184 research outputs found
Robust speaker identification based on neural response in clean and noisy conditions / Md. Atiqul Islam
Full text linkSpeaker identification (SID) is a biometric technique of determining an unknown
speaker's identity using underlying information of his/her speech utterances. It is very
essential for security, crime investigation, forensic test, and telephoning. Robust SID
under noisy conditions is still a challenging topic in the field of speech processing. Most
of the acoustic-feature-based methods fail to achieve robust SID scores under noisy
conditions. However, human performance is very robust in noisy environments. The
physiologically-based computational model of the auditory nerve (AN) proposed by
Zilany and colleagues (2006), which captures almost all of the nonlinearities observed
at the level of auditory periphery, was used in this study to obtain a robust SID
performance. A neural-response-based novel feature was proposed in this study for both
text-dependent and text-independent speaker identification systems. The proposed
feature, referred to as neurogram, was computed from the output of the AN model. The
training and testing speech signals were taken from three renowned text-independent
datasets (YOHO, TIMIT, and TIDIGIT) and a text-dependent audio speech dataset
'UNIVERSITY MALAY A' to evaluate the performance of the proposed system. The
speaker modeling was done using speech signals recorded under clean environment
whereas testing was done in both clean and noisy conditions. The testing speech signals
were contaminated by adding white Gaussian noise, pink noise, and street noise with
signal-to-noise ratios (SNRs) ranging from -5 to 15 dB in steps of 5 dB.
To develop a speaker model, three standard classifiers were employed in this study such
as the Gaussian mixture model (GMM), support vector machine (SVM) and Gaussian
mixture model-Universal background model (GMM-UBM). The performance of the
proposed neural-feature-based speaker identification was compared to the results from
the traditional acoustic-feature-based methods, such a the Mel-frequency cep tra
Evaluation of climate reanalysis and space-borne precipitation products over Bangladesh
No full textThis study aims to quantify the spatial distribution of errors in two climate reanalysis (ERA5 and CFSR) and two satellite (TMPA-RT and TMPA-V7) precipitation products over Bangladesh. The datasets are assessed against ground-based rain gauge observations to capture the extreme rainfall accumulations at daily temporal scale over a 5-year period (January 2010–December 2014). The bias ratio scores indicate that CFSR and TMPA-RT seriously overestimate the rainfall values over much of the study area. Whilst TMPA-V7 performs better than the other precipitation products, all datasets lose their detection skills substantially for higher quantile thresholds (i.e. above 50th and 75th percentiles). With respect to rainfall detection metrics – probability of detection (POD) and volumetric hit index (VHI) – both ERA5 and CFSR show superior performance (in the range 0.9–1.0 for all the analysis grid boxes). All rainfall datasets are equally good in terms of false alarm ratio (FAR) and volumetric FAR (VFAR), even though the lowest values are associated with ERA5 for higher quantiles. All products demonstrate a decrease in skill to capture the amount of rainfall but show satisfactory results to detect the rainfall events when using higher quantile thresholds (i.e. rainfall above the 50th and 75th percentiles) to sample the data before computing product skill.Full Tex
Statistical comparison of satellite-retrieved precipitation products with rain gauge observations over Bangladesh
No full textIn this investigation, six satellite-derived precipitation products namely Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN), Climate Prediction Centre (CPC) Morphing Technique (CMORPH), Integrated Multi-satellite Retrievals for Global Precipitation Measurement (IMERG) final run both non gauge-calibrated (IMERG) and gauge-calibrated (IMERG-GC), and Global Satellite Mapping of Precipitation (GSMaP) for GPM both non gauge-calibrated (GSMaP) and gauge-calibrated (GSMaP-GC) are evaluated over Bangladesh, using ground-based rain gauge observations as reference over a 3 years period from January 2014 to December 2016. Nine widely used categorical and volumetric statistical matrices such as bias, probability of detection, volumetric hit index, false alarm ratio, volumetric false alarm ratio, critical success index, volumetric critical success index, miss index, and volumetric miss index are employed to exploit the performance of the precipitation products in detecting extremes above different quantile thresholds (i.e. 50%, 75%, and 90% quantiles) for various temporal window (i.e. 3 h, 6 h, 12 h, and 24 h). The bias values show that none of the satellite rainfall data sets are ideal for detecting extreme rainfall accumulations. In fact, all products lose their detection skills consistently as the extreme precipitation thresholds (50%, 75%, and 90% quantiles) increase. The results indicate that PERSIANN shows the worst performance over the study region. Overall, GSMaP-GC performs better than the other precipitation products. However, the FAR values of GSMaP are also higher over monsoon and post-monsoon months. The categorical and volumetric scores reveal that the detection skill increases remarkably for all rainfall data sets throughout the year with the increase of extreme quantile thresholds. At higher temporal accumulations, the detection capability of the products also improves considerably, and this improvement is more significant during monsoon period. The performance is relatively poor for all precipitation data sets over the cold months. In general, all six satellite precipitation products are doing well in detecting the occurrence of rainfall but are not so good in estimating the amount of rainfall.Full Tex
Evaluation of TRMM rainfall products for hydrological uses at different scales
No full textThe principal objectives of the thesis are two: firstly to study the accuracy of the satellite
rainfall products in climatologically distinctive places at different scales and secondly to find
the possibility of using satellite-rain gauge blended rainfall products for hydrological
purposes.
