Simulations of glacial/interglacial cycles with simple box-models. The ocean CO2 source during deglaciations

Máster en Oceanografía

Meaningful insights: explainability techniques for black-box models on tabular data

Artificial Intelligence (AI) and Machine Learning (ML) are novel data analysis techniques providing very accurate prediction results. They are widely adopted in a variety of industries to improve efficiency and decision-making, but they are also being used to develop intelligent systems. Their success grounds upon complex mathematical models, whose decisions and rationale are usually difficult to comprehend for human users to the point of being dubbed as black-boxes. This is particularly relevant in sensitive and highly regulated domains. To mitigate and possibly solve this issue, the Explainable AI (XAI) field became prominent in recent years. XAI consists of models and techniques to enable understanding of the intricated patterns discovered by black-box models. In this thesis, we consider model-agnostic XAI techniques, which can be applied to Tabular data, with a particular focus on the Credit Scoring domain. Special attention is dedicated to the LIME framework, for which we propose several modifications to the vanilla algorithm, in particular: a pair of complementary Stability Indices that accurately measure LIME stability, and the OptiLIME policy which helps the practitioner finding the proper balance among explanations' stability and reliability. We subsequently put forward GLEAMS a model-agnostic surrogate interpretable model which requires to be trained only once, while providing both Local and Global explanations of the black-box model. GLEAMS produces feature attributions and what-if scenarios, from both dataset and model perspective. Eventually, we argue that synthetic data are an emerging trend in AI, being more and more used to train complex models instead of original data. To be able to explain the outcomes of such models, we must guarantee that synthetic data are reliable enough to be able to translate their explanations to real-world individuals. To this end we propose DAISYnt, a suite of tests to measure synthetic tabular data quality and privacy.

Modelling data and voice traffic over IP networks using continuous-time Markov models

Common approaches to IP-traffic modelling have featured the use of stochastic models, based on the Markov property, which can be classified into black box and white box models based on the approach used for modelling traffic. White box models, are simple to understand, transparent and have a physical meaning attributed to each of the associated parameters. To exploit this key advantage, this thesis explores the use of simple classic continuous-time Markov models based on a white box approach, to model, not only the network traffic statistics but also the source behaviour with respect to the network and application. The thesis is divided into two parts: The first part focuses on the use of simple Markov and Semi-Markov traffic models, starting from the simplest two-state model moving upwards to n-state models with Poisson and non-Poisson statistics. The thesis then introduces the convenient to use, mathematically derived, Gaussian Markov models which are used to model the measured network IP traffic statistics. As one of the most significant contributions, the thesis establishes the significance of the second-order density statistics as it reveals that, in contrast to first-order density, they carry much more unique information on traffic sources and behaviour. The thesis then exploits the use of Gaussian Markov models to model these unique features and finally shows how the use of simple classic Markov models coupled with use of second-order density statistics provides an excellent tool for capturing maximum traffic detail, which in itself is the essence of good traffic modelling. The second part of the thesis, studies the ON-OFF characteristics of VoIP traffic with reference to accurate measurements of the ON and OFF periods, made from a large multi-lingual database of over 100 hours worth of VoIP call recordings. The impact of the language, prosodic structure and speech rate of the speaker on the statistics of the ON-OFF periods is analysed and relevant conclusions are presented. Finally, an ON-OFF VoIP source model with log-normal transitions is contributed as an ideal candidate to model VoIP traffic and the results of this model are compared with those of previously published work.

