Close Window

Florida State University: "Research to improve forecasts at Cape Canaveral, Florida"

Final Report


1.1 Introduction

This research was a collaborative project between Florida State University (FSU) and the 45th Weather Squadron (45 WS). The overall objective was to improve forecasts of sea and river breezes and their associated convection in the Cape Canaveral area of central Florida. Four specific research phases were conducted during the four-year life of the project:

1.2 Personnel

Personnel at Florida State University performed most of the "hands on" research. Prof. Henry E. Fuelberg was the Principal Investigator at FSU. Four graduate research assistants also participated in the project. Air Force Capt. Jeff Cetola developed the climatology of the sea breeze in the Cape area (Phase 1). Capt. Cetola graduated with an M.S. in meteorology during 1997. Mr. Anil Rao performed mesoscale simulations of various sea breeze phenomena in the Cape area (Phase 2). He obtained his Ph.D. in December 1998. Air Force Capt. Jon Kelly investigated conditions associated with sea breeze-induced thunderstorms in the Cape area and developed guidance tools for forecasting their development (Phase 3). Capt. Kelly received his M.S. degree during the summer of 1998. Capt. Chris Cantrell investigated nocturnal thunderstorms at Cape Canaveral (Phase 4). He graduated during the summer of 1999.

Personnel at the 45th Weather Squadron provided expertise on operational meteorology in general, and on specific local forecast issue in particular. Mr. William P. Roeder was the Co-Principal Investigator at the 45 WS. Mr. Roeder maintained the operational focus of the research, served as liaison between the FSU researchers and personnel at Cape Canaveral, and facilitated the acquisition of needed data. The Applied Meteorology Unit (AMU) that is co-located with the 45 WS provided data and information about their own research activities.

1.3 Summary of Research

Phase 1 - Sea Breeze Climatology

Capt. Jeff Cetola's master's research was entitled "A Climatology of the Sea Breeze at Cape Canaveral, Florida". He utilized several types of data, most of which were obtained from the 45 WS. These data included satellite imagery for 1995 (from the AMU) and 1996 (from the FSU ingest system), observations from the Cape area mesonetwork, radiosonde data for the Cape, surface and upper air analyses from the National Center for Environmental Prediction (from FSU archives), and surface observations to document the locations of thunderstorms in the area. To facilitate the display and manipulation of the mesonet data, the Spaceflight Meteorology Group at NASA's Johnson Space Center provided McIDAS software (the Tower J and X Plot packages). Capt. Cetola examined the data on a day-by-day basis, noting the times of sea breeze passage at several locations near the Cape as well as its degree of inland penetration. A total of 357 days was analyzed. These days were classified as sea-breeze days, non-sea-breeze days, or undetermined. The undetermined days were removed, leaving a total of 317 days. River breezes and other local circulations were analyzed and related to the sea breeze, and the presence of convection was related to sea breeze occurrence and large-scale flow.

An onshore sea breeze was observed on 194 of 317 days (61%) during the warm season. The sea breeze was likely to form on days with large-scale flow from any direction except northeast. The average time of sea-breeze passage at tower 112 was found to be 1528 UTC. The sea breeze penetrated the entire Cape Canaveral tower network (30 km) on 81% of the 194 sea-breeze days. Inland penetration was reduced, and passage time was delayed, for offshore flow greater than 4 m s-1. The river breezes that were observed on 116 days tended to occur during weak large-scale flow. A trailing convergence line was observed behind the sea-breeze front on 30 days. This line also formed on days with weak large-scale forcing. Thunderstorms were observed on 53% of the sea-breeze days, being most likely when the large-scale flow was from the southwest.

Thresholds were established for the onshore and offshore large-scale wind components associated with inland sea-breeze occurrence. Observations indicated an offshore maximum of 12.9 m s-1 and an onshore maximum of 6.7 m s-1 for the Cape Canaveral area. In general, sea breezes occur over the Cape Canaveral area for onshore flow no greater than 6 m s-1 and offshore flow no greater than 10 m s-1. These values are somewhat greater than those derived from previous, numerical studies. Thus, the results indicated that some findings from two-dimensional modeling studies are not applicable to Cape Canaveral's complex land/sea interface.

