No description available.

Overpressure, or abnormally low effective pressures, is hazardous in drilling operations and construction of sea-bottom facilities in deepwater environments. Estimation of the locations of overpressure can improve safety in these operations and significantly reduce overall project costs. Propagation velocities of both seismic P and S wave are sensitive to bulk elastic parameters and density of the sediments, which can be related to porosity, pore fluid content, lithology, and effective pressures. Overpressured areas can be analyzed using 4C seismic reflection data, which includes P-P and P-SV reflections. In this thesis, the effects on compressional (P) and shear (S) wave velocities are investigated to estimate the magnitude and location of excess pore pressure utilizing Eaton’s approach for pressure prediction (Eaton, 1969). Eaton’s (1969) method relates changes in pore pressure to changes in seismic P-wave velocity. The underlying assumption of this method utilizes the ratio of observed P-wave velocity obtained from areas of both normal and abnormal pressure. This velocity ratio evaluated through an empirically determined exponent is then related to the ratio of effective stress under normal and abnormal pressure conditions. Effective stress in a normal pressured condition is greater than the effective stress value in abnormally overpressured conditions. Due to an increased sensitivity of variations in effective pressure to seismic interval velocity, Ebrom et al. (2003) employ a modified Eaton equation to incorporate the S-wave velocity in pore pressure prediction. The data preparation and subsequent observations of seismic P and S wave velocity estimates in this thesis represent a preliminary analysis for pore pressure prediction. Six 2D receiver gathers in the regional dip direction are extracted from six individual ocean-bottom 4C seismic recording nodes for P-P and P-SV velocity analysis. The receiver gathers employed have minimal pre-processing procedures applied. The main processing steps applied were: water bottom mute, 2D rotation of horizontal components to SV and SH orientation, deconvolution, and frequency filtering. Most the processing was performed in Matlab with a volume of scripts designed by research scientists from the University of Texas, Bureau of Economic Geology. In this thesis, fluid pressure prediction is estimated utilizing several 4C multicomponent ocean-bottom nodes in the Atlantis Field in deepwater Gulf of Mexico. Velocity analysis is performed through a ray tracing approach utilizing P-P and P-SV registration. A modified Eaton’s Algorithm is then used for pore pressure prediction using both P and S wave velocity values. I was able to successfully observe both compressional and shear wave velocities to sediment depths of approximately 800 m below the seafloor. Using Hamilton (1972, 1976) and Eberhart-Phillips et al. (1989) regressions as background depth dependent velocity values and well-log derived background effective pressure values from deepwater Gulf of Mexico, I am able to solve for predicted effective pressure for the study area. The results show that the Atlantis subsurface study area experiences a degree of overpressure. / text

Exploiting photoplethysmography signals (PPG) for non-invasive blood pressure (BP) measurement is interesting for various reasons. First, PPG can easily be measured using fingerclip sensors. Second, camera based approaches allow to derive remote PPG (rPPG) signals similar to PPG and therefore provide the opportunity for non-invasive measurements of BP. Various methods relying on machine learning techniques have recently been published. Performances are often reported as the mean average error (MAE) on the data which is problematic. This work aims to analyze the PPG- and rPPG based BP prediction error with respect to the underlying data distribution. First, we train established neural network (NN) architectures and derive an appropriate parameterization of input segments drawn from continuous PPG signals. Second, we use this parameterization to train NNs with a larger PPG dataset and carry out a systematic evaluation of the predicted blood pressure. The analysis revealed a strong systematic increase of the prediction error towards less frequent BP values across NN architectures. Moreover, we tested different train/test set split configurations which underpin the importance of a careful subject-aware dataset assignment to prevent overly optimistic results. Third, we use transfer learning to train the NNs for rPPG based BP prediction. The resulting performances are similar to the PPG-only case. Finally, we apply different personalization techniques and retrain our NNs with subject-specific data for both the PPG-only and rPPG case. Whilst the particular technique is less important, personalization reduces the prediction errors significantly.