the aircraft and automated weather station data. Preliminary analysis of the model results indicated that the high horizontal resolution was required to capture the topographically distorted flow to a resolution where mean shears became dynamically unstable and broke down into transient eddies. It was shown that the horizontal and vertical scales of the propagating eddies (or shedding vortices) were in good agreement with predictions from the linear equations or dispersion relations. The transient nature of the eddies resulted in localized drag similar to that produced by an obstacle in neutral flow. This interaction then resulted in eddies expanding their vertical scale to a height equivalent to the depth predicted for an evanescent mode. The 2-3 km-wide, energy-containing eddies extended to about 600 m AGL (above ground level) in the atmosphere, producing the strong shears and severe turbulence observed on that day. Using Bernoulli's energy form, the three-dimensional spatial structure and transient nature of the regions of flow separation (or separation bubbles) were successfully displayed. We believe this is the first time this simple and highly effective analysis technique has been used. A vertical cross section from the Tropical Storm Russ simulations shows the Bernoulli energy form, including contours of the values of 60 and 80 m 2/s2. These contours at times hug the surface and at other times separate, forming separation bubbles. A horizontal plot of the height of the separation bubbles demonstrates their three-dimensional structure of transient eddies. A paper was submitted to the Journal of the Atmospheric Sciences describing these results.
Using the large-scale initialization of the Clark-Hall model, Clark, Hall, and Robert Kerr continued their study of the 9 December 1992 incident over Evergreen where a DC-8 lost an engine in clear-air turbulence. This work was in collaboration with researchers at NOAA/ETL/FSL. High-resolution simulations using 300 m in the horizontal and 200 m in the vertical for the highest resolution domain were completed this past year. Analysis of the results shows extremely strong shears vorticity and turbulence occurring in the region of the aircraft incident. Analysis of the turbulence fields showed that the Probability Distribution Functions (PDFs) of the perturbation components of the flow form exponential distributions which are in agreement with DNS simulations of well-developed turbulence. When the PDF analysis was rotated and aligned with the mean flow direction, the variances indicate rotors aligned with the direction of the flow, i.e., the variances of the velocity components in the cross flow and vertical direction were about three to four times larger than those in the mean flow direction. Simulations and analysis of this case will continue into FY97. Kerr's experience in DNS (direct number simulation) turbulence, help in the assessment of the nature of this type of turbulence. The results of this high resolution simulation were presented at the Rotating and Stratified Turbulence workshop in Boulder during the summer of 1996.
Roelof Bruintjes (joint appointment with RAP), Hall, Coen, and Clark were actively involved in the Arizona Program during the past year. This work was in collaboration with researchers from NOAA/ETL, University of Arizona, Army Corps. of Engineers, Arizona Department of Water Resources (ADWR), University of Wisconsin, NASA, and NSSL. Two main objectives of this program were to improve the understanding of the processes that determine the spatial and temporal distribution of precipitation, and to assess the potential for artificially enhancing winter precipitation in a mountainous region of central Arizona. As typical approaching winter storms generate flow orthogonal to a series of three mountain ranges, gravity waves and wave clouds containing supercooled liquid water are often produced. Of special interest to the Arizona Program was the interaction of these topographically-induced gravity waves with the ambient upslope flow. It has been suggested that these waves may serve to augment the upslope-forced precipitation that falls onto the Mogollon Rim.
A major thrust of these studies were to compare the observations collected during gravity wave/winter storm events that occurred during the Arizona Field Program 1995 (AP95) with simulations that used the Clark-Hall numerical model. This data includes detailed measurements of the precipitation evolution in both natural and seeded winter storms as they pass through the area. The simulations have exploited model features such as large-scale initialization using RUC (Rapid Update Cycle) data, coarse grain parallelization, and interactive grid nesting with grid refinement in both the horizontal and vertical directions to telescope from the large-scale atmospheric flow to the precipitation formation in individual clouds over the Verde Valley and Mogollon Rim. Two events that caused heavy local precipitation and flash flooding are being examined.
Hall led the study of the 14 February 1995 event, which was characterized by gravity waves that fluctuated in strength and at times ceased entirely. Surface precipitation rates downwind of the gravity wave in the Verde Valley and on the upwind side of the Mogollon Rim were closely correlated with the strength of the gravity wave, in that mid-valley precipition rates peaked immediately after the weakening or collapse of the gravity wave; this appears to be due to supercooled liquid water loading in the wave updraft. When the suspended liquid water increases, the ice and snow mixing ratios also increase immediately downwind and subsequently descend into the low-level upslope cloud over the Mogollon Rim, providing enhanced precipitation on the Rim and decreasing precipitation rates in the valley. Radars, aircraft, and surface stations provided data that agree with these fluctuations in air motion and precipitation.
