The second-order closure for the ensemble-mean dynamics is validated using the approach of direct numerical ensemble simulations of a linear barotropic model with stochastic basic flows in extratropics. For various configurations of the stochastic basic flow and external forcing, the deterministic solutions under the second-order closure capture, with remarkable accuracy, the ensemble means and the associated eddy covariance fields of forced responses simulated by a 500-member numerical ensemble. Thus, the second-order closure is found to be adequate for describing the ensemble-mean linear dynamics with stochastic basic flows. Example of ensemble-mean solution shows the important role played by the stochastic synoptic eddy component of the basic flow in determining the ensemble-mean responses to external forcing. This study supports the notion that linear frameworks of ensemble-mean dynamics under second-order closure are useful tools for describing and understanding the dynamics of the synoptic eddy and the low-frequency flow (SELF) feedback and extratropical atmospheric low-frequency variability. Following a similar concept, the conceptual recharge oscillator model for the El Niño-Southern Oscillation phenomenon (ENSO) is utilized to study the influence of fast variability such as that associated with westerly wind bursts (WWB) on dynamics of ENSO and predictability. The ENSO-WWB interaction is simply represented by stochastic forcing modulated by ENSO-related sea surface temperature (SST) anomalies. An analytical framework is developed to describe the ensemble-mean dynamics of ENSO under stochastic forcing. Numerical ensemble simulations verify the main results derived from the analytical ensemble-mean theory: the state-dependent stochastic forcing enhances the instability of ENSO and its ensemble spread, generates asymmetry in the predictability of the onsets of cold and warm phases of ENSO, and leads to an ensemble-mean bias that may eventually contribute to a climate mean state bias. Clouds contribute greatly to the atmospheric variability within weather systems. Measurements of thermodynamic properties in cloudy airs are required to improve numerical weather forecasting models and for the study of the global radiation and hydrology budget. The Global Positioning System (GPS) radio occultation (RO) technique is not affected by clouds and has a high vertical resolution, making it ideally suited for cloud study. Temperatures retrieved from Constellation Observing System for Meteorology Ionosphere & Climate (COSMIC) RO measurements are compared with two operational weather assimilation models including the Prediction and National Center for Atmospheric Research (NCEP/NCAR) reanalysis and European Centre for Medium-Range Weather Forecasts (ECMWF) analysis. The cloudy GPS ROs during June 2007 and June to September 2006 are identified based on the collocated CloudSat data. Systematic bias of opposite sign between large-scale global analyses and observed RO profiles are found for cloudy and clear-sky conditions. It is also found that GPS wet retrieval lapse rate is nearly constant (~6oC/km) in the vertical while that from ECMWF increases with height from cloud middle to cloud top. A new GPS RO cloudy profile retrieval algorithm is proposed. A relative humidity parameter is introduced through an empirical relationship between CloudSat ice-water content and ECMWF relative humidity. The new cloudy temperature retrieval tends to be warmer than the GPS wet retrieval within the cloud and slightly colder near the cloud top, resulting in a cloudy lapse rate that agrees more closely with that of the ECMWF in the lower part of the cloud and increases with height (but faster than that of the ECMWF), and reaches a value of about 7.6oC/km near the cloud top. When the ice-water content measurements are absent, an empirical value of 0.85 is shown to be a good approximation for the relative humidity parameter. / A Dissertation submitted to the Department of Meteorology in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Spring Semester, 2009. / March 26, 2009. / Cloud, CloudSat, COSMIC / Includes bibliographical references. / Xiaolei Zou, Professor Directing Dissertation; I. Michael Navon, Outside Committee Member; Ming Cai, Committee Member; Robert G. Ellingson, Committee Member; T. N. Krishnamurti, Committee Member.