Thermoaktive Bauteilsysteme - Ein neuer simulationstechnischer Berechnungsansatz

Thermoaktive Bauteilsysteme sind Bauteile, die als Teil der Raumumschließungsflächen über ein integriertes Rohrsystem mit einem Heiz- oder Kühlmedium beaufschlagt werden können und so die Beheizung oder Kühlung des Raumes ermöglichen. Die Konstruktionenvielfalt reicht nach diesem Verständnis von Heiz, bzw. Kühldecken über Geschoßtrenndecken mit kern-integrierten Rohren bis hin zu den Fußbodenheizungen. Die darin enthaltenen extrem trägen Systeme werden bewußt eingesetzt, um Energieangebot und Raumenergiebedarf unter dem Aspekt der rationellen Energieanwendung zeitlich zu entkoppeln, z. B. aktive Bauteilkühlung in der Nacht, passive Raumkühlung über das kühle Bauteil am Tage. Gebäude- und Anlagenkonzepte, die träge reagierende thermoaktive Bauteilsysteme vorsehen, setzen im kompetenten und verantwortungsvollen Planungsprozeß den Einsatz moderner Gebäudesimulationswerkzeuge voraus, um fundierte Aussagen über Behaglichkeit und Energiebedarf treffen zu können. Die thermoaktiven Bauteilsysteme werden innerhalb dieser Werkzeuge durch Berechnungskomponenten repräsentiert, die auf mathematisch-physikalischen Modellen basieren und zur Lösung des bauteilimmanenten mehrdimensionalen instationären Wärmeleitungsproblems dienen. Bisher standen hierfür zwei unterschiedliche prinzipielle Vorgehensweisen zur Lösung zur Verfügung, die der physikalischen Modellbildung entstammen und Grenzen bzgl. abbildbarer Geometrie oder Rechengeschwindigkeit setzen. Die vorliegende Arbeit dokumentiert eine neue Herangehensweise, die als experimentelle Modellbildung bezeichnet wird. Über den Weg der Systemidentifikation können aus experimentell ermittelten Datenreihen die Parameter für ein kompaktes Black-Box-Modell bestimmt werden, das das Eingangs-Ausgangsverhalten des zugehörigen beliebig aufgebauten thermoaktiven Bauteils mit hinreichender Genauigkeit widergibt. Die Meßdatenreihen lassen sich über hochgenaue Berechnungen generieren, die auf Grund ihrer Detailtreue für den unmittelbaren Einsatz in der Gebäudesimulation ungeeignet wären. Die Anwendung der Systemidentifikation auf das zweidimensionale Wärmeleitungsproblem und der Nachweis ihrer Eignung wird an Hand von sechs sehr unterschiedlichen Aufbauten thermoaktiver Bauteilsysteme durchgeführt und bestätigt sehr geringe Temperatur- und Energiebilanzfehler. Vergleiche zwischen via Systemidentifikation ermittelten Black-Box-Modellen und physikalischen Modellen für zwei Fußbodenkonstruktionen zeigen, daß erstgenannte auch als Referenz für Genauigkeitsabschätzungen herangezogen werden können. Die Praktikabilität des neuen Modellierungsansatzes wird an Fallstudien demonstriert, die Ganzjahressimulationen unter Bauteil- und Betriebsvariationen an einem exemplarischen Büroraum betreffen. Dazu erfolgt die Integration des Black-Box-Modells in das kommerzielle Gebäude- und Anlagensimulationsprogramm CARNOT. Die akzeptablen Rechenzeiten für ein Einzonen-Gebäudemodell in Verbindung mit den hohen Genauigkeiten bescheinigen die Eignung der neuen Modellierungsweise.

New results on the thermodynamical properties of the climate system

In this paper the authors exploit two equivalent formulations of the average rate of material entropy production in the climate system to propose an approximate splitting between contributions due to vertical and eminently horizontal processes. This approach is based only on 2D radiative fields at the surface and at the top of atmosphere. Using 2D fields at the top of atmosphere alone, lower bounds to the rate of material entropy production and to the intensity of the Lorenz energy cycle are derived. By introducing a measure of the efficiency of the planetary system with respect to horizontal thermodynamic processes, it is possible to gain insight into a previous intuition on the possibility of defining a baroclinic heat engine extracting work from the meridional heat flux. The approximate formula of the material entropy production is verified and used for studying the global thermodynamic properties of climate models (CMs) included in the Program for Climate Model Diagnosis and Intercomparison (PCMDI)/phase 3 of the Coupled Model Intercomparison Project (CMIP3) dataset in preindustrial climate conditions. It is found that about 90% of the material entropy production is due to vertical processes such as convection, whereas the large-scale meridional heat transport contributes to only about 10% of the total. This suggests that the traditional two-box models used for providing a minimal representation of entropy production in planetary systems are not appropriate, whereas a basic—but conceptually correct—description can be framed in terms of a four-box model. The total material entropy production is typically 55 mW m−2 K−1, with discrepancies on the order of 5%, and CMs’ baroclinic efficiencies are clustered around 0.055. The lower bounds on the intensity of the Lorenz energy cycle featured by CMs are found to be around 1.0–1.5 W m−2, which implies that the derived inequality is rather stringent. When looking at the variability and covariability of the considered thermodynamic quantities, the agreement among CMs is worse, suggesting that the description of feedbacks is more uncertain. The contributions to material entropy production from vertical and horizontal processes are positively correlated, so that no compensation mechanism seems in place. Quite consistently among CMs, the variability of the efficiency of the system is a better proxy for variability of the entropy production due to horizontal processes than that of the large-scale heat flux. The possibility of providing constraints on the 3D dynamics of the fluid envelope based only on 2D observations of radiative fluxes seems promising for the observational study of planets and for testing numerical models.

