Consider a time series model which is stationary apart from a single shift in mean. If the time of a level shift is known, the least squares estimator of the magnitude of this level shift is a minimum variance unbiased estimator. If the time is unknown, however, this estimator is biased. Here, we first carry out extensive simulation studies to determine the relationship between the bias and three parameters of our time series model: the true magnitude of the level shift, the true time point and the autocorrelation of adjacent observations. Thereafter, we use two generalized additive models to generalize the simulation results. Finally, we examine to what extent the bias can be reduced by multiplying the least squares estimator with a shrinkage factor. Our results showed that the bias of the estimated magnitude of the level shift can be reduced when the level shift does not occur close to the beginning or end of the time series. However, it was not possible to simultaneously reduce the bias for all possible time points and magnitudes of the level shift.

Numerical models are powerful and widely used tools for environmental prediction; however, any model prediction contains errors due to imperfect model parameterizations, insufficient model resolution, numerical errors, imperfect initial and boundary conditions etc. A variety of approaches is applied to quantify, correct and minimize these errors including skill assessments, bias correction and formal data assimilation. All of these require observations and benefit from comprehensive data sets. In this thesis, two aspects related to the quantification and correction of errors in biological ocean models are addressed: (i) A new bias correction method for a biological ocean model is evaluated, and (ii) a novel approach for expanding the set of typically available phytoplankton observations is assessed. The bias correction method, referred to as frequency-dependent nudging, was proposed by Thompson et al. (Ocean Modelling, 2006, 13:109-125) and is used to nudge a model only in prescribed frequencies. A desirable feature of this method is that it can preserve high frequency variability that would be dampened with conventional nudging. The method is first applied to an idealized signal consisting of a seasonal cycle and high frequency variability. In this example, frequency-dependent nudging corrected for the imposed seasonal bias without affecting the high-frequency variability. The method is then applied to a non-linear, 1 dimensional (1D) biogeochemical ocean model. Results showed that application of frequency-dependent nudging leads to better biogeochemical estimates than conventional nudging. In order to expand the set of available phytoplankton observations, light measurements from sensors attached on grey seals where assessed to determine if they provide a useful proxy of phytoplankton biomass. A controlled experiment at Bedford Basin showed that attenuation coefficient estimates from light attenuation measurements from seal tags were found to correlate significantly with chlorophyll. On the Scotian Shelf, results of the assessment indicate that seal tags can uncover spatio-temporal patterns related to phytoplankton biomass; however, more research is needed to derive absolute biomass estimates in the region.

This thesis evaluates regional hydrology using output from the Canadian Regional Climate Model (CRCM 4.1) and examines changes in the hydrological processes over the Churchill River Basin (CRB) by employing the Variable Infiltration Capacity (VIC) hydrology model. The CRCM evaluation has been performed by combining the atmospheric and the terrestrial water budget components of the hydrological cycle. The North American Regional Reanalysis (NARR) data are used where direct observations are not available. The outcome of the evaluation reveals the potential of the CRCM for use in long-term hydrological studies. The CRCM atmospheric moisture fluxes and storage tendencies show reasonable agreement with the NARR. The long-term moisture flux over the CRB was found to be generally divergent during summer. A systematic bias is observed in the CRCM precipitation and temperature. A quantile-based mapping of the cumulative distribution function is applied for precipitation adjustments. The temperature correction only involves shifting and scaling to adjust mean and variance. The results indicate that the techniques employed for correction are useful for hydrological studies. Bias-correction is also applied to the CRCM future climate. The CRCM bias-corrected data is then used for hydrological modeling of the CRB. The VIC-simulated streamflow exhibits acceptable agreement with observations. The VIC model's internal variables such as snow and soil moisture indicate that the model is capable of simulating internal process variables adequately. The VIC-simulated snow and soil moisture shows the potential of use as an alternative dataset for hydrological studies. Streamflow along with precipitation and temperature are analyzed for trends. No statistically significant trend is observed in the daily precipitation series. Results suggest that an increase in temperature may reduce accumulation of snow during fall and winter. The flow regime may be in transition from a snowmelt dominated regime to a rainfall dominated regime. Results from future climate simulations of the A2 emission scenario indicate a projected increase of streamflow, while the snow depth and duration exhibit a decrease. Soil moisture response to future climate warming shows an overall increase with a greater likelihood of occurrences of higher soil moisture.

We review several nonparametric regression techniques and discuss their various strengths and weaknesses with an emphasis on derivative estimation and confidence band creation. We develop a generalized C(p) criterion for tuning parameter selection when interest lies in estimating one or more derivatives and the estimator is both linear in the observed responses and self-consistent. We propose a method for constructing simultaneous confidence bands for the mean response and one or more derivatives, where simultaneous now refers both to values of the covariate and to all derivatives under consideration. In addition we generalize the simultaneous confidence bands to account for heteroscedastic noise. Finally, we consider the characterization of nanoparticles and propose a method for identifying a proper subset of the covariate space that is most useful for characterization purposes.