Phase 2. Mesoscale Numerical Simulations of Sea Breeze Phenomena

This phase of the research constituted the Ph.D. dissertation of Mr. Anil Rao. Three-dimensional numerical simulations of land/water circulations near Cape Canaveral, FL were performed using the Advanced Regional Prediction System (ARPS). A three-nested configuration was employed, i.e., at 1.6 km, 400 m, and 100 m. The structure of horizontal convective rolls (HCRs) and their influence on convection along the Cape Canaveral sea breeze front were investigated. In addition, the role of Kelvin-Helmholtz instability (KHI) in determining the location and time of convection behind the sea breeze front was examined. The model configuration attempted to improve on limitations of previous work (e.g., resolution, surface characteristics, initial state, etc.). The model provided a detailed, realistic simulation of the mutual interactions of features

Two simulations were presented in the dissertation. Results from the first indicated the formation of HCRs over the heated land surface at periodic intervals. These HCRs had both large and small spatial scales. When aspect ratios were based on both scales of HCRs, values were smaller than those of most previous observations. However, when only the larger HCRs were considered, agreement was closer. We speculated that previous methods for determining HCR aspect ratios from satellite imagery produced the discrepancy. Specifically, since simulated cloud water was present only above the larger HCRs, the smaller circulations would not be visible from a satellite. The smaller HCRs eventually dissipated or merged with their larger counterparts, intensifying the vertical motion within the larger circulations.

An HCR was shown to intersect the Indian River breeze. The complex orientation of the two convergence zones, a result of the realistic physiography used in the simulation, created a misocyclone at the intersection. The misocyclone produced an area of intense surface convergence that initiated the development of a strong precipitating storm.

Kelvin-Helmholtz instability (KHI) occurred along the top of the IRB interface in the form of billows (KHBs). The updraft portions of the KHBs were preferred regions for post-frontal storms. This is due to the greater vertical transport of moisture in these regions, which then requires less lift for convection initiation. However, post-frontal storms did not develop since additional lift was not present.

The second case exhibited a single precipitating storm that formed behind the sea breeze front. This post-frontal storm developed when an outflow boundary intersected a deep layer of upward motion above the marine air. This region of ascent initially was the remnant of a cell that formed along the sea breeze front, but before the cell decayed, a portion of its upward motion was intensified and displaced by KHI that was occurring on top of the sea breeze interface. This region of enhanced ascent then moved backward with the KHB until the outflow boundary arrived. Thus, KHI was critical in determining the location and time of storm development along the outflow boundary.

Phase 3. Develop Local Thunderstorm Forecasting Tools

Capt. Jon Kelly's M.S. research was entitled "Predicting East Coast Sea Breeze initiated Convection Near Cape Canaveral, Florida". Using data from the 1996-97 warm seasons, Capt. Kelly found 184 sea breeze days in the Cape area. To isolate East Coast Sea Breeze (ECSB) initiated convection, days were eliminated when outside forcing mechanisms other than the ECSB were indicated. This left 120 days on which only the ECSB affected the area. These days were categorized according to convective activity based on NEXRAD reflectivity data and lightning data. The categories were: days when no cells formed along the sea breeze front (NCD), days when cells formed but no cloud-to-ground lightning (CGL) occurred (CD), and days when cells formed and developed into thunderstorms with CGL (LD). Our primary goal was to determine whether one could discriminate between a LD and a CD using 1000 and 1500 UTC radiosonde sounding parameters and surface area-averaged horizontal divergence. Each parameter's ability to forecast a LD versus other days then was assessed. We also documented characteristics of undisturbed ECSB days.

Eight stability parameters from the 1000 and 1500 UTC radiosonde soundings were calculated. Results showed that only the K Index at 1000 UTC statistically discriminated a LD from a CD. Relative humidity (RH) and u and v wind components from 1000 to 400 mb also were calculated for both sounding times. There were no statistically significant differences in RH between LD and CD at any pressure level for either sounding time. Likewise, the u wind component produced no statistically significant discrimination between the LD and CD categories. However, there were significant differences in the v component at almost all levels at both sounding times. The v wind component at 700 mb produced the best statistical separation of all parameters studied. There were no statistically significant differences between the temperature profiles of LD and CD for either sounding time.

The forecasting skill of parameters was measured using the True Skill Statistic (TSS). Forecasting skill was based on a forecast of a LD versus a non- LD, i.e., a CD or a NCD. The K index showed the most forecasting skill, followed by the v component of the 850 mb wind. Surface area-averaged horizontal divergence (AAD) was found to be of little use as a forecasting tool for undisturbed ECSB initiated convection, both in the nowcasting (0-3 h) or longer term (3-12 h) ranges.