In a study of 6 March 1995, Coen examined the sequence of microphysical processes that occurred before and during a cold frontal passage through the AP95 observing network. This event was one of several featured at the 4th International Cloud Modeling Workshop in August 1996. For this 6 March event, the prefrontal period was characterized by strong orographic gravity waves downwind of the Black Hills. Some of these waves, such as one associated with Mingus Mountain, persisted for many hours and produced wave clouds containing large amounts of supercooled liquid water, which was observed by radar and radiometer. Simulations show that ice crystals that formed near the supercooled liquid water suspended in the gravity wave fell across the valley into the liquid upslope cloud, triggering graupel formation. With the approach of the cold front, the large-scale flow speed and direction varied, and led to fluctuations in gravity wave amplitude and wave cloud height and location, affecting precipitation rates and distribution downwind in the Valley and on the Rim. During frontal passage, the effects of which are modified strongly by the rugged and complex terrain, upslope clouds produced heavy precipitation (5-7 cm measured by surface stations) on the snowpack on the Mogollon Rim.
In both these studies, the simulations agree quite well with the rich set of data, both in the location and amplitude of the gravity wave, supercooled liquid water contents, and precipitation rates on the Rim. They also demonstrate the range of factors that affect the distribution of precipitation, from changing large-scale conditions to the complicated effect of small-scale topographic features.
Christopher Davis (joint appointment with RAP) conducted an analysis of diurnally forced, anticyclonic circulations over northeast Colorado with the aid of the Winter Icing and Storms Project 1994 (WISP94) data set. Nineteen cases of the so-called Longmont anticyclone were found, sixteen of which formed in the afternoon and dissipated after sunset. Two different types of circulations were identified, based on upstream flow direction. The air within Type 1 circulations originates to the west of the Continental Divide and contains little moisture. Type 2 circulations form in more moist, northerly flow and are sometimes associated with snowbands. Both types tend to form during the afternoon and dissipate after sunset, although Type 2 events may follow frontal passages and occur at night. Case studies of one event of each type suggest that anticyclonic vorticity generation occurs within the lowest kilometer above ground level (AGL) when that layer is nearly vertically mixed in both potential temperature and velocity. An analogy with vorticity generation in mixed layer models was considered, and it was shown that the conditions for generating negative vorticity in those models were satisfied in each observed case. The mixed-layer mechanism was also favored as it naturally explains the diurnal tendency of the circulations, and may therefore explain the observed, late-day snowfall maximum along the Front Range of Colorado.
Steep orography along coastal zones can lead to a variety of cool-season weather events such as high winds associated with barrier jets and heavy orographically enhanced rainfall. Scott Braun (visitor, University of Washington), Richard Rotunno, and Joesph Klemp completed a numerical study using a two-dimensional, Boussinesq terrain-following coordinate model in which an initially steady, geostrophically balanced front upstream of the coast moves toward the coast at a prescribed velocity. They determined that the motion of fronts can be significantly retarded along the coast by the well-known flow deceleration that occurs on the windward slope of an obstacle in a rotating stratified fluid. They found that across-frontal circulations induced by frontogenesis/frontolysis caused by the front flowing over the mountain are small compared to the changes in the mountain circulation caused by the stability perturbations associated with the front. They also established that the strength of the along-shore winds during frontal passage at the coast is approximately determined by a superposition of the southerly barrier jet and the frontal jets, implying that a barrier jet forming ahead of a front can combine with a prefrontal jet to produce very strong winds in the coastal zone prior to frontal passage.
Klemp, Skamarock, and Rotunno found in the past that the two-layer shallow water equations are useful for studying disturbances propagating along the inversion capping the marine boundary layer, and trapped against the coastal mountains of the west coast of the U.S. However, these equations lack the physics necessary to investigate the relation between the evolution of synoptic-scale flow and its interactions with the coastal mountain barrier that ultimately trigger the marine-layer disturbance. For this reason, they recently conducted numerical simulations with a continuously stratified model for flow above a uniform coastal barrier to investigate the initiation mechanisms for the coastally trapped disturbances. One proposed initiation mechanism is anomalous offshore (downslope) flow that leads to lee troughing. To simplify the problem, they initially used a two-dimensional (no along-coast variation) version of the model. They initialized the model with a thermally balanced northerly flow over the ocean side of a plateau representing the coastal mountains. To simulate the synoptic evolution, an easterly wind is gradually imposed. They found that the evolving flow over the plateau creates several features favorable to the development of southerly coastal surges. The initial northerly flow is advected westward, away from the coast, thereby removing the opposing flow that would impede a southerly surge. The descending flow on the lee side of the plateau creates warming that significantly lowers the low-level pressure. The Coriolis effect is an essential element of this lee troughing, allowing subsiding flow to remain in balance. Without rotational effects, these values of offshore wind, terrain height, and ambient static stability produce a low Froude number flow that would separate from the slope, and lee troughing would not occur.