This dissertation examines the causes of variations in tropical cyclone precipitation (TCP) and the relationship between TCP and river discharge in Texas. The dissertation has three major objectives: 1) investigate the spatial and temporal variations of TCP in Texas from 1950 to 2009, 2) construct seasonal statistical forecast models for TCP and identify the primary factors controlling TCP in Texas, and 3) examine how TCP contributes to the extreme precipitation and river discharge in watersheds surrounding the city of Houston. An automated extraction method is developed to identify TCP from 60 years of precipitation data from Cooperative Observing Network gauges. Texas receives an average of 123.5 mm of TCP/year, which is ~13% of the state’s mean annual precipitation. September is the month with the most TCP, and it receives an average of 18.5 mm. Long-term trends (>50 years) in TCP are evident at some locations, but there are no statistically significant long-term trends in aggregated annual TCP metrics. Despite the lack of long-term trends, TCP metrics show some spectral power at periodicities of ~2-3 years, ~5-8 years, and >10 years. Areas within 400 km of the coast have higher risk of extreme daily TCP (>100 mm), but inland Texas can also occasionally experience extreme TCP. In some areas in southeastern Texas the probability of receiving >100 mm of daily TCP in any given year is ~0.30 (i.e., daily TCP exceeds 100 mm, on average, 1 out every 3 years). The best seasonal forecast models of TCP can explain >20% variance based on three or fewer predictors. ENSO is the most important control of TCP in Texas. La Niña, the major driver in all TCP models, reduces the vertical wind shear in the Caribbean and tropical Atlantic and therefore generates more precipitating storms in Texas. Maximum Potential Velocity (MPV) in the Gulf of Mexico and vorticity in the Atlantic Hurricane Development Region (MDR) are also important predictors of TCP and they can increase the R2 by ~0.2. The negative relationship between MPV and vorticity with the TCP are due to the fact that TCs with weaker wind speed and slower translation speed tend to contribute much more to both extreme and total TCP. Sea level pressure in the Gulf of Mexico, SST in the Caribbean and North Atlantic Oscillation are also identified as useful predictors in some of the models. TCP is associated with many of the annual maximum discharge events in watersheds near Houston. Urbanization can significantly increase river discharge generated by TCP. Both the annual maximum discharge and 90 percentile discharge have increased significantly in many watersheds in Houston. Although no long-term trend can be observed in the TCP and TCP-related extreme discharge, there may be an increased risk of floods from TCP because of the statistically significant increases in annual maximum discharge that have been observed. There are also increased uncertainties in flood risk because extreme precipitation, including TCP, is projected to become more variable in the future.

Tropical cyclones (TCs) are destructive, natural occurring phenomena that can cause the loss of lives, extensive structural damage, and negative economic impacts. A major hazard associated with these tropical systems is rainfall, which can result in flood conditions, contributing to the death and destruction. The role rainfall plays in the severity of the TC aftermath emphasizes the importance for models to produce reliable precipitation forecasts. Hurricane model precipitation forecasts can be improved through precipitation verification as the model weaknesses are identified. In this study, the Hurricane Analysis and Forecast System (HAFS), an experimental NOAA hurricane model, is evaluated for its skillfulness in forecasting TC precipitation. An object-based verification method is used as it is demonstrated to more accurately represent the model skill compared to traditional point-based verification methods. A 600 km search radius is implemented to capture the TC rainfall and the objects are defined by 2, 5, and 10 mm/hr rain rate thresholds. The 2 mm/hr threshold is chosen to predominantly represent stratiform precipitation, and the 5 and 10 mm/hr thresholds are used as approximate thresholds between stratiform and convective precipitation. Shape metrics such as area, closure, dispersion, and fragmentation, are calculated for the forecast and observed objects and compared using a Mann Whitney U test. The evaluation showed that model precipitation characteristics were consistent with storms that are too intense due to forecast precipitation being too central and enclosed around the TC center at the 2 mm/hr threshold, and too cohesive at the 10 mm/hr threshold. Changes in the model skill with lead time were also investigated. The model spin-up negatively impacted the model skill up to six hours at the 2 mm/hr threshold and up to three hours at the 5 mm/hr threshold, and the skill was not affected by the spin-up at the 10 mm/hr threshold. This indicates that the model took longer to realistically depict stratiform precipitation compared to convective precipitation. The model skill also worsened after 48 hours at the 2 and 10 mm/hr thresholds when the precipitation tended to be too cohesive. Future work will apply the object-based verification method to evaluate the TC precipitation forecasts of the Basin-Scale Hurricane Weather Research and Forecasting (HWRF-B) model. / Master of Science / Tropical cyclone (TC) precipitation can impose serious threats, such as flood conditions, which