Browsing by Title

Parisian cycling increased importantly in the past twenty years. Reviewing fifty years of Parisian transport planning and details of the bike-sharing programme Vélib’, I argue in the background research that municipal biking planning and the public bicycles Vélib’ can explain this development of urban biking (1997-2015). The city also has high ambitions for the biking modal share, aiming for fifteen per cent of all Parisian transport by 2020. I want to discover what the determinants of Parisian biking are, and if the latter can be modelled and predicted; thereby, I can verify if predictions match municipal objectives for 2020. I calculate correlations between the Parisian cycling index and its possible determinants with annual values on biking and other variables from 1997 to 2015 in the first part of the analysis (chapter V). This analysis shows that cycling infrastructure, Vélib’ memberships and gasoline price are the strongest positive biking determinants, while car traffic is the strongest negative determinant. In the second part of the analysis, knowing these determinants, I can find multiple linear regression models with high R-squared values (around 0,97 and 0,98) and low standard errors. The best regression model combines linear infrastructure, car traffic volume and Vélib’ memberships. The predictions in the last part of the analysis chapter reveal that in the current tendencies, the Parisian biking modal share will reach about 7 per cent by 2020, instead of the 15 per cent aimed. But I illustrate how the objective can be accomplished, by either improving drastically one of the determinants or the three of them simultaneously to reach a modal share of 15 per cent. The results and the models found appear to be more satisfactory and accurate than the ones of previous researches, presented in the literature review. The findings may be useful for public authorities and decision-makers during processes of biking planning, and it might contribute to future research in this topic.

The feasibility of quantitatively measuring ultrasound in air with a Schlieren arrangement has been demonstrated before, but previous work demonstrating calibration of the system combined with computation to yield the 3D pressure field does not exist. The present work demonstrates the feasibility of this both in theory and practice, and characterizes the setup used to gain the results. Elementary ray optical and Schlieren theory is exhibited to support the claims. Derivation of ray optical equations related to quantitative Schlieren measurements are shown step by step to help understand the basics. A numerical example based on the theoretical results is then displayed: Synthetic Schlieren images are computed for a theoretical ultrasonic field using direct numerical integration, then the ultrasonic field is recovered from the Synthetic Schlieren images using the inverse Abel transform. Accuracy of the inverse transform is evaluated in presence of synthetic noise. The Schlieren arrangement, including the optics, optomechanics, and electronics, to produce the results is explained along with the stroboscopic use of the light source to freeze ultrasound in the photographs. Postprocessing methods such as background subtraction and median and Gaussian filtering are used. The repeatability and uncertainty of the calibration is examined by performing repeated calibration while translating or rotating the calibration targets. The ultrasound fields emitted by three transducers (100 kHz, 175 kHz, and 300 kHz) when driven by 5 cycle sine bursts at 400 Vpp are measured at two different points in time. The measured 3D pressure fields measured for each transducer are shown along with a line profile near the acoustic axis. Pressure amplitudes range near 1 kPa, which is near the acoustic pressure, are seen. Nonlinearity is seen in the waveforms as expected for such high pressures. Noise estimates from the numerical example suggest that the pressure amplitudes have an uncertainty of 10% due to noise in the photographs. Calibration experiments suggest that additional uncertainty of about 2% per degree of freedom (Z, X, rotation) is to be expected unless especial care is taken. The worst-case uncertainty is estimated to be 18%. Limitations and advantages of the method are discussed. As Schlieren is a non-contacting method it is advantageous over microphone measurements, which may affect the field they are measuring. As every photograph measures the whole field, no scanning of the measurement device is required, such as with a microphone or with an LDV. Suggestions to improve the measurement setup are provided.

In this thesis we investigate the feasibility of machine learning methods for estimating the type and the weight of individual food items from images taken of customers’ plates at a buffet- style restaurant. The images were collected in collaboration with the University of Turku and Flavoria, a public lunch-line restaurant, where a camera was mounted above the cashier to automatically take a photo of the foods chosen by the customer when they went to pay. For each image, an existing system of scales at the restaurant provided the weights for each individual food item. We describe suitable model architectures and training setups for the weight estimation and food identification tasks and explain the models’ theoretical background. Furthermore we propose and compare two methods for utilizing a restaurant’s daily menu information for improving model performance in both tasks. We show that the models perform well in comparison to baseline methods and reach accuracy on par with other similar work. Additionally, as the images were captured automatically, in some of the images the food was occluded or blurry, or the image contained sensitive customer information. To address this we present computer vision techniques for preprocessing and filtering the images. We publish the dataset containing the preprocessed images along with the corresponding individual food weights for use in future research. The main results of the project have been published as a peer-reviewed article in the International Conference in Pattern Recognition Systems 2022. The article received the best paper award of the conference.