A hurricane's destructive nature must be evaluated on as small a scale as possible to reveal the various types of mesoscale circulations that are embedded within the storm's overall wind field. This research develops a technique to estimate small scale surface wind speeds in hurricanes crossing Florida, and thereby identify areas of anomalous winds. Level II Doppler radar data are analyzed onto a high resolution (1 degree radial, 0.5 km gate) grid. An algorithm is developed to estimate the total wind speed from a combination of radial velocity and quality controlled reflectivity. These variables are utilized by identifying the location of the eye and then using radial velocities and an assumed symmetric wind field about the eye to estimate the total wind field over the entire radar scan area. Once the total wind field is computed along a scan, reduction factors are used to transpose the winds at the varying beam altitudes down to the surface using similarity theory. Case studies of Hurricanes Jeanne (2004), Frances (2004), Wilma (2005), Irene (1999), Ivan (2004) and Charlie (2004) are investigated. The success of the algorithm depends greatly on the ability of the Weather Surveillance Radar 88 Doppler (WSR-88D) to sample the velocity data and the ability to properly unfold it. Hurricane Wilma is an example of dry air being entrained into the cyclone, which produces an inadequate concentration of targets to provide a velocity profile, thereby resulting in poor results. Computed wind speeds are compared with National Weather Service (NWS) ASOS observations and independent wind observations supplied by the University of Florida. The estimated winds and those from the two datasets exhibit reasonable agreement; however, additional validation is needed to determine the actual skill of the algorithm. The observed data indicate that gust factors are not optimally estimated by applying a uniform percentage of the total wind speed. Further investigation is needed to determine the proper procedure for estimating wind gusts. Results also show that the algorithm can be used with some confidence to diagnose the damage potential for embedded tornadic cells located within the land-falling hurricane. / A Dissertation submitted to the Department of Earth, Ocean, and Atmospheric Science in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Spring Semester, 2011. / March 7, 2011. / Surface Winds, Florida, Doppler Radar, Hurricane / Includes bibliographical references. / Henry E. Fuelberg, Professor Directing Dissertation; Anthony Stallins, University Representative; Robert E. Hart, Committee Member; Guosheng Liu, Committee Member; Mark Bourassa, Committee Member.

Both observed and modeled global warming pattern shows a large surface polar warming and a large upper atmospheric warming in the tropics. This pattern leads to an amplification (reduction) of the temperature gradient at upper levels (surface). Physical processes behind this temperature change are the external radiative forcing, and subsequent feedback processes that may amplify or dampen the climate response. This unique warming pattern suggests that high latitudes are very sensitive to climate change and also the area where the largest warming projection uncertainties occur. The objective of this study is to apply a new coupled atmosphere-surface climate feedback-response analysis method to quantify the contributions of the external forcing alone (doubling of carbon dioxide), and subsequent feedback processes to the 3-D global warming pattern in the GFDL_CM2.0 model. The feedbacks under consideration include the water vapor feedback, surface albedo feedback, surface turbulent heat flux feedback, and the sum of the change in cloud radiative forcing (CRF), vertical convective, and large-scale scale dynamical feedback. The partial temperature changes due to the external forcing and due to individual feedbacks are additive and their sum converges toward the temperature change produced by the original GFDL_CM2.0 global warming simulations. Therefore, our attributions of the global warming patter to individual thermodynamic and dynamical processes are mathematically robust and physically meaningful. The partial temperature change due to the water vapor feedback is found to be the largest contributor to the globally averaged surface warming. It is twice as large as the warming due to the external radiative forcing alone. The surface albedo feedback and change in surface cloud radiative forcing increase the surface temperature by a smaller amount. In addition, the changes in atmospheric cloud forcing and large-scale dynamics, as well as the surface turbulent heat flux feedback, contribute to an overall damping the surface warming. In terms of spatial pattern of global warming, the external forcing alone would cause a large surface warming in the extratropics. The water vapor feedback strengthens the tropical warming substantially and the ice/snow albedo feedback contributes to polar warming amplification. The atmospheric dynamical feedbacks associated with the enhancement of vertical convection in the tropics acts to amplify the warming in the upper troposphere at the expense of reducing the warming in the lower troposphere and at the surface in the tropics. The dynamical feedbacks due to the strengthening of the poleward energy transport contribute to a warming in the entire troposphere and the surface in high latitudes. At the surface and in the lower troposphere, the additional warming brought by the change in circulations strengthens the warming due to thermodynamical forcings (e.g., external forcing, water vapor feedback, and ice albedo feedback). In the upper troposphere, the warming brought by the change in circulations dominates the cooling due to thermodynamical forcings. As a result, the entire troposphere becomes warmer. The stratospheric cooling is entirely due to the external radiative forcing. / A Dissertation submitted to the Department of Meteorology in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester, 2009. / July 23, 2009. / Climate Change, Radiative Forcing, Climate Response, Feedbacks, Sensitivity / Includes bibliographical references. / Ming Cai, Professor Directing Dissertation; William Dewar, Outside Committee Member; Xiaolei Zou, Committee Member; Paul Ruscher, Committee Member; Mark Bourassa, Committee Member.

