Abstract: Transitions observed in the dynamical patterns of vehicular traffic, for instance, as a result of changes in traffic density, form an important class of phenomena that is sought to be explained by large-scale modeling using many interacting agents. While the dynamics of highway traffic has been the subject of intense investigation over the last few decades, there is as yet comparatively little understanding of the patterns of urban traffic. The macroscopic collective behavior of cars in the network of roads inside a city is marked by relatively high vehicular densities and the presence of signals that coordinate movement of cross-flowing traffic traveling along several directions. We have devised a novel kinetic Monte Carlo simulation approach for studying the dynamics of urban traffic congestion, allowing study of continuous-time, continuous-space traffic flow, which contrast with the dominant paradigm of cellular automata models. Well-known results of such discrete models for traffic flow in the absence of any intersections can be easily reproduced in the framework. More importantly, the behavior in the presence of an intersection where cross-flowing traffic is regulated by a signal is seen to produce novel features. The fundamental diagram of traffic flow in the presence of a signal shows a broad plateau indicating that the flow is almost independent of small variations in vehicle density for an intermediate range of densities. This is unlike the case where there are no intersections, where a sharp transition is observed between free flow behavior and jamming on changing vehicle density. The distribution of congestion times shows a power-law scaling regime over an extended range for the stochastic case when exponential-like right skewed probability distributions are used. These results are then compared with empirically observed power-law behavior in congestion time distributions for urban traffic obtained from the cities of Delhi, Bengaluru and Mumbai.

Abstract: Understanding growth patterns in complex systems of cities through modeling is an intensive branch of quantitative geography. Complex agent-based models have been recently provided promising results by multi- modeling and intensive computation for pattern discovery and calibration. However simple interaction-based extensions of seminal models of growth (such as the Gibrat model) have not yet been tested and calibrated against real datasets. We propose a spatial model of urban growth extending the Gibrat model by adding the contributions of gravity-based interactions to expected growth rates. Moments derivation for the stochastic model allows to implement a deterministic version on expectancies. Working with the Pumain-INED harmonized database for French cities (population of urban areas for 1831-1999), the 4-parameter interaction model is calibrated through intensive computation on grid, using the OpenMole software, yielding e.g. the characteristic interaction distance at different periods. We then add a second order term aimed at integrating interactions between physical transportation networks and cities, through a feedback of physical flows on traversed cities.It allows to obtain better fits and reproduce stylized facts such as hierarchy inversions and apparition of the “tunnel effect” with the development of railway network. We furthermore introduce a novel method to assess the impact of adding parameters to a simulation model on the effectively gained information, as an extension of Akaike Information Criterion to simulation models. This empirical AIC is estimated by comparing AICs for statistical models, with same parameter number, fitting best behavior space obtained by exploration. It confirms that our extension provide a gain of information on the French city system. This contribution provides a renewing insight on simple models of urban growth for system of cities, that proves to have good explicative potentialities. It also introduce a methodology to tackle the open question of quantifying overfitting in simulation models.

Abstract: In recent years, it has been shown that many seemingly unrelated natural phenomena; earthquake magnitudes, word frequency, astronomical masses and city sizes to name a few, can be asymptotically described by a small collection of empirical distributions. Of these distributions, perhaps the most prolific is Zipf’s Law. When applied to the size distribution of cities, Zipf’s Law states that the population of a city is inversely proportional to its rank and this has been shown to apply to city sizes both globally and historically. The existence of this global distribution of city sizes places a constraint on models of city growth. The most widely accepted model is proportionate random growth which constrains growth rates to be identically distributed and independent of city size. Despite proportionate random growth being the accepted mechanism behind the evolution of city sizes, there is no consensus on a model that describes the underlying stochastic processes governing city growth rates. Furthermore, it is noted that Zipf’s Law is only present in the tail of the distribution of city sizes and does not fit the distribution as a whole suggesting proportionate random growth alone is not a complete model. Here we present a model of births and deaths that is able to both reproduce Zipf’s Law in the upper tail and account for its absence in the distribution of smaller cities. We demonstrate that the observed proportionate random growth is a consequence of the interaction of these processes. The model is validated using census data on counties in the United States. Our results can be applied to other systems in which Zipf’s law arises from the interaction of underlying processes and may provide an explanation as to why this distribution occurs across such a diverse set of natural phenomena.

Abstract: Dengue fever is a multi-serotype mosquito-borne disease that is steadily increasing in incidence worldwide and sharing animal vectors with other rapidly spreading viruses like Zika virs. In an Epidemic Prediction Initiative context, a new computational method is proposed as a new approach for constructing Stochastic Generalized Linear Models (SGLM) with multiple diversely lagged input factors based on the aim to improve prediction accuracy. The proposed computational method uses mutual information (MI) to evaluate the dependencies between predictive and outcome variables at different time lags. The window with the highest MI in the time series of each predictive variable is selected as the input of a negative binomial SGLM that predicts the weekly incidence of DF. More precisely, total cases, outbreak timing and magnitude are the variables used to design the most accurate predictive model. Global Sensitivity and Uncertainty Analysis (GSUA) is applied to attribute the variability of the output to each predictive factor and their interactions. Results reflect the micro/meso ecosystem dependence on Dengue fever incidence. For instance, temperature and humidity are more important in urban settings like in San Juan; NDVI is more important in rural settings like in Iquitos, Peru. For both study sites, annual and inter-annual trends and autoregressive components are the most influential independent variables. MI allows one to construct a varied lag factor model that can both investigate the universal epidemiology of a disease and make useful and site-dependent fine-resolution predictions. Yet, the mutual information based SGLM is proposed as a powerful epidemic prediction model not just for Dengue fever but also for any other environmental dependent infectious diseases.

Abstract: Ecosystem health issues abound worldwide with environmental implications, and impact for animal and human populations. The complexity of addressing problems systemically in the policy arena on one side, and the lack of use of computational technologies for quantitative public policy on the other side have determined a worsening of ecosystem health. We propose to enhance existing adaptive management efforts with an integrated decision-analytical and environmental dynamic model that can guide the strategic selection of robust ecosystem restoration alternative plans. The model can inform the need to adjust these alternatives in the course of action based on continuously acquired monitoring information and changing stakeholder values. We demonstrate an application of enhanced adaptive management for a wetland restoration case study inspired by the Florida Everglades restoration effort. This has implication for the environment, animal and human health and embraces the sustainability paradigm quantitatively. In terms of diseases we particularly look into waterborne and water-based diseases (Zika, Dengue Chickengunya, West Nile and Yellow Fever for instance). In relation to the Everglades, we find that alternatives designed to reconstruct the pre-drainage flow may have a positive ecological and animal health impact, but may also have high operational costs and only marginally contribute to meeting other objectives such as reduction of flooding that has catastrophic human impact and morbidities in terms of deaths and infectious disease symptoms. Enhanced adaptive management allows managers to guide investment in ecosystem modeling and monitoring efforts through scenario and value of information analyses to support optimal restoration strategies in the face of uncertain and changing information. Thus, the model allows decision makers to explore the full landscape of possible scenarios before taking decisions and to dynamically design the system considering stakeholder values, economical and political constraints, ecosystem dynamics and surprises.