Capt. Kelly conducted several follow-on tasks before he left FSU. These tasks will permit his research to be combined with that of Capt. Jeff Cetola (Phase 1) to form a journal manuscript for Weather and Forecasting. Those calculations took longer than anticipated; therefore, undergraduate student Todd Lericos completed them. Also, at the request of Mr. Roeder, Capt. Kelly calculated values of the Neumann-Pfeffer Index to compare with values of stability indices that he presented in his thesis. The N-P Index currently is used by forecasters at the 45WS.

Phase 4. Investigation of Nocturnal Thunderstorms

Capt. Chris Cantrell's M.S. research explored nocturnal convection in the Cape Canaveral area. Two aspects of nocturnal convection over Florida were examined: A climatology of all warm season nocturnal cloud-to-ground lightning (CGL) flashes over east central Florida, and nocturnal CGL that develops near Cape Canaveral, Florida during undisturbed conditions. Nocturnal storms forming during undisturbed flow are difficult to forecast because they form over the Kennedy Space Center/Cape Canaveral Air Station with little or no warning.

Seven years of warm season radiosonde data (1992 - 1998) were used to categorize nights into one of five wind regimes based on the 1000 - 700 mb mean vector wind. Cloud-to-ground (CGL) flashes measured by the National Lightning Detection Network (NLDN) were used to examine the spatial distribution of lightning. Flashes were found to occur on 64 percent of the nights with available soundings (525 nights). Spatial maps of CGL flashes were constructed for each of the five wind categories. CGL flash statistics for the categories were related to values of lapse rate and moisture. Streamline analysis was used to document flow patterns for the five wind regimes.

Flow parallel to the peninsula from the south was found to have the greatest percent occurrence of nights with CGL flashes. Enhanced midlevel moisture (700 - 500 mb) increased the percent occurrence of nights having CGL flashes. Midlevel lapse rates (850 - 500 mb) showed little variation, and the more unstable midlevel lapse rates did not increase the likelihood of nights with CGL flashes. Streamline analyses revealed that differences in synoptic flow are crucial in determining thermodynamic characteristics of the air flowing over the study area as well as the spatial distribution and number of CGL flashes.

Data from the local lightning detection network at the Kennedy Space Center/ Cape Canaveral Air Station revealed 74 nights during the warm season months of 1995 - 1997 with cloud-to-ground lightning within our 40-km radius of study. Surface analyses then were used to eliminate nights with disturbed flow, while radar data were analyzed to eliminate pre-existing convection. A data set of nights without CGL also was constructed. Stability, temperature, wind, and moisture parameters of the two categories were examined statistically.

Wind speed and u and v wind components did not statistically discriminate between nights with or without CGL. Results indicated that temperatures at individual levels also did not show statistically significant differences. The 1000 - 800 mb lapse rate was the only temperature related parameter that statistically discriminated between the two categories. Relative humidity at 850, 800, and 750 mb all showed statistically significant discrimination between categories. Only two stability parameters (K-Index and Showalter Index) exhibited statistically significant discrimination between CGL and NCGL nights. The K-Index was found to have the greatest forecast skill of all the standard stability parameters. A Modified K-Index was created using the 1000 - 800 mb lapse rate and the 850 - 700 mb moisture. This new index provided the best statistical discrimination and forecast skill of all the wind and thermodynamic parameters examined.


The following exchanges were conducted at or by the 45WS in addition to the research activities cited in Section 1.3:


Cantrell, C.E., 1999: Predicting warm season nocturnal cloud-to-ground lightning near Cape Canaveral, Florida, M.S. Thesis, Florida State University, 71 pp.

Cetola, J.D., 1997: A climatology of the sea breeze at Cape Canaveral, Florida, M.S. Thesis, Florida State University, 56 pp.

Kelly, J.L., 1998: Predicting east coast sea breeze initiated convection near Cape Canaveral, Florida, M.S. Thesis, Florida State University, 49 pp.

Kelly, J.L., H.E. Fuelberg, and W.P. Roeder, 1998: Thunderstorm predictive signatures for the east coast sea breeze (ECSB) at Cape Canaveral, Air Station (CCAS) and the Kennedy Space Center (KSC), Preprints, 19th Conf. Severe Local Storms, Amer. Meteor. Soc., Minneapolis, 677-680.

