Abstract

This study presents a comparison of an observed mesoscale sea breeze circulation and numerical experiments with a mesoscale model simulation of the same event. The observed dataset was from the Jacksonville Area Sea Breeze Experiment (JASBEX), conducted as a cooperative data gathering effort by many agencies and individuals interested in the coastal sea breeze of the Atlantic Coast. The data was collected from 23 July to 25 August 1995 covering the east coast areas of Northeast Florida and Southeast Georgia. The mesoscale model (Penn State/University Corporation for Atmospheric Research MM5) incorporates nonhydrostatic dynamics, diffusion, a multi-level planetary boundary layer (PBL), surface friction, and the split, semi-implicit time-integration scheme. In its nonhydrostatic configuration, separate predictive equations are required for the pressure and vertical velocity terms. Rather than the full pressure term, the local perturbation pressure is the predictive variable. A two-way interactive triple-nested domain configuration with grid spacings of 36, 12, and 4 km is used for the simulation presented in this study. The important physical processes evident during selected days of the dataset (including both sea breeze and river breeze) are discussed and compared with mesoscale modeling studies initialized on synoptic data. Critical analysis of synoptically driven model results and the mesoscale data are generated with a view toward future quasi-operational implementation of numerical mesoscale models of the sea breeze supported by a mesoscale data network.

With the rapid advances in computational power available in the near term, and the criticality of the sea breeze convergence to the initial convection problem, we may see the use of such quasi operational models as convective forecast aides along the Southeast and Gulf Coast sooner than we might have imagined only five years ago.

1. Introduction

In July and August of 1995 the National Weather Service office in Jacksonville, Florida (NWS JAX) led a group of cooperating organizations from several government agencies, academia and even some individual volunteers, in conducting the Jacksonville Area Sea Breeze Experiment (JASBEX). This experiment was a joint effort between NWS JAX, the local Naval Atlantic Oceanography and Meteorological Command active and reserve units, the Florida Air National Guard, local universities including Jacksonville University and Florida State University, and data collection efforts by several additional state and federal agencies along with some local area cooperative volunteers. The data network was designed to be sufficiently dense to observe the atmospheric mesoscale scale features, including the development of secondary circulations such as a sea breeze or river breeze. The experiment consisted of surface and upper air data collection with sites located in Northeast Florida and Southeast Georgia during the weeks of July 24-28, and August 14-18, of 1995.

The main scientific objectives of this experiment were to study the local sea breeze and convective interaction, and to verify the rainfall estimates recorded by the new WSR-88D NEXRAD Doppler Radar (88D). This paper is concerned with the first objective, and presents initial results and computer simulations of the land sea breeze system (LSBS) for July 26 and 27 of 1995. This paper presents and compares mesoscale numerical model simulations with collected data from the 88D Doppler Radar and JASBEX surface data for these dates. This study, which is the first to specifically focus on the LSBS of the northeastern Florida and southeastern Georgia coastlines (and possibly the first to compare a high-resolution numerical model simulation, 88D radar imagery and a mesoscale surface sea breeze network), we used a triple domain (double-nested) primitive-equation mesoscale numerical model. The mesoscale model was set up with a point resolution of 36 km over an outer domain of 860 km by 860 km, with an innermost grid resolution of 4 km spanning an area of 280 km by 270 km. The innermost grid is essentially the JASBEX domain.

The long-range goals of these modeling studies are to understand the complex mechanisms which initiate and sustain sea breeze convection, and to apply this understanding to more accurately forecast the location and intensity of such convective initiation. Further, we hope to contribute to efforts involved in validating the need for research and development of coastal sea breeze mesoscale numerical modeling studies in the Southeast, where such circulations are significant to the development of severe weather events, which occur almost daily in the warm season.
Due to such frequent occurrence, Florida is historically the preferred site for sea breeze and convective studies. Among these, we can mention the Thunderstorm Project in 1946 (Byers and Rodebush, 1948), the Florida Area Cumulus Experiment (FACE) in 1971, 73, and 75 (Ulanski and Garstang, 1978), the Convective and Precipitation/Electrification Experiment (CAPE) in 1991 (Rubes et al., 1993), and the Tallahassee Area Sea Breeze Experiment (TASBEX) in 1994 (Herbster, 1996). However, the specific complexity of the sea breeze circulation of the North Florida-South Georgia coast has been only recently studied (Tunney, 1996) though it has long been a subject of discussion among local operational forecasters.

