Cirrus cloud forecasting is of particular importance to various Department of Defense programs. This thesis takes a case study approach to study Air Force Weather Agency Mesoscale Model 5 (AFWA MM5) skill in forecasting cirrus clouds, which are not represented explicitly by the model (ice water mixing ratio is used as a surrogate.) Two cases are selected for study. For each case, an initial forecast time of interest is determined which serves as the beginning point for the case study. GOES data and 3-hourly MM5 data are then obtained at 3- hourly intervals to coincide with model forecast time steps between the initial time through the 30-hour forecast. A standard analysis is performed on all data to determine general atmospheric structure for each case at each 3- hourly point. Following this, the model's relative humidity with respect to ice, explicit ice water content, vertical velocity, and other fields are considered to determine if the model possesses the proper dynamical factors for cirrus formation. Finally, model coverage of ice cloud is compared to the ABL cloud mask results to determine how well the model s ice cloud forecasts verify against each 3-hourly observed ice water field taken from the GOES data. Results indicate that the MM5 underforecasts cirrus coverage, and that the 90% relative humidity field with respect to ice may be a better approximation of observed cirrus coverage than the ice water field.

The three dimensional distribution of clouds is of great interest to the Air Force, and to the aviation community in general. The Stochastic Cloud Forecast Model (SCFM) is a novel, global cloud model currently operated at the Air Force Weather Agency (AFWA) which diagnoses cloud cover statistically using a minimal set of predictors from global numerical forecasts. Currently the four predictors are pressure, temperature, vertical velocity, and relative humidity. In this thesis, 330 sets of predictors are compared in the SCFM-R, a research version of the model programmed for this thesis. There are some differences in the SCFM and the SCFM-R that yield important information. It is found that the SCFM is very sensitive to how cloud cover in the boundary layer is diagnosed. An analysis of the diagnosis method used to initialize the model revealed a bias for over-diagnosing cloud at lower levels and under-diagnosing cloud at upper levels. Also, it is recommended that AFWA consider exchanging temperature for another predictor more related to moisture, such as cloud water, and that relative humidity is included as relative humidity to the fourth power. Other recommendations include improving the method for diagnosing cloud cover in the boundary layer and improving the model initial condition.

model-derived wind gusts determined by each method are then compared to wind reports from the Storm Prediction Centerâ s severe storm reports archive and reports from observing stations. Model-derived wind gusts are then compared to the observed wind gusts for varying times of day and observed wind gust ranges. Wind gust frequency plots are examined for each wind method to determine accuracy and to characterize any patterns. The T1 method was the most accurate overall for this study, but was shown to be less sensitive to varying atmospheric conditions. The T2 method was the least accurate of the three methods during all situations. The WINDEX method performed well in most situations and was nearly as accurate as the T1 method, while WINDEX also proved to be the most sensitive of the three to varying mesoscale conditions.

In this thesis, a wavelet-based prediction method is developed for concealing packet-loss effects in wireless channels. The proposed method utilizes a wavelet decomposition algorithm in order to process the data and then applies the well known linear prediction technique to estimate one or more approximation coefficients as necessary at the lowest resolution level. The predicted sample stream is produced by using the predicted approximation coefficients and by exploiting certain sample value patterns in the detail coefficients. In order to test the effectiveness of the proposed scheme, a wireless channel based on a three-state Markov model is developed and simulated. Simulation results for transmission of image and speech packet streams over a wireless channel are reported for both the wavelet-based prediction and direct linear prediction. In all the simulations run in this work, the wavelet-based method outperformed the direct linear prediction method. / Hellenic Navy author.