can result in death and severe damage. Due to these negative consequences associated with TC rainfall, it is important for affected populations to be sufficiently prepared once these TCs make landfall. Hurricane models play a large role in the preparations that are made as they predict the location and intensity of TC rainfall, which influences the peoples' choices in taking precautionary measures. Therefore, hurricane models need to be accurate, and comparing the forecast precipitation to the observed precipitation allows for areas in which the model performs poorly to be identified. Model developers can then be informed of the areas that need to be improved. In this study, the precipitation forecasts from the Hurricane Analysis and Forecast System (HAFS) model, a hurricane model that is currently under development, are evaluated. The shape and size of the forecast and observed precipitation are quantified for light, moderate, and heavy precipitation using metrics such as area, perimeter, and elongation. The values of these metrics for the forecast and observed precipitation are compared using a statistical test. The results show that the hurricane model tended to forecast storms that are too weak due to forecast precipitation being too close to the TC center, too wrapped around the TC center, and too connected. The hurricane model is also evaluated for the accuracy of its forecasts with time from model initialization. The model had a harder time representing lighter precipitation than heavier precipitation during the first 6 hours after initialization. A decrease in the accuracy of the model forecasts was also shown 48 hours after initialization due to the general degradation of model accuracy with time after initialization. Future work will evaluate the TC precipitation forecasts of another hurricane model, the Basin-Scale Hurricane Weather Research and Forecasting (HWRF-B) model.

Hurricane Matthew (2016) was a category 5 hurricane that interacted with remote terrain over northern South America in the early stages of its life cycle. Because tropical cyclone (TC) precipitation and convection are known to be crucial factors in the understanding and forecasting of TC intensity, this study investigates how this terrain impacted Hurricane Matthew's rainband structure. Remote terrain is hypothesized to play a role in the strength of TC rainband convection by modifying the thermodynamic environment such that subsiding dry air advects over an extremely moist ocean surface layer leading to increased moist static instability. To investigate this hypothesis, this study utilizes the Advanced Research Weather and Research Forecasting Model (WRF-ARW) to create a high-resolution (2-km horizontal grid spacing) control simulation (CTL) of Hurricane Matthew and a second experimental simulation with a 50% reduction of terrain height over the topography of northern South America (T50). This study focuses on a particular convective rainband positioned downstream of the terrain that displayed prolonged robust convection during the initial stages of Hurricane Matthew's life cycle. Results indicate that characteristics of this robust rainband are consistent with prior research on an inner core rainband called a principal rainband. This rainband does not display differences in intensity in the two simulations but is located closer to the TC center and more persistent in the control simulation. In the region downstream of the topography, significantly (p < 0.05) drier conditions exist in the control simulation, which is consistent with the hypothesis that downslope motion would lead to a drier air mass. TC structural changes are also apparent, with a weaker TC in the reduced topography simulation. This research emphasizes the potentially important role of terrain distant from the TC center with possible influences on TC rainband convection and warm core structure. Conclusions of this research are limited due to the small sample size of a single case study. An ensemble modeling study and additional cases are needed for a more thorough conclusion on the impact of remote terrain on TC structure. / Master of Science / Predicting the intensity of hurricanes remains a monumental challenge for hurricane forecasters. Many factors can influence the intensity of hurricanes, including the strength, frequency, and spatial distribution of hurricane rainbands (band of precipitation). The hypothesis for this study is that terrain distant from the hurricane center can alter the hurricane environment and cause more frequent and stronger rainbands to form. To assess this hypothesis, I use a weather model to simulate Hurricane Matthew (2016) while it was interacting with remote terrain over northern South America on September 30 - October 1, 2016. Then I use the same model, but with terrain height reduced by 50% over northern South America and analyze the similarities and differences in the hurricane structure and rainband patterns. The results of this study suggest that terrain did not alter the peak rain rates in the hurricane rainbands but may have caused more frequent, widespread, and prolonged precipitation. Also, differences in hurricane structure were apparent when comparing the two model simulations. The reduced terrain simulation produced a weaker hurricane, lending some evidence to support the hypothesis that terrain may have played a role in altering the hurricane structure. These results demonstrate the potential importance of distant terrain on forecasting hurricane precipitation and intensity.