The simplest sea breeze conceptual model is one of differential diurnal heating rates; contrasting the coastal landmass with that of the adjacent ocean. This heating contrast gives rise to sloping pressure surfaces and both coastal thermal and pressure gradients which favor inland surface flow and rising vertical motion in the heated air over the adjacent landmass. This in turn, initiates the sea breeze circulation. The sea breeze secondary circulation develops as cooler air over the ocean is accelerated inland in response to the pressure gradient, and return offshore flow aloft develops subsidence over the ocean completing the secondary circulation.
Recent research has shown that the coastal sea breeze is a much more complex circulation than this simple conceptual model, a situation where dynamic, surface, and radiative processes are all working simultaneously and that clouds can play a significant role. Each process is in turn strongly influenced by local variations in terrain, vegetation, moisture distribution, and coastal shape (Xian and Pielke, 1991; Rubes et al., 1993; Herbster, 1996, Wai et al., 1996). Therefore only sophisticated numerical model studies, which have sufficient atmospheric physics, can realistically reflect how all these different processes work together to generate the sea breeze circulation.

The LSBS in Northeast Florida and Southeast Georgia is particularly complex, and the importance of the local terrain, surface fluxes, and flow interactions between the sea breeze circulation and the broad St Johns River, the Okefenokee swamp, and coastal estuarian marshes, are particularly acute in this area. JASBEX was conducted to gather a mesoscale data set of the sea breeze which would support high resolution numerical model simulations of these complex interactions given adequately resolved LSBS flow. Even sophisticated models fall short of fully simulating such complex interactions, and yet, if carefully crafted, they can lead us to a physically correct interpretation of the LSBS forced convective initiation process.

The mesoscale numerical model used for the simulations presented here is the fifth generation, Pennsylvania State University and University Corporation for Atmospheric Research (PSU/UCAR) Mesoscale Modeling system (MM5) described by Grell et al., 1994. Specific sea breeze modifications and adaptations for Florida simulations were developed at Florida State University by Herbster (1996). An abbreviated model description is found in section two below. In section three, we describe the dataset and how it was collected. Next, we compare the data with the model simulation. Finally, we summarize with a list of proposed studies and research projects for the future.

2. Model Description

For the simulation presented here, Version 1 of the MM5 was used. The MM5 incorporates nonhydrostatic dynamics, diffusion, a multi-level planetary boundary layer (PBL), surface friction, and the split, semi-implicit time-integration scheme. In its nonhydrostatic configuration, separate predictive equations are required for the pressure and vertical velocity terms. Rather than the full pressure term, the local perturbation pressure is the predictive variable. Therefore, the model variables in the MM5 are pressure perturbation (p’), the three momentum components (u, v, w) , temperature (T), and specific humidity (q). The model uses a terrain-following sigma coordinate system defined entirely from a reference state [po(z), To(z), (z)], where sigma and reference pressure are defined by

sigma = (po - pt/p*), p* = ps - pt

and ps and pt (pt = 100mb) are the surface and top pressures of the model, respectively. The MM5 includes a flux form for advection, and its variables are coupled with p*. The MM5 uses the Arakawa B-grid staggering of the horizontal velocity variables (u, v) with respect to the other fields (T, q, p’). While vertical velocity is defined on the full surfaces, all other variables are defined halfway between these levels. A relaxation lateral boundary condition and a radiation upper boundary condition are employed. For detailed descriptions of the MM5 model in general, refer to Grell et al., 1994, and Dudhia (1993).

A two-way, interactive, double-nested domain configuration was used for the simulation presented in this study, with all nest locations held fixed throughout the simulation. For this study, the model is integrated with horizontal resolutions of 36 km, 12 km, and 4 km for the three domains, as is shown in Fig. 1. However, for the comparison study in this paper, only simulations from the innermost grid (4 km) are shown. Over land, the surface temperature is calculated from a surface energy budget based on the "force-restore'' method originally developed by Blackadar (1976, 1979) and further developed by Zhang and Anthes (1982). The performance of this method has been found to be quite adequate for simulating the diurnal variation of the ground temperature and heat flux (Deardorff 1978). A more recent version of the MM5 (Version 2) does include a multi-level soil temperature algorithm, but that version was not available at the time this simulation was conducted.