Bistable systems with stochastic noise: virtues and limits of effective one-dimensional Langevin equations

The understanding of the statistical properties and of the dynamics of multistable systems is gaining more and more importance in a vast variety of scientific fields. This is especially relevant for the investigation of the tipping points of complex systems. Sometimes, in order to understand the time series of given observables exhibiting bimodal distributions, simple one-dimensional Langevin models are fitted to reproduce the observed statistical properties, and used to investing-ate the projected dynamics of the observable. This is of great relevance for studying potential catastrophic changes in the properties of the underlying system or resonant behaviours like those related to stochastic resonance-like mechanisms. In this paper, we propose a framework for encasing this kind of studies, using simple box models of the oceanic circulation and choosing as observable the strength of the thermohaline circulation. We study the statistical properties of the transitions between the two modes of operation of the thermohaline circulation under symmetric boundary forcings and test their agreement with simplified one-dimensional phenomenological theories. We extend our analysis to include stochastic resonance-like amplification processes. We conclude that fitted one-dimensional Langevin models, when closely scrutinised, may result to be more ad-hoc than they seem, lacking robustness and/or well-posedness. They should be treated with care, more as an empiric descriptive tool than as methodology with predictive power.

Impact of oceanic processes on the carbon cycle during the last termination

During the last termination (from ~18 000 years ago to ~9000 years ago), the climate significantly warmed and the ice sheets melted. Simultaneously, atmospheric CO2 increased from ~190 ppm to ~260 ppm. Although this CO2 rise plays an important role in the deglacial warming, the reasons for its evolution are difficult to explain. Only box models have been used to run transient simulations of this carbon cycle transition, but by forcing the model with data constrained scenarios of the evolution of temperature, sea level, sea ice, NADW formation, Southern Ocean vertical mixing and biological carbon pump. More complex models (including GCMs) have investigated some of these mechanisms but they have only been used to try and explain LGM versus present day steady-state climates. In this study we use a coupled climate-carbon model of intermediate complexity to explore the role of three oceanic processes in transient simulations: the sinking of brines, stratification-dependent diffusion and iron fertilization. Carbonate compensation is accounted for in these simulations. We show that neither iron fertilization nor the sinking of brines alone can account for the evolution of CO2, and that only the combination of the sinking of brines and interactive diffusion can simultaneously simulate the increase in deep Southern Ocean δ13C. The scenario that agrees best with the data takes into account all mechanisms and favours a rapid cessation of the sinking of brines around 18 000 years ago, when the Antarctic ice sheet extent was at its maximum. In this scenario, we make the hypothesis that sea ice formation was then shifted to the open ocean where the salty water is quickly mixed with fresher water, which prevents deep sinking of salty water and therefore breaks down the deep stratification and releases carbon from the abyss. Based on this scenario, it is possible to simulate both the amplitude and timing of the long-term CO2 increase during the last termination in agreement with ice core data. The atmospheric δ13C appears to be highly sensitive to changes in the terrestrial biosphere, underlining the need to better constrain the vegetation evolution during the termination.

Part of the global DOC versus AOU (dissolved organic carbon/apparent oxygen utilization) data compilation, Pacific Ocean

Total organic carbon (TOC) was analyzed on four transects along 140°W in 1992 using a high temperature combustion/discrete injection (HTC/DI) analyzer. For two of the transects, the analyses were conducted on-board ship. Mixed-layer concentrations of organic carbon varied from about 80 µM C at either end of the transect (12°N and 12°S) to about 60 µM C at the equator. Total organic carbon concentrations decreased rapidly below the mixed-layer to about 38-40 µM C at 1000 m across the transect. Little variation was observed below this depth; deep water concentrations below 2000 m were virtually monotonic at about 36 µM C. Repeat measurements made on subsequent cruises consistently found the same concentrations at 1000 m or deeper, but substantial variations were observed in the mixed-layer and the upper water column above 400 m depth. Linear mixing models of total organic carbon versus sigmaT exhibited zones of organic carbon formation and consumption. TOC was found to be inversely correlated with apparent oxygen utilization (AOU) in the region between the mixed-layer and the oxygen minimum. In the mixed-layer, TOC concentrations varied seasonally. Part of the variations in TOC at the equator was driven by changes in the upwelling rate in response to variations in physical forcing related to an El Niño and to the passage of tropical instability waves. TOC export fluxes, calculated from simple box models, averaged 8±4 mmol C/m**2/day at the equator and also varied seasonally. These export fluxes account for 50-75% of the total carbon deficit and are consistent with other estimates and model predictions.