Tässä tutkielmassa tarkastellaan Cantorin joukkoa ja sen soveltamista muutamiin matemaattisiin tarkoituksiin. Cantorin joukon määrittelyssä otetaan huomioon sen monet topologiset ominaisuudet. Tarkoituksena on muotoilla Cantorin joukkoon liittyvät matemaattiset erikoisuudet mahdollisimman ymmärrettävällä tavalla matematiikkaa vain vähän opiskelleelle. Cantorin joukko on hämmentänyt matemaatikkoja sen ensimmäisestä esiintymisestä lähtien. Joukko muodostetaan vaiheittain poistamalla yksikkövälistä [0, 1] avoimia kolmanneksia. Prosessin ensimmäisessä vaiheessa välistä [0, 1] poistetaan väli (1/3, 2/3). Seuraavassa vaiheessa kahdesta välistä [0, 1/3] ja [2/3, 1] poistetaan jälleen niiden keskimmäiset kolmannekset eli välit (1/9, 2/9) ja (7/9, 8/9). Kun tätä prosessia jatketaan loputtomasti, välistä [0, 1] lopulta jäljelle jäävät pisteet muodostavat Cantorin joukon. Cantorin joukkoon kuuluvat ainakin kaikkien poistettujen välien päätepisteet. Yhtenä joukon erikoisuutena on kuitenkin se, että siihen kuuluu vielä ylinumeroituvasti ääretön määrä pisteitä, jotka eivät ole poistettujen välien päätepisteitä. Tämän seurauksena myös Cantorin joukko on siis ylinumeroituvasti ääretön. Topologiset ominaisuudet ovat Cantorin joukolla myös erikoisia. Voidaan osoittaa, että joukolla ei ole sisäpisteitä eli pisteitä, joilla olisi jokin Cantorin joukkoon kuuluva ympäristö. Lisäksi voidaan osoittaa, että jokainen Cantorin joukon piste on kasautumispiste eli piste, jonka jokaisessa ympäristössä on jokin toinen Cantorin joukon piste. Pirunporrasfunktionakin tunnetun Cantorin funktion lähtö- ja maalijoukko on yksikköväli [0, 1] eli gamma : [0, 1] -> [0, 1]. Funktion määrittely aloitetaan kuitenkin usein Cantorin joukon avulla. Nimen pirunporrasfunktio on saanut portaikkoa muistuttavasta kuvaajastaan. Vaikka gamma muistuttaa portaikkoa ja ensisilmäyksellä vaikuttaa katkonaiselta, niin se on kuitenkin jatkuva ja jopa tasaisesti jatkuva. Viimeisenä asiana tässä tutkielmassa esitetään lyhyesti Cantorin joukkoon liittyvä Lebesguen käyrä. Lebesguen käyrää sanotaan avaruuden täyttäväksi käyräksi, koska se kulkee jokaisen maalijoukkonsa, tässä tapauksessa yksikköneliön [0; 1] x [0; 1], pisteen kautta.

The large variety of shapes, functions, and conformations of proteins explains the challenges in protein identification and separation in solution. Gel electrophoresis, and more specifically SDS-PAGE, is an established technique applied to large proteins. Capillary gel electrophoresis offers the advantage of miniaturization coupled with higher speed of analysis and sample throughput. Protein analysis in silica capillaries is affected by the presence of charges on the silica surface, causing absorption, reliability and repeatability issues. Electroosmotic flow may also contribute, requiring buffer additives such as sodium dodecyl sulphate and dynamic or permanent coatings on the capillary wall. Coatings employed in capillary zone electrophoresis can be neutral or charged. They include copolymers such as poly((1-vinylpyrrolidone)-co-(2-dimethylaminoethyl methacrylate)), polyacrylamide, diazoresin as a coupling agent to form covalently bound coatings with polyvinyl alcohol, carboxyl fullerene and graphene oxide, and proprietary coatings. Capillary gel electrophoresis offers the advantages of a sieving matrix to enhance the separation of large proteins with similar charge-to-mass ratio. Dilute and semi-dilute polymer solutions with or without self-coating properties can be employed, such as polyacrylamide, polydimethyl acrylamide, and hydrophilic cellulose derivatives such as hydroxypropyl cellulose and hydroxyethyl cellulose. UV detection is enhanced by stacking techniques such as field-amplified sample stacking and the addition of sodium chloride to the sample. The focus of experimental laboratory works for this thesis was the identification of BSA, the lipoproteins HDL and LDL, and an apolipoprotein ApoB-100 using capillary gel electrophoresis. The polymer solution employed was cross-linked polyacrylamide. Instrumental factors such as run voltage, injection type, presence of sieving matrix and SDS in the BGE, sample concentration and presence of NaCl in the sample were examined. Repeatability was an issue throughout the study caused by current instability, although SDS in the BGE and addition of NaCl to the sample prior to injection had a positive effect. Stability of the BGE and the sieving matrix, together with addition of NaCl to the sample could be explored further.