The Planetary Boundary Layer (PBL) is handled with a multiple-level first order closure scheme, originally developed by Blackadar (1976, 1979), and using standard similarity theory for the calculation of surface heat and moisture fluxes. As a first order closure scheme, this method provides an adequate estimate of the diurnal variation of the PBL for this study. An added benefit to using this scheme is that the computational demands are much lower than would be required for a higher order scheme. A second order PBL closure scheme for the MM5 is being developed, but was not available at the time we concluded this study.

The atmospheric radiation parametrization consists of separate long wave and shortwave schemes that interact with the atmosphere, cloud and precipitation fields, as well as the surface (Dudhia 1989). Radiation calculations were performed every thirty minutes for all model simulations presented here. The radiation interacts with an explicit micro physical treatment of moisture variables, which are included with separate cloud water, rainwater, snow, and ice after Dudhia (1989). This scheme only becomes active whenever grid-scale saturation is reached, and was used in all three domains. Grell's (1993) implicit cumulus parametrization scheme was utilized in the 36 km and 12 km resolution domains, while no implicit scheme was used in the highest resolution, 4 km domain, explicit microphysics were calculated.

The MM5 allows for the inclusion of terrain and land use information. For the coarse (36 km) domain, the terrain data are interpolated from the five minute (9.25 km) resolution data set from the National Geophysical Data Center. For the two nested domains (12 km, 4 km), the terrain data are interpolated from the Defense Mapping Agency 30 second (0.925 km) resolution data set. All three domains utilize the PSU-NCAR global ten minute (18.5 km) resolution land use data set. This data set consists of thirteen different categories to identify the primary characteristics of land use within a grid box. Parameters which are dependent upon the land use category are albedo, moisture availability, emissivity, roughness length, and thermal inertia.

The JASBEX simulation was initialized with archived synoptic data for the period of 0000 UTC 26 July 1995 through 0000 UTC 28 July 1995. The data are from the National Center for Environmental Prediction (NCEP), formerly the National Meteorological Center (NMC), large scale analyses, which are at a 2.5 by 2.5 degree resolution. These data are available at the standard pressure levels of 1000, 850, 700, 500, 400, 300, 250, 200, 150, and 100 mb, the model top. A relaxation boundary condition is used which allows for the inclusion of large scale information along four grid points around the perimeter of the coarse domain based on the 12 hour archived analyzed fields. To enhance the mesoscale information content of the initialization fields, any archived observations from standard synoptic surface, ship, buoy, and upper air sites within the area of interest, were included before the interpolation to the three model domains was made. The mesoscale data collected during the JASBEX Intensive Observation Period (IOP) were not included in these initializations based on synoptic observations, and therefore comprise a set of observations which are independent of the simulation.

3. Data

JASBEX consisted of a network of stations taking surface and upper air observations during two intensive observation periods defined earlier. Fig. 2a shows a map of the NWS JAX County Warning Area with the three letter identifiers of synoptic reporting stations across the area. The stations that were part of the JASBEX mesoscale data network are shown in Fig. 2b. Fig. 2c shows the stations used in this paper. Future studies will involve the use of the entire data network. The data collected during the experiment consisted of surface observations of temperature, DuPont, relative humidity, wind speed and direction, pressure, cloud cover, and rainfall using both human observers and automated sites. Additionally, four sites were tasked with the collection of upper air data.

This study presents the comparison of time series of temperature, dew point, and wind speed and direction for stations shown in Fig. 2c, the 88D reflectivity images, and model simulations for specific times. The model simulations used are from the innermost grid (4 Km resolution) that is part of the triple nested grid domain explained in the model section.

Specifically, the simulations shown in this study show the horizontal wind at 0.53 km, the vertical circulation, surface temperature, cloud mixing ratio (CMR) greater than .0002 kg/kg, and rain mixing ratio (RMR) greater than .00194 Kg/Kg. These, of course, are not the only parameters generated by the model. They were selected for display as they show the onset of the sea breeze, the cloud field associated with it (cloud mixing ratio), and the precipitating clouds (rain mixing ratio).