Seawater carbonate chemistry and net ecosystem calcification and production in a coral community Kaneohe Bay, Oahu, Hawaii, 2011

Net ecosystem calcification rates (NEC) and net photosynthesis (NP) were determined from CO2 seawater parameters on the barrier coral reef of Kaneohe Bay, Oahu, Hawaii. Autosamplers were deployed to collect samples on the barrier reef every 2 hours for six 48-hour deployments, two each in June 2008, August 2009, and January/February 2010. NEC on the Kaneohe Bay barrier reef increased throughout the day and decreased at night. Net calcification continued at low rates at night except for six time periods when net dissolution was measured. The barrier reef was generally net photosynthetic (positive NP) during the day and net respiring (negative NP) at night. NP controlled the diel cycles of the partial pressure of CO2 (pCO2) and aragonite saturation state resulting in high daytime aragonite saturation state levels when calcification rates were at their peak. However, the NEC and NP diel cycles can become decoupled for short periods of time (several hours) without affecting calcification rates. On a net daily basis, net ecosystem production (NEP) of the barrier reef was found to be sometimes net photosynthetic and sometimes net respiring and ranged from -378 to 80 mmol m-2 d-1 when calculated using simple box models. Daily NEC of the barrier reef was positive (net calcification) for all deployments and ranged from 174 to 331 mmol CaCO3 m-2 d-1. Daily NEC was strongly negatively correlated with average daily pCO2 (R2 = 0.76) which ranged from 431 to 622 µatm. Daily NEC of the Kaneohe Bay barrier reef is similar to or higher than daily NEC measured on other coral reefs even though aragonite saturation state levels (mean aragonite saturation state = 2.85) are some of the lowest measured in coral reef ecosystems. It appears that while calcification rate and ?arag are correlated within a single coral reef ecosystem, this relationship does not necessarily hold between different coral reef systems. It can be expected that ocean acidification will not affect coral reefs uniformly and that some may be more sensitive to increasing pCO2 levels than others.

Black-box decomposition approach for computational hemodynamics: One-dimensional models

Increasing efforts exist in integrating different levels of detail in models of the cardiovascular system. For instance, one-dimensional representations are employed to model the systemic circulation. In this context, effective and black-box-type decomposition strategies for one-dimensional networks are needed, so as to: (i) employ domain decomposition strategies for large systemic models (1D-1D coupling) and (ii) provide the conceptual basis for dimensionally-heterogeneous representations (1D-3D coupling, among various possibilities). The strategy proposed in this article works for both of these two scenarios, though the several applications shown to illustrate its performance focus on the 1D-1D coupling case. A one-dimensional network is decomposed in such a way that each coupling point connects two (and not more) of the sub-networks. At each of the M connection points two unknowns are defined: the flow rate and pressure. These 2M unknowns are determined by 2M equations, since each sub-network provides one (non-linear) equation per coupling point. It is shown how to build the 2M x 2M non-linear system with arbitrary and independent choice of boundary conditions for each of the sub-networks. The idea is then to solve this non-linear system until convergence, which guarantees strong coupling of the complete network. In other words, if the non-linear solver converges at each time step, the solution coincides with what would be obtained by monolithically modeling the whole network. The decomposition thus imposes no stability restriction on the choice of the time step size. Effective iterative strategies for the non-linear system that preserve the black-box character of the decomposition are then explored. Several variants of matrix-free Broyden`s and Newton-GMRES algorithms are assessed as numerical solvers by comparing their performance on sub-critical wave propagation problems which range from academic test cases to realistic cardiovascular applications. A specific variant of Broyden`s algorithm is identified and recommended on the basis of its computer cost and reliability. (C) 2010 Elsevier B.V. All rights reserved.

Comparative study of dental arch width in plaster models, photocopies and digitized images

The aim of this study was to comparatively assess dental arch width, in the canine and molar regions, by means of direct measurements from plaster models, photocopies and digitized images of the models. The sample consisted of 130 pairs of plaster models, photocopies and digitized images of the models of white patients (n = 65), both genders, with Class I and Class II Division 1 malocclusions, treated by standard Edgewise mechanics and extraction of the four first premolars. Maxillary and mandibular intercanine and intermolar widths were measured by a calibrated examiner, prior to and after orthodontic treatment, using the three modes of reproduction of the dental arches. Dispersion of the data relative to pre- and posttreatment intra-arch linear measurements (mm) was represented as box plots. The three measuring methods were compared by one-way ANOVA for repeated measurements (α = 0.05). Initial / final mean values varied as follows: 33.94 to 34.29 mm / 34.49 to 34.66 mm (maxillary intercanine width); 26.23 to 26.26 mm / 26.77 to 26.84 mm (mandibular intercanine width); 49.55 to 49.66 mm / 47.28 to 47.45 mm (maxillary intermolar width) and 43.28 to 43.41 mm / 40.29 to 40.46 mm (mandibular intermolar width). There were no statistically significant differences between mean dental arch widths estimated by the three studied methods, prior to and after orthodontic treatment. It may be concluded that photocopies and digitized images of the plaster models provided reliable reproductions of the dental arches for obtaining transversal intra-arch measurements.