This study investigates the impact of the synoptic-scale flow on the propagation and intensity of the sea breeze front at Eglin Air Force Base. The 925 mb wind direction and speed from the 12 UTC Tallahassee sounding was used to categorize 509 summer days as having an offshore, onshore, or coast parallel synoptic-scale flow regimes. Days with similar synoptic-scale flows were then composited together to create hourly surface analyses for each regime. Sea breeze frontogenesis, location and intensity were analyzed on hourly plots of temperature, winds and frontogenesis. Results indicate that the most intense sea breeze fronts formed under 3-5 1 ms- offshore, 7-9 1 ms- coast parallel easterly, and 3-5 1 ms- coast parallel westerly synoptic-scale flow while the weakest fronts formed under 0-3 1 ms- onshore and coast parallel westerly flow. The inland penetration of the sea breeze front was restricted under offshore flow but propagated through the Eglin Range Complex under onshore flow. The intensity of the sea breeze front was found to be a balance between convergence (frontogenetic) and turbulent mixing (frontolytic). Under onshore flow the sea breeze front formed late in the afternoon when convergence at the front was maximized and turbulent mixing decreased. Under offshore flow, the strongest sea breeze fronts formed early in the afternoon due to strong convergence between offshore and onshore winds and weak turbulent mixing.

The Air Force Weather Agency (AFWA) version of the Advanced Research Weather Research and Forecasting (ARW) model with a moving 4-km nested grid is examined for 10 track and intensity predictions of six western North Pacific tropical cyclones during 2005. In three of the 10 integrations, the ARW vortex tracker algorithm based on the 500-mb height minimum failed to appropriately move the nest and thus lost track of the storm vortex. For the other seven cases, the ARW track forecasts are more skillful than the AFWA MM5 forecasts and (except at 12 h) the CLIPER-type forecasts. The ARW intensity forecasts were less skillful than the MM5 and CLIPER-type forecasts at all forecast intervals, and were severely degraded by a large negative bias at the initial time. The deficiency in these intensity forecasts is shown to be related to model spin-up (lasting 12-54 h) problems caused by the lack of a bogus vortex and a cold start initialization from the interpolation of the NCEP Global Forecast System (GFS) analysis to the 12- km and 4-km grids. Thus, a more appropriate initial vortex representation will be required to improve intensity forecasts.

D.Com. (Economics) / Successful forecasting and a understanding of the forecast. study is to develop and propose a South African Airways which will the Working Budget. To do this the must be evaluated, extended and to satisfy the needs of the airline of this model for to project methodology where necessary The purpose forecasting enable it existing changed industry. Information on budgets in general, the definition and purpose thereof, as well as the use of the budget as a tool for management was readily available. Similarly information on forecasting and forecasting models could be found, but no pUblished research indicating a method or model for the forecasting of the Working Budget in South African Airways was available, nor was any evidence found of the use of a more sophisticated model other than forecasting a budget based on historical tendencies. In chapter one the purpose of the study is defined followed by a short description of the contents of each chapter. Chapter two underlines the importance of the managerial functions in a company. The level of management will determine the degree of the functions to be performed, whether it be planning, organising, directing, activating or controlling. Management tools are available to ease the task, and one of these tools are budgets, which could be regarded as the most important tool. Chapter three and four contain an expanded review of budgets and budgeting. Whilst the basic principle of budgeting remains applicable to all types and sizes of business, the methods and application could vary. A matter to be settled ab initio is the period that the budget parts : part one and part two the information can be should cover. Three time periods may be drawn : a yearly budget; five-year plan; or a long-term plan. The main categories of budgets are then detailed, namely the sales, purchasing and financial budgets. Chapter four continues with a description of historical data flow, and how the data flow will influence the model to be used. Finally the South African Airways budget system, which is divided into a Capital and Working Budget, is described.