This thesis contains an observational analysis of the genesis of Tropical Storm Allison (2001). Using a paradigm of tropical cyclone formation as the superposition of potential vorticity (PV) anomalies, the importance of different scales of PV merger to various aspects of Allison’s formation is discussed. While only the case of Allison is discussed in great detail, other studies have also documented PV superposition on various scales, and superposition could be important for most tropical cyclones. Preceding Allison’s genesis, PV superposition on the large scale destabilized the atmosphere and increased low-level cyclonic vorticity. This presented a more favorable environment for the formation of MCV-type PV anomalies and smaller, surface-based, meso-&#946;-scale vortices. Although these vortices eventually merged to form a more concentrated vortex with stronger surface pressure gradients, the merger happened well after landfall of Allison and no strengthening ensued. The unstable, vorticity-rich environment was also favorable for the development of even smaller, meso-&#947;-scale vortices that accompanied deep convective cells within one of Allison’s meso-&#946;-scale vortices. The observations herein suggest that the meso-&#947;- scale convective cells and vortices are the respective source of PV production and building blocks for the meso-&#946;-scale vortices. Finally, this thesis discusses issues related to the multiple vortex nature of tropical cyclone formation. For instance, the tracking of developing tropical cyclones is greatly complicated by the presence of multiple vortices. For these cases, the paradigm of a single cyclone center is inappropriate and alternative tracking methods are introduced.

The movement of tropical cyclones (TC) is studied numerically based on a two-dimensional barotropic model, using a previously developed non-oscillatory balanced scheme. The model of TC used here takes an exponential form, and its size and strength are selected to be of a middle scale. Without a background flow, TCs move in the northwest direction due to the beta effect. The amplitudes of high wavenumber modes of the asymmetric flow, that are believed to be responsible for the TC drift, are computed using Fourier analysis. The amplitude of wavenumber one and two modes are dominant, so they are indicators of beta conversion of energy. Also, the effect of the monsoon trough on the TC movement is investigated. The results show a sudden change of the TC propagation path, consistent with earlier work. These two studies correspond to previous works. Here, the effect of equatorially trapped waves such as Kelvin, Rossby, and Mixed Rossby Gravity, on the TC path is newly studied by varying the wavenumber and wave speed of the underlying waves. The effect of the waves is considered because they are believed to contribute to cyclogenesis. For studying the effect, the barotropic flow induced by these waves via momentum transport and its variation were simulated for 50 days, and some patterns are found in the change of maximum wind speed. At a given time during the simulation, a TC is injected and the effect of the background wave is analyzed. Using the wavefield of 11 cases from 10 days to 30 days, the trajectories are calculated, and their patterns appear to be stochastic. So, the patterns are identified by calculating the mean path and its spread. The trajectories of TCs are different for different time of the waves. Kelvin waves make small variations on the length and direction of the trajectory of TCs. On the contrary, Rossby waves cause a dramatic change in the TC path and yield longer trajectories. Meanwhile, TCs in MRG waves keep fairly the same direction and usually have longer traveling distance. These changes vary by wave conditions. Therefore, the three kinds of waves have different effects on the trajectories of the TC. For some peculiar cases, the movements are explained based on wavefields. / Graduate

The vertical structures of tropospheric temperature and moisture over the oceans have not been well observed to date. The Atmospheric Infrared Sounder (AIRS) aboard NASA?s Aqua satellite offers the opportunity to provide observed soundings of these variables. This thesis focuses on the validation and application of AIRS soundings in the tropical troposphere over the Atlantic Ocean, with emphasis on the Saharan Air Layer (SAL). SAL outbreaks occur every few days, producing a warm air mass that is particularly dry at the middle levels. These westward-propagating plumes inhibit convection and are thereby thought to possess a detrimental effect on African easterly waves and tropical cyclones (TCs). First, AIRS soundings are compared with concurrent Global Positioning System (GPS) dropwindsonde data released from NOAA?s Gulfstream-IV jet aircraft, for three TC cases. In SAL environments, temperature soundings from both instruments are usually consistent. Additionally, AIRS is able to capture the very dry air