Rao, P.A., and H.E. Fuelberg, 1998: A numerical modeling study to investigate the role of Kelvin-Helmholtz instability on convection near the Cape Canaveral sea breeze front, Preprints, 19th Conf. Severe Local Storms, Amer. Meteor. Soc., Minneapolis, 380-383.

Rao, P. A., 1998: A numerical modeling investigation of the Cape Canaveral Land/Water Circulations, Ph.D. dissertation, Florida State University, 100 pp.

Rao, P.A., H.E. Fuelberg, and K.K. Droegemeier, 1999: High resolution modeling of the Cape Canaveral land/water circulations and associated features, Mon. Wea. Rev., 127, 1808-1821.

Rao, P.A., and H.E. Fuelberg, 2000: An investigation of convection behind the Cape Canaveral sea breeze front. Accepted by Mon. Wea. Rev.

Roeder, W. P., F. J. Merceret, S. J. Sokol, S. T. Heckman, and G. E. Taylor, 1996: Meteorological operational research requirements on the central Florida Atlantic Coast in support of the United States space program. Preprints, 15th Conf. Weather Analysis and Forecasting, Norfolk, Amer. Meteor. Soc., 455-458.

Roeder, W. P., 1997: 45 WS Forecast Improvement Initiatives Via External Agencies, 22nd Annual National Weather Association Meeting, 19-24 October 1997, 1 pp.

Roeder, W.P., and C. Pinder, 1998: Empirical techniques for central Florida lightning forecasts in support of America's space program, Preprints, 16th Conference on Weather Analysis and Forecasting, Amer. Meteor. Soc., 475-477.

Roeder, W.P., C. Pinder, and A. A. Guiffrida, 1998: Lightning forecasting at 45th Weather Squadron, 23rd Annual Meeting, National Weather Association, October 1998.

Roeder, W.P. and D.E. Harms, 1999: 45th Weather Squadron forecast improvement initiatives via external agencies: An update, 24th Annual Meeting of the National Weather Association, 16-22 October 1999, 1 pp.


4.1 FSU Perspective

It is valuable for FSU faculty and graduate students, both Air Force and civilian, to be exposed to real life research problems such as encountered by the 45 WS. Furthermore, by working with the 45 WS, the FSU participants can utilize some of the best data sources and computer software available anywhere. Prof. Fuelberg regularly teaches the FSU graduate course in mesometeorology. Some of the knowledge gained from this COMET research has been used to update lectures on the sea breeze. In addition, the course's lab on sea breezes was modified to include use of data from the Cape area mesonetwork. Those data were displayed using the TWRJ and XPLOT software that were obtained for FSU by the 45 WS. Finally, some of the research topics for the Cape area also are applicable for the Tallahassee area. Thus, the AWS COMET research enriches the NWS COMET research that Prof. Fuelberg performs with the local NWS office.

Mr. Roeder, and all of the 45 WS participants, have been very helpful to us. They have provided a myriad of data as well as expert guidance on local forecasting issues and how our collaborative research can provide improvements to their operational forecasting.

FSU has not encountered any serious problems. We have been able to collaborate with the 45 WS and COMET to seek solutions to any minor problems that have developed.

4.2 45 WS Perspective

PHASE-1 (Sea Breeze And Thunderstorm Climatology, M.S. Thesis):

Operational Impacts: Improved forecasts of sea breeze front timing and in-land penetration, and the associated thunderstorms. Added results to 45 WS Terminal Forecast Reference Notebook, the primary reference for local forecasters and used in training new forecasters. We were so pleased with these results; we extended the topic into Phase-3 of this project.

PHASE-2 (Advanced Modeling Of Post Sea Breeze Convection, Ph.D. Dissertation):

Operational Impacts: Improved our capability to forecast post sea breeze convection. Learned many valuable lessons that will facilitate advanced local meso-scale numerical modeling in the future.

PHASE-3 (Thunderstorm Predictive Signatures Under Easterly Flow/Continued Sea Breeze Climatology, M.S. Thesis):

Operational Impacts: Improved our forecasts of thunderstorm probability under the easterly flow regime. Improved forecasts of sea breeze front timing and in-land penetration, and the associated thunderstorms. Added results to 45 WS Terminal Forecast Reference Notebook

PHASE-4 (Nocturnal Thunderstorm Predictive Signatures, M.S. Thesis):

Operational Impacts: Improved our capability to forecast nocturnal thunderstorms with no apparent driving mechanisms.