Tropical cyclones can form from many different precursors, including baroclinic systems. The process of an extratropical system evolving into a warm core tropical cyclone is defined by Davis and Bosart (2004) as a Tropical Transition (TT) with further classification of systems into Weak Extratropical Cylclones (WEC) and Strong Extratropical Cyclones (SEC). It is difficult to predict which systems will make the transition and which will not, but the description of a common type of TT occurring along a front will aid forecasters in identifying systems that might undergo TT. A wind speed and SST relationship thought to be necessary for this type of transition is discussed. QuikSCAT and other satellite data are used to locate TT cases forming along fronts and track their transformation into tropical systems. Frontal TT is identified as a subset of SEC TT and the evolution from a frontal wave to a tropical system is described in five stages. A frontal wave with stronger northerly wind and weaker southerly wind is the first stage in the frontal cyclogenesis. As the extratropical cyclogenesis continues in the next two stages, bent back warm front stage and instant occlusion stage, the warmer air of the bent back front becomes surrounded by cooler air . Next, in the subtropical stage the latent heat release energy from the ocean surface begins ascent and forms a shallow warm core. As the energy from surface heat fluxes translates to convection within the system, the warm core extends further into the upper levels of the atmosphere in the final, tropical stage of TT. Model data from MM5 simulations of three storms, Noel (2001), Peter (2003) and Gaston (2004) are analyzed to illustrate the five stages of frontal TT. Noel is found to have the most baroclinic origin of the three and Gaston the least. / A Thesis submitted to the Department of Meteorology in partial fulfillment of the requirements for the degree of Masters of Science. / Summer Semester, 2007. / May 21, 2007. / Noel(2001), Gaston(2004), Front, QuikSCAT, Peter(2003), Tropical Transition / Includes bibliographical references. / Mark Bourassa, Professor Directing Thesis; Robert Hart, Committee Member; Phillip Cunningham, Committee Member.