in the middle levels, but it generally underestimates the moisture in the boundary layer and often misses the sharp vertical moisture gradient at the SAL base (~850 hPa). In the moist tropical boundary layer, AIRS also exhibits a dry bias. Cloud cover also prevents AIRS from accurately sampling the low-level moisture. Next, total precipitable water is derived from AIRS soundings and averaged over daily, monthly and seasonal timescales. The significant monthly and interannual variability of the moisture distribution is found to be consistent with expectations. A peak in the probability density function of mixing ratio corresponding to dry air is observed in the lower-mid troposphere in early summer, consistent with the increased frequency of SAL outbreaks during this period. Finally, the relationship between dry air derived from AIRS and TC intensity is explored. As the amount of dry air increases, particularly in the southeast and northeast quadrants of the TC, the TC becomes more likely to weaken. In the presence of high wind shear or low sea surface temperature, the likelihood of weakening increases further. While these results highlight some shortcomings of the AIRS data, their importance and uniqueness are emphasized via new applications of AIRS soundings over data sparse regions.

This thesis seeks to identify locations in which errors in numerical model initial conditions may compromise skill in tropical cyclone (TC) track forecasts. Two major TCs that made landfall in 2008 are analyzed: Hurricane Ike and Typhoon Sinlaku. In order to examine the sensitivity of the TC to selected synoptic features, a vorticity perturbation technique is developed. Within a chosen radius and atmospheric depth, the vorticity is amplified or decreased, followed by a re-balancing of the fields. The following questions are proposed: (1) How does the TC track vary with respect to initial perturbations of differing amplitude, spatial scale and distance to the storm? (2) How does the evolving perturbation act to modify the synoptic environment surrounding the TC, and thereby the track? (3) Is it best to follow an objective technique to determine the sensitive areas, or is it better to use a subjective method based on fundamental synoptic reasoning? Utilizing the Weather Research and Forecasting (WRF) model, the ?control? simulation for each TC is found to replicate forecast errors evident in the operational global models. For Sinlaku, this includes a premature recurvature in the forecast. For Ike, this comprises a landfall too far south along the Texas coast due to no recurvature being forecast. The size, magnitude and location of vorticity perturbations to the control analysis are chosen subjectively. For Sinlaku, these locations include a large mid-latitude shortwave trough around 3000 km to the north-northwest, a smaller upper-level shortwave immediately to the north, a low-level monsoon trough to the west-southwest, a weak tropical storm to the northeast, and a local perturbation in the immediate environment. It is found that WRF forecasts of Sinlaku exhibit high sensitivity, with large modifications to its track arising from the perturbation of each selected targets in the synoptic environment. The greatest improvement in the track forecast occurs by weakening the vorticity associated with each of two shortwaves to the north of Sinlaku, suggesting that either or both of the shortwaves may have been initialized too strongly in the model analysis, thereby contributing to an erroneous recurvature. For Ike, the perturbation locations include a large mid-latitude shortwave trough 2500 km to its north, an upper-level cutoff low to the east-northeast, a low-level shortwave trough to the northwest, a tropical storm in the East Pacific, and a local perturbation in the immediate environment. In contrast to Sinlaku, the perturbation of synoptic targets around Ike produces less sensitivity, likely due to the fact that Ike is not in a position of imminent recurvature. The only perturbation that leads to an accurate 4-day forecast of recurvature and landfall in North Texas is the strengthening of the large mid-latitude shortwave trough, suggesting that the shortwave may have been initialized too weakly in the operational models. Finally, a comparison of targets selected objectively by the Ensemble Transform Kalman Filter (ETKF) versus the above subjectively-chosen targets suggests that while the ETKF effectively indicates similar target regions to those selected subjectively, it may be less effective in ranking the relative sensitivities of those targets. Overall, it is found that the TC track is more sensitive to perturbations of larger amplitude and spatial scale, and less so to the distance between the perturbation and the TC, and sensitivity is confined to specific regions of the flow. The perturbation methodology employed here may be used to offer suggestions of locations in which extra high-density satellite data may be assimilated.