Three case study areas Catalunya, Bangladesh, and South Africa have been chosen for the
analysis using the satellite rainfall products (TMPA) and rain gauge records for the period
from January 2005 to December 2009. The areal pattern of rainfall has been presented using
satellite rainfall products over the case study areas. Both daily and monthly products are
showing good agreement with rain gauge records although it is highly variable with space and
seasonality. From the results, it can be shown that TRMM satellite identified the seasonal
variability of rainfall. Moreover, the mean TRMM rainfall products show same pattern as like
mean rain gauge observations in daily and monthly scale in all case study areas.Finally, a blending technique is applied (originally used for radar-rain gauge blending) to
conform satellite rainfall products to rain gauge observations. This blended product is also
tested against the rain gauge records to verify the improvement of the blended rainfall
products over the original satellite products. Results of blended rainfall products enlightens
few aspects or issues that should consider before applying blending technique including
density of rain gauge network and resolution of TRMM pixel
Development of satellite-derived precipitation products for water resources management
No full textRainfall is highly variable both spatially and temporally, which is mainly connected to large-scale atmospheric and oceanic phenomena. Hydrologic extremes (i.e. floods and droughts) are linked to the variability in rainfall across a range of spatiotemporal scales. Heavy rainfall events on short time intervals (e.g., flash floods), in particular, often
impose potential risk to life, infrastructure, and private property. Sub-hourly rainfall extremes are also recognised as important drivers of soil erosion. Severe soil erosion can occur in the arid and semiarid regions, especially after long dry periods. Soil erosion is an important risk factor for pastoral and agricultural production as well as downstream
water quality. To address the impacts of sub-hourly rainfall extremes in the design of water sensitive urban and rural infrastructure and the development of suitable land
management practices, the availability of sub-hourly rainfall data with the adequate record length and spatial coverage is vital. However, rainfall data at sub-hourly timescales are absent or limited in terms of the record length and spatial coverage in much of the world including remote areas of Australia. Given the availability of satellite-derived precipitation products (SPPs) at fine spatiotemporal resolutions and quasi-global coverage, the objectives of this thesis were to develop SPPs for use in water resources management. The products/approaches are developed in this thesis using Australia as a test location. The SPPs were developed by: (a) evaluating several SPPs over the diverse precipitation regimes of Australia, (b) investigating new methods of using SPPs to derive intensity-frequency-duration (IFD) curves for sub-hourly durations, and (c) exploring the utility of SPPs in estimating sub-hourly rainfall intensities as input to rangeland runoff models.Thesis (PhD Doctorate)Doctor of PhilosophySchool of Eng & Built EnvGriffith SciencesFull Tex
Cyanobacterial harmful algal bloom modeling in eutrophic water bodies
Full text linkMaster of ScienceDepartment of Biological & Agricultural EngineeringAleksey Y. SheshukovHarmful algal bloom (or HAB) is a global phenomenon in the rising trend of environmental concerns that impacts public health and the economy through declining water quality and toxicity. A rapid increase in cyanobacteria concentrations in water bodies is a primary cause of HABs. Enhanced eutrophication and warmer climate are considered vital driving factors for the proliferation of HAB events in the United States and worldwide. Dynamic modeling of cyanobacteria concentrations can help manage and reduce the impact of toxic blooms by better understanding the conditions for cyanobacteria growth and providing recommendations for early advisory warnings to the public for eutrophic water bodies in the agriculture dominated watersheds of the Midwest. In this study, sub-daily time series of cyanobacteria concentration and other environmental, physical-chemical variables were collected at the USGS sites in southcentral Kansas at Cheney Reservoir near the City of Wichita and in northeast Kansas at Kansas River near Wamego. Statistical analysis of the data revealed positive correlations between cyanobacteria concentration and water temperature, irradiation, phosphorus concentration, and storage volume. Correlation of dissolved oxygen depletion with cyanobacteria growth indicated an adverse impact of HABs on aquatic systems. A process-based mathematical framework for the kinetics of cyanobacteria growth was implemented at two sites considering bacteria natural growth, non-predatory loss, outflow washout, and accounting for the changes in water temperature (T), solar irradiance (I), and available nutrients (phosphorus [P] and nitrogen [N]). Four models were developed to facilitate examination of potential data limitation in sampling and continuous observations: (i) T-based, (ii) T, I-based, (iii) T, I, P- based, and (iv) complete four-factor model (T, I, P, N-based). The models were calibrated using continuous observations in 2013 - 2014 with time intervals from 2 days to 15 days (NSE = 0.41 to 0.71), and validated for 2018 (NSE = 0.56).
Simulations revealed model efficiency in short-term (one day to bi-weekly) forecasting of cyanobacteria concentration for both nutrient-rich sites. The performance of TIP-based and TIPN-based models was found acceptable for long-term forecasting in the Cheney Reservoir. Data sampling at a 15-day interval was found adequate for the forecasting of cyanobacteria growth. A stochastic modeling approach was applied to the TIPN model that converted a kinetic growth model to a modified Fokker-Planck equation for the probability density function of the cyanobacteria concentration to account for variability in influent nutrient concentrations and their impact on HABs. Several single storm event scenarios were simulated to evaluate the impact of high nutrient runoff into the lake on cyanobacteria. Stochastic model simulations showed that mechanistic modeling forecasting uncertainty increased along time propagation and higher uncertainty in initial concentrations of the cyanobacteria. The process-based mechanistic model was found to be useful for simulating future HAB events in the data-scarce eutrophic conditions, and preliminary insights into the stochastic modeling approach showed potential for future modeling direction under variable nutrient lake condition
Analysis for genetic loci controlling protoscolex development in the Echinococcus multilocularis infection using congenic mice
No full text
- …