The El Nino-Southern Oscillation (ENSO), the North Atlantic Oscillation (NAO), the Pacific Decadal Oscillation (PDO), and the Polar Vortex Oscillation (PVO) produce conditions favorable for monthly extreme temperatures and precipitation. These climate modes produce upper-level teleconnection patterns that favor regional droughts, floods, heat waves, and cold spells, and these extremes impact agriculture, energy, forestry, and transportation. The above sectors prefer the knowledge of the worst (and sometimes the best) case scenarios. This study examines the extreme scenarios for each phase and the combination of phases that produce the greatest monthly extremes. Data from Canada, Mexico, and the United States are gathered from the Historical Climatology Network (HCN). Monthly data are simulated by the utilization of a Monte Carlo model. This Monte Carlo method simulates monthly data by the stochastic selection of daily data with identical ENSO, PDO, and PVO (NAO) characteristics. In order to test the quality of the Monte Carlo simulation, the simulations are compared with the observations using only PDO and PVO. It has been found that temperatures and precipitation in the simulation are similar to the model. Statistics tests have favored similarities between simulations and observations in most cases. Daily data are selected in blocks of four to eight days in order to conserve temporal correlation. Because the polar vortex occurs only during the cold season, the PVO is used during January, and the NAO is used during other months. The simulated data are arranged, and the tenth and ninetieth percentiles are analyzed. The magnitudes of temperature and precipitation anomalies are the greatest in the western Canada and the southeastern United States during winter, and these anomalies are located near the Pacific North American (PNA) extrema. Western Canada has its coldest (warmest) Januaries when the PDO and PVO are low (high). The southeastern United States has its coldest Januaries with high PDO and low PVO and warmest Januaries with low PDO and high PVO. Although extremes occur during El Nino or La Nina, many stations have the highest or lowest temperatures during neutral ENSO. In California and the Gulf Coast, the driest (wettest) Januaries tend to occur during low (high) PDO, and the reverse occurs in Tennessee, Kentucky, Ohio, and Indiana. Summertime anomalies, on the other hand, are weak because temperature variance is low. Phase combinations that form the wettest (driest) Julies form spatially incoherent patterns. The magnitudes of the temperature and precipitation anomalies and the corresponding phase combinations vary regionally and seasonally. Composite maps of geopotential heights across North America are plot for low, median, and high temperatures at six selected sites and for low, median, and high precipitation at the same sites. The greatest fluctuations occur near the six sites and over some of the loci of the PNA pattern. Geopotential heights tend to decrease (increase) over the target stations during the cold (warm) cases, and the results for precipitation are variable. / A Dissertation submitted to the Department of Meteorology in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Summer Semester, 2007. / April 24, 2007. / NAO, PDO, ENSO, Climate Variability, Extremes, Stochastic / Includes bibliographical references. / James J. O'Brien, Professor Directing Dissertation; Bernd Berg, Outside Committee Member; Jon Ahlquist, Committee Member; Ming Cai, Committee Member; Paul Ruscher, Committee Member.

Rossby waves propagating along the extratropical tropopause have a dominant impact on weather at the surface in midlatitudes via baroclinic instability. While forecasts of waves by numerical weather prediction systems are limited by intrinsic unpredictability due to initial condition sensitivity, the statistical properties of these features should not contain biases if the models are optimally designed. In this study, statistical properties of upper level waves in forecasts are evaluated for the seasons 2008-2012. Wavelet transforms are used to analyze Rossby wave amplitude as a function of both wavelength and longitude. The wavelet transforms are applied to both analysis and forecast potential vorticity. The difference between the two reveals model error. This process is done for three operational centers: ECMWF, UKMET, and NCEP, over 5 winter seasons. A comparison is made of 5 dependencies: forecast center, interseasonal variation, longitude, lead time, and wavelength. Additionally, a wave breaking diagnostic is developed as a supplemental tool for determining wave structure errors in operational models. Breaking frequency determines how many points of longitude in the wave pattern on average per day were breaking. The analysis reveals: 1) On average, Rossby wave amplitude is under forecast in all 5 dependencies, statistically significant at the 95% confidence level, 2) The largest errors are found in regions of climatologically higher wave activity, 3) errors in ECMWF are smaller and accumulate more steadily than UKMET and NCEP, which demonstrate large error very early in forecasts, 4) The errors represent biases in the model, not a regression to the mean, and 5) There are large errors in the wave breaking frequency for all forecast centers. The results imply an inherent issue in operational models with forecasting wave amplitude. How the errors vary across the 5 dependencies and how the work may be expanded for further study are discussed. / A Thesis submitted to the Department of Earth, Ocean, and Atmospheric Science in partial fulfillment of the requirements for the degree of Master of Science. / Spring Semester, 2015. / March 5, 2015. / analysis, forecast, model, Rossby wave, tropopause, wavelet / Includes bibliographical references. / Jeff Chagnon, Professor Directing Thesis; Robert Hart, Committee Member; Ming Cai, Committee Member.