Tropical cyclone (TC) intensity change is governed by internal dynamics (e.g. eyewall contraction, eyewall replacement cycles, interactions of the inner-core with the rainbands) and environmental conditions (e.g. vertical wind shear, moisture distribution, and surface properties). This study aims to gain a better understanding of the physical mechanisms responsible for TC intensity changes with a particular focus to those related to the vertical wind shear and surface properties by using high resolution, full physics numerical simulations. First, the effects of the vertical wind shear on a rapidly intensifying storm and its subsequent weakening are examined. Second, a fully coupled atmosphere-wave-ocean model with a sea spray parameterization is used to study the impact of sea spray on the hurricane boundary layer. The coupled model consists of three components: the high resolution, non-hydrostatic, fifth generation Pennsylvania State University-NCAR mesoscale model (MM5), the NOAA/NCEPWAVEWATCH III (WW3) ocean surface wave model, and theWHOI threedimensional upper ocean circulation model (3DPWP). Sea spray parameterizations were developed at NOAA/ESRL and modified by the author to be introduced in uncoupled and coupled simulations. The model simulations are conducted in both uncoupled and coupled modes to isolate various physical processes influencing TC intensity. The very high-resolutionMM5 simulation of Hurricane Lili (at 0.5 km grid resolution) showed a rapid intensification associated with a contracting eyewall. Changes in both the magnitude and the direction of the vertical wind shear associated with an approaching upper-tropospheric trough were responsible for the weakening of the storm before landfall. Hurricane Lili weakened in a 5-10 m/s vertical wind shear environment. The simulated storm experienced wind shear direction normal to the storm motion, which produced a strong wavenumber one rainfall asymmetry in the downshear-left quadrant of the storm. The rainfall asymmetry was confirmed by various observations from the TRMM satellite and the WSR-88D ground radar in the coastal region. The increasing vertical wind shear induced a vertical tilt of the vortex with a time lag of about 5-6 hours after the wavenumber one rainfall asymmetry was first observed in the model simulation. Other key factors controlling intensity and intensity change in tropical cyclones are the air-sea fluxes. Accurate measurement and parameterization of air-sea fluxes under hurricane conditions are challenging. Although recent studies have shown that the momentum exchange coefficient levels off at high wind speed, little is known about the high wind behavior of the exchange coefficient for enthalpy flux. One of the largest uncertainties is the potential impact of sea spray. The current sea spray parameterizations are closely tied to wind speed and tend to overestimate the mediated heat fluxes by sea spray in the hurricane boundary layer. The sea spray generation depends not only on the wind speed but also on the variable wave state. A new spray parameterization based on the surface wave energy dissipation is introduced in the coupled model. In the coupled simulations, the wave energy dissipation is used to quantify the amount of wave breaking related to the generation of sea spray. The spray parameterization coupled to the waves may be an improvement compared to sea spray parameterizations that depends on wind speed only.

This document proposes an objective technique to estimate the intensity and predict the formation of tropical cyclones using infrared satellite imagery. As the tropical cyclone develops from an unstructured cloud cluster and intensifies the cloud structures become more axisymmetric around an identified reference point or center. This methodology processes the image gradient to measure the level of symmetry of cloud structures, which characterizes the degree of cloud organization of the tropical cyclone.The center of a cloud system is calculated by projecting and accumulating parallel lines to the gradient vectors. The point where the highest number of line intersections is located pinpoints a common point where the corresponding gradients are directed. This location is used as the center of the system. Next, a procedure that characterizes the departure of the weather system structure from axisymmetry is implemented. The deviation angle of each gradient vector relative to a radial line projected from the center is calculated. The variance of the set of deviation angles enclosed by a circular area around the center describes the axisymmetry of the system, and its behavior through time depicts its dynamics. Results are presented that show the time series of the deviation angle variances is well correlated with the National Hurricane Center best-track estimates.The formation of tropical cyclones is detected by extending the deviation-angle variance technique, it is calculated using every pixel in the scene as the center of the cloud system. Low angle variances indicate structures with high levels of axisimmetry, and these values are compared to a set of thresholds to determine whether a cloud structure can be considered as a vortex. The first detection in a sequence indicates a nascent storm. It was found that 86% of the tropical cyclones during 2004 and 2005 were detected 27 h on average before the National Hurricane Center classified them as tropical storms (33kt).Finally, two procedures to locate the center of a tropical cyclone are compared to the National Hurricane Center best-track center database. Results show that both techniques provide similar accuracy, which increases as the tropical cyclone evolves.