The time series were first compared to 88D images to identify areas of sea breeze involved convective initiation. These in turn, are then contrasted against the cloud and precipitation fields simulated by the model. Future studies, include the analysis of satellite data to more accurately assess the onset of the sea breeze. It is also important to understand that the data collected during the sea breeze experiment was not used during the initialization of the mesoscale model. The model, as described in the model section, was initialized using NMC model generated prognostic fields and the synoptic scale network of data. Nonetheless, the comparison of high resolution (4km) model results with the collected data is considered a measure of the potential utility of high resolution models in areas of complex sea breezes.

One feature of the WSR-88D data is the prominent interference spike extending southeast from the radar which will not be present in other data since it is electronic interference rather than valid radar reflectivity. The time periods selected represent the late afternoon of two days when convection developed in North Florida. Also, the apparent boundary in the reflectivity images across Charlton and Baker Counties is a geographic feature known as the Trail Ridge.

4. Results

In this section we compare the complex LSBS from the mesoscale observations, Doppler radar imagery, and the MM5 computer simulation runs from the synoptic data.

a. Comparison of the LSBS on July 26, 1995

We begin by comparing the 88D reflectivity images with the time series of the meteorological fields. Figure 3 shows the 88D reflectivity image corresponding to July 26, 1995 valid at 1852Z. In this image the sea breeze shows up as a weak line of enhanced reflectivity from 10 to 25 dbz at the eastern edge of the ground clutter extending along the coast from Camden County in SE Georgia through Flagler County in Florida parallel to, but east of the St. Johns River. Figure 4. shows the same image, but valid at 2055Z. This image shows the sea breeze from Central Camden County in Georgia through the border of Putnam and Flagler Counties in Florida with convection along the St. Johns River and additional convection inland of the sea breeze front, near Gainesville (GNV).

Comparing these images with surface data for Fernandina Harbor Marina (FHM), Kings Bay (KNBQ), and St Augustine (SGJ), all show winds shifting from southwest to southeast and dewpoints rising prior to 1900Z (Figs. 5a, b, c). The St. Augustine data shows the earliest sign of a sea breeze (between 15Z and 16Z) which coincides well with Fig. 3, where the sea breeze appears well inland over St. Johns County two to three hours later. While the temperature field does not reflect the expected cooling as a sign of sea breeze passage at FHM and SGJ, all coastal stations clearly show the wind shift from light westerly winds to the southeast along with increasing wind speeds (Fig.5). The synoptic flow on the 26th was west to southwest across the area.

A fourth station, Green Cove Springs (GCS, Fig. 5d), shows an earlier wind shift to the south and subsequent cooling and moisture increase from 1700Z to 1800Z earlier than FHM and KNBQ. The sea breeze at this time was located east of GCS (refer to Figs. 1 and 3 for location); however, looking closer at Fig. 3, there is another boundary further inland (well ahead of the sea breeze) depicted in this image across Southern Clay, Putnam, and Eastern Marion Counties in FL. This represents a river breeze from a wide section of the St Johns River moving westward ahead of the actual sea breeze front, and initiating some convection in Clay, Putnam, and Marion counties. The fact that SGJ show a sea breeze signature about the same time as GCS (Fig. 5) is further evidence that this reflectivity boundary is the river breeze. The authors speculate that the convective activity over Flagler County in Fig. 3 is the result of convergence between the eastward moving branch of the St. Johns River breeze and the westward moving subareas.

Figure 6, depicts the model simulation of horizontal winds, vertical circulations, CMR, RMR, and surface temperature at 20Z over Southeast Georgia and Northeast Florida. Comparing this simulation with Figs. 3 and 4, which show the 88D reflectivity images one hour before and after 20Z, it is evident that the model captures the sea breeze moving inland in the horizontal wind and associated cloud field (CMR) in the same general location as shown by the 88D images and in agreement with the observed time series. Notice also that the model shows a line of convergence further inland from the sea breeze with some cloud development as well as convective activity as depicted by the CMR and RMR, respectively, and the rain cooled surface. It is certainly encouraging to see the model has the right idea of the overall picture. However, it is far from certain if this convective activity is related to a river breeze as the resolution of the model is insufficient to resolve the St. Johns River breeze.

b. Comparison of the LSBS on July 27, 1995

Figure 7 shows the 88D reflectivity image for July 27, 1995 valid at 1759Z. This image shows the sea breeze located across inland sections of Glynn and Camden Counties in Southeast Georgia and through Nassau, Duval, and Western St Johns and Flagler Counties parallel to the St Johns River. Since the synoptic flow on the 27th was mostly out of the south to southeast, the sea breeze moved faster and further inland than the 26th. Figure 8 shows the reflectivity image valid at 1903Z. In this image the sea breeze is located across inland sections from Charlton County in Georgia through Marion County in FL with convection along and behind the sea breeze front.