The hyper-spectral Cross-track Infrared Sounder (CrIS) on board Suomi National Polar-orbiting Partnership (NPP) supports a continuing advance in numerical weather prediction (NWP) for improved short- to medium-range weather forecast skills. The assimilation of CrIS brightness temperature observations in NWP modeling systems requires the data biases be properly estimated and removed from data. Both the solar radiation reflected by sea surface into the satellite viewing direction and the solar pumping that deviates the stratosphere from the local thermal equilibrium (LTE) introduce the significant biases in CrIS infrared shortwave observations. In this study, the effects of sun glint and nonlocal thermal equilibrium (NLTE) on CrIS data biases are assessed quantitatively. It is found that the newly-developed sun glint and NLTE models can dramatically reduce the CrIS data biases at infrared shortwave band during daytime. However, the biases still remain relatively large for CrIS infrared shortwave stratospheric channels after the NLTE correction. A further study confirms that the bias residuals after the NLTE correction mainly come from the input temperature profiles to the Radiative Transfer Model (RTM) not the RTM itself. It is found that the temperature profiles from the National Center for Environmental Prediction (NCEP) Global Forecast System (GFS), which serve as input to the Community Radiative Transfer Model (CRTM), have large cold biases in the upper stratosphere, leading to the large bias remnants of stratospheric channels. Compared with the temperature profiles from ERA Interim reanalysis, the cold biases of GFS temperature profiles increase with altitude and reach about 10 K near 1 hPa. / A Thesis submitted to the Department of Earth, Oceanic and Atmospheric Sciences in partial fulfillment of the Master of Science. / Spring Semester, 2015. / February 5, 2015. / Includes bibliographical references. / Ming Cai, Professor Directing Thesis; Peter Ray, Committee Member; Zhaohua Wu, Committee Member.

As reported by the University of Central Florida (UCF), Florida nearly leads the nation in fatal vehicle crashes due to fog and smoke conditions. Between 2002 and 2009, 299 deaths were due to vehicle crashes related to fog and smoke conditions. This is more than the amount of deaths by hurricanes and lightning strikes combined. It may be possible to reduce the number of fatalities and crashes by implementing an effective early warning system. A warning of impending fog conditions would allow DOT and other agencies the ability to monitor specific locations. However, fog is both spatially and temporally variable and surface observation equipment is widely dispersed. The challenge lies in the ability to forecast and detect the occurrence of fog from surface observations far removed from the location of fog occurrence. The spatial variability of fog frequency over the state of Florida is explored based on an evaluation of GOES Imager satellite data. A nighttime fog detection algorithm employing a bispectral thresholding technique involving brightness temperature differences (BTD) between two channels: 4 (10.7-μm) and 2 (3.9-μm) is presented. The performance of the fog product is validated using one year of AWOS/ASOS station observations in the period right before daybreak, showing moderate skill. The frequency of fog in Florida for 2012 is analyzed through application of this technique and is compared to interpolated fog frequencies based on ground observations. Seasonal and annual bias corrections are implemented to calibrate the satellite fog product observations and provide spatially continuous data of fog occurrence in Florida. While the satellite-derived fog product generally overestimated fog frequency, the pattern of fog occurrences agreed with the general spatial patterns found in station-derived climatologies, providing encouraging results. This analysis sets a basis for a satellite-based fog climatology that provides spatially continuous information of underlying fog dynamics. Future work involving assessments of satellite fog products over a multi-year period, as well as improved spatial resolution in the forthcoming GOES-R, will assist in furthering knowledge regarding regional fog risks and potentials. / A Thesis submitted to the Department of Earth, Ocean, and Atmospheric Sciences in partial fulfillment of the requirements for the degree of Master of Science. / Spring Semester, 2015. / April 7, 2015. / Algorithm, Climatology, Florida, Fog, Remote Sensing, Satellite / Includes bibliographical references. / Peter S. Ray, Professor Directing Thesis; Ming Cai, Committee Member; Zhaohua Wu, Committee Member.