The number of people belonging to a language minority in Finland is increasing and people are becoming more and more spatially mobile. This has also led to an increase in transnationals and higher rates of cross-border mobility. With new methods involving social media big data, we can map spatial mobility patterns in new ways and deepen the understanding of how people relate to space. Differences in spatial mobility can for example give us an indication of the rate of integration into society. Some claim that a more spatially mobile life is a sign of success, but can we see differences in spatial mobility between people in Finland? The three language minorities considered in this thesis are Swedish, Russian, and Estonian. The history and culture of these groups are different as well as their status in Finnish society. Swedish speakers, with a national language status, have a different role in society, but do this well integrated minority differ from the other ones spatially? By using Twitter data and looking at the spatial mobility within Finland, we see where differences occur between language groups. To understand how strong ties the language groups have with neighbouring countries, we look at cross-border mobility to Estonia, Russia, and Sweden. The results show that there are differences in the spatial mobility of language minorities in Finland. Estonian speakers most frequently visit Estonia, while at the same time they are less mobile within Finland. The variation was large for Russian speakers, with some visiting Russia often and others almost never. Swedish speakers seem to have relatively weak ties to Sweden, compared to the other language groups and have very similar spatial mobility to the majority Finnish speaking population.

Urban areas account for 70% of worldwide energy-related CO2 emissions and play a significant role in the global carbon budget. With the enhanced consumption of fossil fuel and the dramatic change in land use related to urbanization, control and mitigation of CO2 emissions in the urban area is becoming a major concern for urban dwellers and city managers. It is of great importance and demand to estimate the local CO2 emissions in urban areas to assess the effectiveness of mitigation regulation. Surface Urban Energy and Water Balance Scheme (SUEWS) incorporated with a CO2 exchange module provides an advanced method to model total urban CO2 flux and quantify the different local-scale emission sectors involving transportation, human metabolism, buildings and vegetation. Using appropriate input data such as detailed site information and meteorological condition, it can simulate the local or neighbourhood scale CO2 emissions in a specific period, or even under a future scenario. In this study, the SUEWS model is implemented in an urban region, Jätkäsaari, which is an extension of Helsinki city centre, to simulate anthropogenic and biogenic CO2 emissions in the past and future. The construction of this district started in 2009 and was planned to be completed in 2030. Therefore, this region is a good case to investigate the impacts of urban planning on urban CO2 emissions. Based on the urban surface information, meteorological data, and abundant emission parameters, a simulation in this 1650 × 1400 m area with the spatial resolution of 50 × 50 m and the time resolution of an hour was conducted with the aim to get information on the total annual CO2 emissions, and the temporal and spatial variability of CO2 fluxes from different sources and sink in 2008 and 2030. The positive CO2 fluxes indicate the CO2 sources, while the negative indicate the CO2 sinks. In both of the previous and future case, the spatial variation of net CO2 fluxes in Jätkäsaari is dominated by the distribution of traffic and human activities. From April to September, the vegetation acts as the CO2 sink with negative net ecosystem exchange. In 2008, the modelled cumulative CO2 flux is 3.0 kt CO2 year-1, consisting of 1.9 kt CO2 year-1 from metabolism, 1.9 kt CO2 year-1 from traffic, 0.5 kt CO2 year-1 from soil and vegetation respiration, as well as -1.3 kt CO2 year-1 from photosynthesis. In 2030, the total annual CO2 emissions increase to 11.1 kt CO2 year-1 because of the rising traffic volume and amount of inhabitants. Road traffic became the dominant CO2 sources, accounting for 53% of the total emissions. For the diurnal variation, in 2008, the study area remains the CO2 sources with the exception of summertime morning when the net CO2 flux is negative, while in 2030, the net CO2 flux is positive in the whole day.