July 27 is a case where the LSBS signature is difficult to identify in the mesoscale data set. Skies were mostly cloudy and convective activity was abundant across the study area. The authors believe that due to rain induced surface cooling associated with the activity and the prevailing southeasterly flow during the day, the surface data did not indicate a sea breeze signature. This is evident in Figs. 9a, b, c, d, e, and f (dew points for MAN were not available). After reviewing all the 88D time sequences, it is believed that the fluctuations in temperature and dew point in some of the stations is dominated by secondary circulations associated with the convective activity, rather than sea breeze effects.

Figures 10 and 11 show the model simulations valid at 19Z and 20Z. The model shows the sea breeze pushing inland at 19Z as depicted by the cloud field (CMR) thus moving it a bit slower than shown by the 88D. Fig. 11 shows the convective activity just pushing inland by 20Z as depicted by the RMR. Overall, the model does a good job in simulating the convective activity offshore as well as inland, but is a bit slow in moving this convective activity onshore.

While the MM5 does a surprisingly good job of forecasting sea breeze generation at greater than 24 hours, the authors believe that significant improvement could be achieved easily. The factors which we believe most severely limit the effectiveness of the model in a 24 hour forecast of the onset and spatial location of the sea breeze convective activity are: 1) the model is not representing the St. Johns River and Okefenokee Swamp in its treatment of the surface processes and 2) the resolution of the domain used to run this simulation (4 Km) is not fine enough to resolve the river breeze development and its impact on the convective initiation. Future improvements in resolution and parametrization of surface processes, in particular the surface moisture flux, are planned for the next study phase. For example, by introducing permanently wetted surface conditions at grid points along the river and in the Okefenokee Swamp, we may help generate the latent heat flux and other surface processes currently missing in the model. This should significantly enhance the river breeze.

5. Summary and Conclusions

The preliminary results shown in this brief paper are certainly encouraging. Summer time sea breeze convective activity is a daily event with important effects on the local economy and social life. Being able to use a mesoscale model to forecast (even in a probabilistic sense) the onset of convective activity, in temporal as well as spatial dimensions, is a promising new addition to short term forecasting. Based on the results seen here, a slightly more sophisticated model may lead to greatly improved mesoscale thunderstorm initiation forecasts out to 24-30 hours. Post initiation modeling of subsequent thunderstorm activity is not a credible possibility, due to the chaotic nature of the convective process once it is initiated.

Though the potential exists to accomplish the goal of improving short term forecasting by the use of mesoscale numerical models, there are some model improvements that are required before that goal can be achieved. First, we need to develop a reliable mesoscale data network (mesonet) that takes observations on a daily basis to use for model post-initialization as well as validation. In fact, data gathering at this scale already exists, what is necessary is to collect all sources of data and input the data electronically into a single database which can be used by the mesoscale numerical model. This is still not a trivial task. Second, local surface fluxes and antecedent soil moisture effects need to be provided to the model. Specifically, the important effects of the estuarine wetlands as well as swamps and rivers need to be physically represented in the model as permanently wetted surfaces. Additionally, improved soil moisture parametrization through the input of the surface wetness from remote sensors from the previous diurnal cycle at a scale which represents the cumulus activity (1-2 km) should substantially improve the surface flux partitioning between sensible and latent heating. Current and future tools to ingest NEXRAD Doppler radar precipitation and GOES soundings will provide the necessary capability.