Fog and smoke can combine to form dangerous zero visibility conditions along roadways throughout the state of Florida. The ability to forecast when and where fog will occur is problematic. Fog can occur over large and small scales, and is dependent on many meteorological and geographic variables. This study used Automated Weather Observation Stations (AWOS) and Automated Surface Observing Systems (ASOS) throughout the state of Florida to develop a climatology to ascertain what conditions are necessary for radiation fog development. Forecasted dewpoint depression, wind speed, cooling rates, the derived vertical hydrolapse, and other variables were shown to all affect fog formation. Using this information, a fog forecasting model was developed. The model was used to determine a three-hour binary forecast for the early morning hours, every day, at the location of the mesonet stations used. The model would predict fog if meteorological conditions preceding the forecasting time met a series of threshold levels. The goal was to make the model easy to deploy so that law enforcement can make a fast decision of whether to warn the public about potentially dangerous road conditions. The model was compared to other forecasting techniques such as the Model Output Statistics (MOS) fog product and climatology. After comparing the model to reference forecasts, it was found that the model outperformed climatology by a significant margin and was able to detect more fog events than MOS. However, the model had a higher false alarm rate and lower percent forecasts correct compared to MOS . / A Thesis submitted to the Department of Earth, Ocean, and Atmospheric Science in partial fulfillment of the requirements for the degree of Master of Science. / Summer Semester, 2014. / July 17, 2014. / Includes bibliographical references. / Peter Ray, Professor Directing Thesis; Jeffrey Chagnon, Committee Member; Robert Hart, Committee Member.

Previous studies have analyzed various atmospheric tornado parameters in a Tropical Cyclone (TC) environment. This study focuses on the evolution of these parameters through a TC landfall. The TCTOR dataset, which assigns all TC tornadoes to their respective TC, is used to group qualifying events from a pool of 1201 tornadoes during the period of 1995-2010 into eight time intervals relative to TC landfall. The environment is then analyzed using seven operationally used tornado parameters. A statistical, spatial, and sounding analysis is performed to determine how the tornadic environment evolves over time after landfall. Analysis shows that statistically significant differences in the mean value of each parameter are found between pre-landfall, post-landfall, and various time interval comparisons. Composite field charts and case studies show that the wind shear parameters at different vertical layers help explain tornado concentrations in space at different time intervals. In addition, a comparison of composite field charts is made between the larger pool of 32 TCs in the ALL composite and the 10 TCs representing the lowest tercile, with respect to the total number of tornadoes produced. This comparison shows higher magnitudes of shear parameters in the ALL composite. Combined with model derived soundings of three prolific tornado producing TCs, this study shows that the increase in shear in the lowest layer (0-1 km) is the best diagnostic tool to explain the increase in tornado occurrences at TC landfall. This finding supports prior research, which showed that low level shear maxima coincided with tornado locations. The increase in shear in the 0-3 km and 0-6 km layers at later time intervals is found to be the best diagnostic tool to explain the secondary increase in tornado occurrences after 24 hours past TC landfall. Additionally, 24 hours after TC landfall appears to be the critical time that separates weaker TC tornadoes at prior time intervals from stronger ones that resemble mid-latitude cyclone tornadoes that occur after, based on parameter values, hodograph analysis, and conceptual models. Lastly, the Significant Tornado Parameter (STP), used with discretion, is shown to work well in diagnosing tornado occurrence in some time intervals but proves to be a poor tool in others. / A Thesis submitted to the Department of Earth, Ocean, and Atmospheric Science in partial fulfillment of the requirements for the degree of Master of Science. / Fall Semester, 2014. / November 3, 2014. / hurricanes, NARR, parameters, STP, tornadoes, wind shear / Includes bibliographical references. / Peter Ray, Professor Directing Thesis; Robert Hart, Committee Member; Mark Bourassa, Committee Member.