This thesis evaluates regional hydrology using output from the Canadian Regional Climate Model (CRCM 4.1) and examines changes in the hydrological processes over the Churchill River Basin (CRB) by employing the Variable Infiltration Capacity (VIC) hydrology model. The CRCM evaluation has been performed by combining the atmospheric and the terrestrial water budget components of the hydrological cycle. The North American Regional Reanalysis (NARR) data are used where direct observations are not available. The outcome of the evaluation reveals the potential of the CRCM for use in long-term hydrological studies. The CRCM atmospheric moisture fluxes and storage tendencies show reasonable agreement with the NARR. The long-term moisture flux over the CRB was found to be generally divergent during summer. A systematic bias is observed in the CRCM precipitation and temperature. A quantile-based mapping of the cumulative distribution function is applied for precipitation adjustments. The temperature correction only involves shifting and scaling to adjust mean and variance. The results indicate that the techniques employed for correction are useful for hydrological studies. Bias-correction is also applied to the CRCM future climate. The CRCM bias-corrected data is then used for hydrological modeling of the CRB. The VIC-simulated streamflow exhibits acceptable agreement with observations. The VIC model's internal variables such as snow and soil moisture indicate that the model is capable of simulating internal process variables adequately. The VIC-simulated snow and soil moisture shows the potential of use as an alternative dataset for hydrological studies. Streamflow along with precipitation and temperature are analyzed for trends. No statistically significant trend is observed in the daily precipitation series. Results suggest that an increase in temperature may reduce accumulation of snow during fall and winter. The flow regime may be in transition from a snowmelt dominated regime to a rainfall dominated regime. Results from future climate simulations of the A2 emission scenario indicate a projected increase of streamflow, while the snow depth and duration exhibit a decrease. Soil moisture response to future climate warming shows an overall increase with a greater likelihood of occurrences of higher soil moisture.

Inductively coupled plasma mass spectrometry is a well-established technique for the measurement of isotope ratios. Double focussing mass analysers enable increased resolution to be applied to separate spectroscopic interferences, or the use of multi-collector detection techniques for high precision isotope ratio determinations. For the Central Science Laboratory (CSL), trace elements team, methods were developed for Zn and Fe isotope ratio measurements in acid digested faecal samples from a human nutritional study. For Zn, a novel high resolutionlmulticollector combination was employed; for Fe a single collector, high resolution method was used. In both cases, samples from the nutritional study known to contain the analytes in natural isotopic abundance were used to correct for the mass bias. Two independent methods for determining Zn and Fe isotope ratios were used to validate the measurement strategies. The team at CSL are also involved in the authentication of food products. Isotope ratio and elemental concentration data were used to determine the geographical origin of rice samples, and to distinguish between traditional and modem Basmati rice grown in India and Pakistan. NERC Isotope Geosciences Laboratory are primarily concerned with the achievable accuracy and precision of an isotope ratio measurement. Use of a mass bias correction expression appropriate to the ICP-MS instrument is essential for high quality isotope ratio measurements. Cd and Sn were used to study the variation of the mass bias in a double focussing ICP-MS system with time, absolute mass and mass difference. It was proposed that mass bias should be considered as a result of the change in the instrument response with mass, and not a fundamental parameter in its own right. A method for determination of the best mass bias correction model for an individual instrument, through examination of the instrument response function was developed.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18690号 / 工博第3968号 / 新制||工||1611(附属図書館) / 31623 / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 立川 康人, 教授 中北 英一, 准教授 KIM Sunmin / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第19976号 / 工博第4220号 / 新制||工||1653(附属図書館) / 33072 / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 立川 康人, 教授 堀 智晴, 准教授 KIM SUNMIN / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第20680号 / 工博第4377号 / 新制||工||1680(附属図書館) / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 平石 哲也, 教授 中北 英一, 准教授 森 信人 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Global hydrological models can fill crucial gaps for providing essential information on water resources management, flood and drought forecasting, and assessing the impacts of climate change. However, these models face several challenges that must be addressed to ensure their applicability at local scales. These challenges include effectively managing Big Data, proper communication, adoption, and achieving accuracy in their results. Achieving accuracy in global hydrological models is critical for acceptance in decision-making, but poses the most significant challenge due to the extensive amount of observed data required and the complexity of obtaining and preparing such data for model evaluation. In this study, I conducted an evaluation of the GEOGloWS ECMWF Streamflow Services (GESS) historical simulation and forecast. The evaluation revealed the presence of systematic biases inherent in global models, which restrict their accuracy and reliability for local applications. To address this limitation, I propose a bias correction methodology that uses local data and employs a quantile-mapping approach to correct the systematic biases in the GESS model. I applied this methodology to the +40 years historical simulation dataset and forecast files released between January 1, 2014, and December 31, 2019, demonstrating its effectiveness in correcting the magnitude and seasonality of simulated streamflow values. Additionally, to enhance communication and adoption of the GESS model, I developed a web application called Historical Validation Tool (HVT) that processes and visualizes observed and simulated historical stream discharge data from the GESS model, performs bias correction on the historical simulation, computes goodness-of-fit metrics, and applies forward bias correction to subsequent forecasts. This web application was customized specifically for Brazil, Colombia, Ecuador, and Peru within the framework of the NASA SERVIR Amazonia Project. HVT enables users from these countries to get adjusted GESS historical simulations and forecasts, enhancing the reliability of GESS modeling results at the local scale. The results demonstrate that the bias correction method significantly improves the accuracy of the GESS historical simulation and forecast, as evidenced by the Kling Gupta Efficiency, making it a valuable tool for hydrological studies and water resources management. Furthermore, HVT with its user-friendly graphical interface, rapid performance, and flood alert capabilities, effectively communicates the improvements in GESS historical and forecasted data.