Finally, the effects of the different types of synoptic flow regimes on the LSBS domain need to be studied through a systematic series of numerical simulations when the above listed model improvements have been incorporated. This study shows different signatures of the sea breeze in the surface data when the synoptic flow changes from a westerly to an easterly component. The classical notion that the sea breeze signature always includes a shift in the wind field, falling temperatures, and rising dew points is not valid as was made evident in the surface mesoscale data for the 27th. The synoptic flow and cloud field also plays an important role when forecasting how deep inland the sea breeze will penetrate and how strong its signature will be, clearly affecting the character of the associated convection as shown in this study and previous research (Xian and Pielke, 1991; Wai et al. , 1996).

Acknowledgements

All data sets used for the initial and boundary conditions in the modeling component of this study are available from the NCAR data archives. Modeling research was supported by an appointment to the COMET Fellowship Program sponsored by the National Weather Service and administered by the University Corporation for Atmospheric Research under a Cooperative Agreement with the National Oceanic and Atmospheric Administration (NOAA Award Nos. NA37WD0018-01 or NA67WD0097). Computational support has been provided by the National Center for Atmospheric Research where the preprocessing of the model run was conducted and by the Florida State University's Academic Computing and Network Services for supporting the model simulation on their Silicon Graphics Power Challenge Machine cluster.

Partial funding for JASBEX was provided by the National Weather Service Southern Region, the Naval Oceanography and Meteorology Command, and a COMET Partners Project which funded some of the data gathering and data entry. Thanks to Al Sandrik of the National Weather Service for sharing leadership of the field phase of JASBEX, to LCDR Barbara Ives for leadership of Naval Reserve participation, and LCDR Jerry Macke for active Navy participation. The Florida State University (FSU) Department of Meteorology provided some of the equipment used to launch and track the PIBALS. FSU students Jeremy Walworth and David Knollhoff were instrumental in training PIBAL teams, and data conversion of the PIBAL data.

Jacksonville University students, particularly Kim Parks, are gratefully acknowledged for their data entry efforts. Finally we wish to acknowledge our numerous military, students and volunteer participants (both government and private citizens), without whose help, JASBEX and this study would not have been possible.

Authors:

Pat Welsh graduated from the U. S. Naval Academy (USNA) where he was one of the first graduates with a major in Oceanography, later he graduated with a Masters of Science degree in Meteorology from the U. S. Naval Postgraduate School in Monterey, California. His thesis work was part of the first shipboard attempt to measure high frequency atmospheric turbulence from a moving ship platform. Later, he concurrently served as an instructor at USNA, and the Laboratory and Technician Manager in the USNA Oceanography Department. Pat has a broad range of scientific interests including turbulent flows, nonlinear acoustics, remote sensing of the earth's atmosphere and oceans by acoustic, laser, radar and satellite sensors, and the complex biogeochemical cycles of carbon and nitrogen compounds. He completed his dissertation at Florida State University, under a NASA - Florida Space Grant Consortium Fellowship, again dealing with turbulent convective boundary layers.

Currently, as Science and Operations Officer for NWS Jacksonville, Dr. Welsh has been heavily involved in the opening and outfitting of the new Jacksonville NEXRAD facility, and training the staff in the use of advanced technologies in operational forecasting. In addition to his long standing interests in boundary layers and remote sensing of the atmosphere and oceans, his current research focus is on tornadic storms within hurricanes, use of the Doppler radar in mesoscale severe weather forecasting, and sea breeze convection. He is a member of the NWA and AMS.

Pablo Santos is a native of Bayamon, Puerto Rico. He was awarded a Bachelor of Science degree (Summa Cum Laude) in Physics from the University of Puerto Rico, San Juan, in June 1992. In April 1995, he completed his masters’ degree in Meteorology at the Florida State University. He attended graduate school under support of the NASA's Graduate Student Researcher Program (GSRP) at Goddard Space Flight Center (GSFC). During his graduate work he attended two conferences sponsored by the GSRP program at GSFC where he presented results from his research. He also worked, during the summer of 1994, at GSFC as part of the Graduate Student Summer Program (GSSP) sponsored by the University Space Research Association (USRA). Pablo has co-authored five abstract publications and three paper publications submitted to the Journal of Applied Meteorology. He is currently a forecaster at NWS Jacksonville, and a doctoral candidate at Florida State University. Pablo is a member of the NWA and AMS.