FLOW REGIME STRATIFIED LIGHTNING CLIMATOLOGY (Senior's Project, Cross-feed From FSU / NWS Tallahassee COMET Outreach Project):

Although this project was not done under the FSU/45 WS project, 45 WS definitely benefited from this work. It is an excellent example of inter-agency cooperation and synergistic interaction between operational research programs, especially within the COMET Outreach Program-they benefited from the FSU / 45 WS sea breeze studies and other 45 WS research, we benefited from the product. Mr. Watson, Scientific and Operations Officer from NWS/Tallahassee attended the final report meeting at 45 WS.

Ironically, we were trying to find a tool to train new forecasters on the importance of the position of the sub-tropical ridge in strongly influencing where thunderstorms form in Florida summer (strongly influences whether the east coast or west coast sea breeze will dominate). This pictorial flow regime stratified lightning climatology satisfies this need admirably.

Agreed to participate in extending the lightning climatology to include RAOB stratification and add numerical probabilities for 45 WS lightning warning areas. We will provide CCAS RAOB data and consult on RAOB signatures in forecasting CCAS/KSC, including the new index developed by the Air Force Institute of Technology, which in turn had benefited from earlier FSU/45 WS work (multiple synergistic interactions between multiple projects!). In exchange, 45 WS will get numerical lightning probabilities for our lightning warning areas stratified by flow regime. This will become a forecasting aid for the 45 WS's daily "7-Day Planning Forecast".

Operational Impacts: Much needed training aid on role of subtropical ridge. Subjective aid to daily lightning probability forecasts. The extension of this project, soon to be delivered, will provide stratified numerical probabilities for 45 WS warning areas as objective forecasts for daily lightning probability forecasts.


FSU's work helped 45 WS develop a new thunderstorm probability index with the Air Force Institute of Technology. This new index combines RAOB data, flow regime, patterns of persistence, and climatology. It has 49.8% better skill than our old index along with 68% better Probability Of Detection, 59% better (lower) False Alarm Rate, 51% better Critical Success Index (aka Threat Score), 57% better Hit Rate (percentage correct yes/no forecasts), and 10% better (lower) bias. The new index also beats the old index at every performance metric evaluated at every probability range. These amounts and breadth of improvement are phenomenal within operational meteorology. We consider this the biggest improvement in local thunderstorm probability forecasting in 30 years, perhaps the biggest improvement ever! Although FSU didn't do the actual development work, they definitely share in the credit.

FSU's work also helped in the Air Force Institute of Technology project to improve thunderstorm probability forecasts for day-2 through day-7 of the daily "7-Day Planning Forecast" using the NCEP numerical models. Results on this project are not yet finalized. - The FSU work has also helped guide 45 WS's Internal research efforts.

OVERALL: This project has been an outstanding success! The Range Weather Operations Flight Commander cited the FSU project as the "best of our operational research projects…makes other projects like the Air Force Institute of Technology weather program look like a waste of money." The FSU team showed outstanding operational focus in their work and cooperation with 45 WS. We also appreciate their understanding of real-world operational priorities in providing support to this project. Although we can not isolate how much of our improved forecasting and warnings can be attributed to FSU, I have personally witnessed warnings issued with better accuracy and better lead-times based specifically on FSU results. The FSU / 45 WS cooperative effort enjoyed the highest-level of enthusiasm from 45 WS. A Launch Weather Officer said, "we should continue extending the sea breeze thunderstorm climatology and thunderstorm predictive signatures." Such enthusiasm and appreciation for operational research is very rare from our operational meteorologists!

A brief synopsis of the major benefits to 45 WS follows. There are many smaller benefits are not itemized here. The COMET Outreach Program director was absolutely correct when she proclaimed the FSU / 45 WS project "a shining example of how the program should work."

We gratefully acknowledge the contributions of many agencies. Without their extraordinary cooperation and invaluable assistance, the FSU/45 WS project would not have been nearly as successful. Computer Sciences Raytheon provided archived weather data and software support. The Applied Meteorology Unit provided technical guidance and data sets. The Spaceflight Meteorology Group provided software support. And ACTA Corp. provided meteorological consultation and represented the 45th Space Wing's dispersion interests.

In summary, I believe our Commander said it best, "this is by far the best program I've come across in my career (27 years) as far as a cooperative arrangement between an operational weather unit and university researchers . . . a good return on the money."

The 45WS did not encounter major problems.