In this study the prediction capabilities of Artificial Neural Networks and typical econometric methods are compared. This is done in the domains of Finance and Economics. Initially, the Neural Networks are shown to outperform traditional econometric models in forecasting nonlinear behaviour. The comparison is extended to indicate that the accuracy of share price forecasting is not necessarily improved when applying Neural Networks rather than traditional time series analysis. Finally, Neural Networks are used to forecast the South African inflation rates, and its performance is compared to that of vector error correcting models, which apparently outperform Artificial Neural Networks. / Prof. D.J. Marais

M.Com. (Econometrics) / The main purpose of this study is the combining of forecasts with special reference to major macroeconomic series of South Africa. The study is based on econometric principles and makes use of three macro-economic variables, forecasted with four forecasting techniques. The macroeconomic variables which have been selected are the consumer price index, consumer expenditure on durable and semi-durable products and real M3 money supply. Forecasts of these variables have been generated by applying the Box-Jenkins ARIMA technique, Holt's two parameter exponential smoothing, the regression approach and mUltiplicative decomposition. Subsequently, the results of each individual forecast are combined in order to determine if forecasting errors can be minimized. Traditionally, forecasting involves the identification and application of the best forecasting model. However, in the search for this unique model, it often happens that some important independent information contained in one of the other models, is discarded. To prevent this from happening, researchers have investigated the idea of combining forecasts. A number of researchers used the results from different techniques as inputs into the combination of forecasts. In spite of the differences in their conclusions, three basic principles have been identified in the combination of forecasts, namely: i The considered forecasts should represent the widest range of forecasting techniques possible. Inferior forecasts should be identified. Predictable errors should be modelled and incorporated into a new forecast series. Finally, a method of combining the selected forecasts needs to be chosen. The best way of selecting a m ethod is probably by experimenting to find the best fit over the historical data. Having generated individual forecasts, these are combined by considering the specifications of the three combination methods. The first combination method is the combination of forecasts via weighted averages. The use of weighted averages to combine forecasts allows consideration of the relative accuracy of the individual methods and of the covariances of forecast errors among the methods. Secondly, the combination of exponential smoothing and Box-Jenkins is considered. Past errors of each of the original forecasts are used to determine the weights to attach to the two original forecasts in forming the combined forecasts. Finally, the regression approach is used to combine individual forecasts. Granger en Ramanathan (1984) have shown that weights can be obtained by regressing actual values of the variables of interest on the individual forecasts, without including a constant and with the restriction that weights add up to one. The performance of combination relative to the individual forecasts have been tested, given that the efficiency criterion is the minimization of the mean square errors. The results of both the individual and the combined forecasting methods are acceptable. Although some of the methods prove to be more accurate than others, the conclusion can be made that reliable forecasts are generated by individual and combined forecasting methods. It is up to the researcher to decide whether he wants to use an individual or combined method since the difference, if any, in the root mean square percentage errors (RMSPE) are insignificantly small.

Researchers have spent years understanding resource utilization to improve scheduling, load balancing, and system management through short-term prediction of resource utilization. Early research focused primarily on single operating systems; later, interest shifted to distributed systems and, finally, into web services. In each case researchers sought to more effectively use available resources. Since schedulers are required to manage the execution of multiple programs every second, short-term prediction has focused on time frames ranging from fractions of a second to several minutes. The recent increase in the number of research studies about web services has occurred because of the explosive growth and reliance on these services by most businesses. As demand has moved from static to dynamic content, the load on machine resources has grown exponentially, periodically resulting in temporary loss of service. To address these short-term denial-of-service issues, researchers have tried short-term prediction to manage scheduling of service requests. What researchers have not considered is that the same methods used for single step short-term prediction can also be used for long-term prediction if a coarse granularity of samples is used. Instead of using one or more samples per second, a coarser aggregate of minutes or hours more accurately emulates the long-term patterns. This research has shown that simple moving averages and exponential moving averages as a prediction technique can be used to more accurately predict hourly, daily, and weekly seasonal patterns of resource utilization for web servers. Additionally, this research provides a foundation where using a resource prediction within a confidence interval range could be more useful to an administrator or system software than a single prediction point. When the focus shifts to a range, a set of probabilities can establish normal function within that system. For distributed systems, it will provide the ability to notify other systems when resource utilization is no longer normal before that system is unable to issue a notice of overloading. For web systems it can be used to provide a warning, permitting the instantiation of a second system to begin load balancing during unscheduled heavy loads. In both cases, the availability of the system can be improved by predicting a resource utilization level and the confidence interval within which that resource use has historically fallen.