Christopher Herbster completed his Master's in Meteorology at the Florida State University in 1990, and then had an opportunity to work with Dr. Anne Thompson of the NASA Goddard Space Flight Center through a NASA fellowship. It is through this experience that Christopher was first exposed to numerical modeling. After the completion of the NASA fellowship, Christopher began to work on the current sea breeze topic. In July of 1994, Christopher and Dr. Paul Ruscher conducted the Tallahassee Area Sea Breeze Experiment (TASBEX), which comprised a significant component of this current work. In the fall of 1994, Christopher was a visiting scientist at the National Center for Atmospheric Research (NCAR), where he worked with Dr. Bill Kuo's Mesoscale and Microscale Meteorology (MMM) Division. After completion of his dissertation on numerical modeling the Florida sea breeze and convection, Chris worked with the Tallahassee National Weather Service Office and the Cooperative Institute for Tropical Meteorology (CITM), at FSU, through a funded postdoctoral fellowship under the Cooperative Program for Operational Meteorology, Education and Training (COMET) Outreach Program. Dr. Herbster is a member of the Meteorology faculty at Embry Riddle Aeronautical University in Daytona Beach, FL.

Blackadar, A. K., 1976: Modeling the nocturnal boundary layer. Preprints of Third Symposium on Atmospheric Turbulence and Air Quality , Raleigh, NC, 19-22 October 1976, Amer. Meteor. Soc., Boston, 46-49.

, 1979: High resolution models of the planetary boundary layer. Advances in Environmental Science and Engineering , 1, No. 1, Pfafflin and Ziegler, Eds. Gordon and Briech Sci. Publ., New York, 50-85.

Byers, H. R., and H. Rodebush, 1948: Causes of Thunderstorms of the Florida Peninsula. J. Meteor., 5, 275-280.

Deardorff, J. W., 1978: Efficient prediction of ground surface temperature and moisture, with inclusion of a layer of vegetation. J. Geophys. Res. , 83 , 1889-1903.

Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46 , 3077-3107.

, 1993: A Nonhydrostatic version of the Penn State-NCAR Mesoscale Model. Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev. 121 , 1493-1513.

Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus
parametrizations. Mon. Wea. Rev. , 121 , 764-787.

, J. Dudhia, and D. R. Stauffer, 1994: Description of the Fifth-Generation Penn State/NCAR Mesoscale Model (MM5). NCAR Technical Note NCAR/TN-398 + STR, National Center for Atmospheric Research, Boulder, CO, 138 pp.

Herbster, C. G., 1996: An observational and modeling study of the Florida panhandle and Big Bend sea breezes. Ph.D. Dissertation, Dept. Of Meteorology, Florida State University, Tallahassee, FL 32306, pp TBD.

Rubes, M. T., H. J. Cooper, and E. A. Smith, 1993: A Study of the Merritt Island, Florida Sea Breeze Flow Regimes and Their Effect on Surface Heat and Moisture Fluxes. NASA Contractor Report 4537, Marshall Space Flight Center, Hunstville, Alabama, 141 pp.

Tunney, D. A., 1996: Numerical studies of the Georgia Coast Sea Breeze. Master's Thesis, Department of Meteorology, Florida State University, Tallahassee, FL 32303-3034, 166 pp.

Ulanski, S. L., and M. Garstang, 1978: The role of surface divergence and vorticity in the life cycle of convective rainfall. Part I: Observation and analysis. J. Atmos. Sci., 35, 1047-1062.

Wai, M. M.-K., P.T.Welsh, and W.-M. Wa, 1996: Interaction of secondary circulations with the summer monsoon and diurnal rainfall over Hong Kong. Boundary Layer Meteor., 5, 1-24.

Xian, Z., and R. A. Pielke, 1991: The effects of width of landmasses on the
development of sea breezes. J. Appl. Met. , 30 , 1280-1304.

Zhang, D. L., and R. A. Anthes, 1982: A high-resolution model of the planetary boundary layer-sensitivity tests and comparisons with SESAME-79 data. J. Appl. Meteor. , 21, 1594-1609.

¹ Current Affiliation: National Weather Service 13701 Fang Dr, Jacksonville, FL 32218

² Current Affiliation: National Weather Service 13701 Fang Dr, Jacksonville, FL 32218

³ Current Affiliation: Embry Riddle Aeronautical University, Dept. of Meteorology, Daytona Beach, FL .