Cartographic interaction, the dialogue between a human and a map, is a process enabling indispensable ways of reasoning with spatial information. Interactive maps are digital applications, increasingly often made with web technologies. Studying and crafting cartographic interaction calls for user-inclusive studies designed around interactive map use, also necessitating the assessment of the rapidly evolving technologies enabling interactive maps. This study combines the technology- and user-centric aspects of cartographic interaction. I ask what the performance bottlenecks of a web map application are, and how different web mapping libraries compare as a platform for real-time cartographic interaction. I also ask how users interact with a highly interactive map interface, and whether cartographic interaction changes the way they perceive the mapped phenomenon. Developing a web map application, a map interface to a massive dataset on spatial accessibility (the Helsinki region Travel Time Matrix), is central to this study. I answer my technology-centric questions by assessing the technological aspects of interactive maps through the development process. To answer my user-centric questions, I carry out a user survey (n=31) by combining the web map application with an online questionnaire. My results show that the geometrical complexity of data, i.e. the number and detail of geometries to render, was the main factor limiting map responsiveness. Notable differences between web mapping libraries existed in the context of dynamic real-time interaction. Survey participants preferred to use the most dynamic mode of map interaction, and perceived the mapped phenomenon differently depending on how they interacted with the map. These results illustrate the dependence between map interface capabilities and technological design choices such as data simplification and software selection. The results also support the wider call for more dynamic map interfaces, indicating that real-time cartographic interaction can be a functional approach to exploring complex data. As a whole, the results highlight the need for the ongoing study of both mapping technologies and map use in order to discover and utilize the potential of cartographic interaction.

Nowadays even medium-range (~6 days) forecasts are mostly reliable but occasionally the quality of the forecasts collapses suddenly. During a collapse or a bust, the actual forecast is worse than a ‘forecast’ made by using climatological mean values. In this study sudden collapse of predictability will be investigated by using one example case from April 2011. OpenIFS NWP model and ERA-Interim reanalysis were used as primary tools. 13 deterministic forecasts with the best available initial conditions were run but one forecast initialized on the worst day is particularly concentrated on. One ensemble forecast of five members also initialized on the worst day is also investigated in this study. Output of OpenIFS was compared to ERA-Interim. Previous studies have shown that the reasons for European forecast busts can be found in North America. Therefore, the aim of this study is to determine if the incorrect representation of convection over North America lead to a forecast bust over Europe. Besides the main goal, this study discusses how the errors originating from North American convection lead to a forecast bust in Europe 6 days later and this study will also be looking for cause of the forecast bust from initial conditions of ensemble forecast. In this case the sudden collapse of predictability in Europe is caused by NWP models predicting change of weather regime wrong. Also OpenIFS predicts formation of a blocking high over Northern Europe although there are no signs of blocking in reanalysis. In Northern America, where the source of the error is, forecast of evolution of a cluster of thunderstorms fails so also convective forcing to large scale dynamics fails. The error grows and is transported to Europe by Rossby waves. Although none of the members of the ensemble forecast was able to forecast the weather properly in Europe, so much deviation was obtained in the outcomes that comparison of the initial conditions was meaningful. The most important finding was that deeper trough over the Rocky Mountains improves the forecast in Europe. This study was able to show evidence that misrepresented convection over North America caused the forecast to fail in Europe. Moreover, this study was able to clarify how the errors caused by misrepresented convection evolved and lead to the forecast bust in Europe. The error at the beginning of the forecast in North America grows so fast that it is unlikely that it would be due to model parameterizations but the initial conditions must contain errors. These failed forecast are difficult to avoid completely but the easiest way to reduce them is to improve quality of the observations in the Rocky mountains.

A case study that studied the performance impact of a node.js component when it was refactored from monolith environment into independent service. The performance study studied the response time of the blocking part of JavaScript code in the component. The non blocking part of the code and the added network overhead from the refactoring were excluded from the performance review. Literature review didn’t show any related research that studied the performance impact of a node.js component when it was refactored from monolith into microservices. Many found studies were found that studied the response time and throughput of REST API build with node.js with comparisons to other programming languages. A study were found that related to refactoring an application from monolith into microservices. None of the found studies were directly related to the studied case. It was noted that the response time of the component improved by 46.5% when it was refactored from monolith into microservice. It is possible that when a node.js monolith application grows it starts to affect the throughput of the event loop affecting performance critical components. For the case component it was beneficial to refactor it into independent service in order to gain the 92.6ms in the mean response time.