Hurricane Florence (2018) was unique due to its slow storm motion during landfall, causing convective rainbands to produce high amounts of precipitation along the coast of North Carolina. This study focuses on the relationship between precipitation asymmetries and atmospheric stability surrounding the tropical cyclone (TC) during the landfall period of a nearly-stationary TC. Previous research with idealized hurricane simulations suggests that atmospheric stability may vary surrounding a TC during landfall, with the atmosphere destabilizing offshore and stabilizing onshore. However, this finding has not been studied using a realistic approach. Due to Hurricane Florence's slow motion, the storm was situated at the land-ocean boundary for multiple days, providing an ideal opportunity to examine the role of atmospheric stability in modifying hurricane precipitation during landfall. This study uses the Advanced Research Weather Research and Forecasting (WRF-ARW) version 3.6.1 to produce high-resolution simulations to examine the variations in precipitation and atmospheric stability surrounding Hurricane Florence. Precipitation accumulation at different temporal scales was used to determine that asymmetries existed during the landfall period. Observed and model-simulated Convective Available Potential Energy (CAPE) were used to measure stability surrounding the TC. Simulated CAPE indicates that there was a significant difference between stability right- and left-of-track. In addition to a control simulation, two experimental simulations were conducted by modifying the land surface to vary the heat and moisture exchange coefficient (HS) and hold the surface roughness (Z0) constant. By isolating the HS to be more moist or dry, the altered low-level moisture was hypothesized to cause the precipitation and convection distributions to become more symmetrical or asymmetrical, respectively. The results from the experimental simulations showed that the altered land surface affects the relative humidity from the surface to 950 mb, which has an immediate impact on stability off-shore left-of-track. Overall, the precipitation and stability asymmetries were not significantly impacted by the altered near-surface moisture, indicating other physical factors contribute to the asymmetries. The results of this study provide insight into the role of atmospheric instability in generating asymmetrical precipitation distributions in landfalling TCs, which may be particularly important in slow-moving TCs like Hurricane Florence. / Master of Science / Landfalling tropical weather systems such as hurricanes can significantly impact coastal communities due to severe flooding and damaging winds. Hurricane Florence (2018) affected coastal and inland communities in North Carolina and South Carolina when the storm produced a significant amount of precipitation over the coastal region. During landfall, the center of Hurricane Florence moved slowly parallel to the coastline, which creates a suitable time frame to isolate and study the influence of landfall on precipitation asymmetries. Precipitation asymmetry occurs when more rainfall falls on one side of the hurricane; for example, heavier precipitation tends to occur on the right side of a hurricane during the landfall period. Hurricane rainbands that are responsible for producing heavy precipitation form in areas where there is higher moisture near the surface while lighter precipitation forms in areas where there is drier air near the surface. This study focuses on the relationship between land surface moisture and spatial variations of precipitation during the hurricane landfall period by studying observations and model simulations of Hurricane Florence. The model simulation of Hurricane Florence found that more precipitation fell on the right side of the storm, indicating that there was precipitation asymmetry. In order to understand how the precipitation asymmetries form, the model simulation of Hurricane Florence was modified to create two experiments. In the first experiment, the land surface was altered to have a moister land surface, which should cause the hurricane precipitation to be more symmetrical. In the second experiment, the land surface was altered to have a drier land surface, which should cause stronger precipitation asymmetry. However, the results did not match this expectation. Instead, both experiments simulated asymmetrical precipitation with more precipitation falling on the right side of each storm during the landfall period. These results suggest that the modified land surface moisture did not have a significant impact on the formation of precipitation asymmetries. Other factors are therefore suggested to have a more dominant influence on the development of precipitation. Overall, this work can support future studies by ruling out the impact of land surface moisture on a hurricane's precipitation formation during the landfall period.