Additive versus multiplicative models in ecologic regression

Much research in environmental epidemiology relies on aggregate-level information on exposure to potentially toxic substances and on relevant covariates. We compare the use of additive (linear) and multiplicative (log-linear) regression models for the analysis of such data. We illustrate how both additive and multiplicative models can be fit to aggregate-level data sets in which disease incidence is the dependent variable, and contrast these results with similar models fitted to individual-level data. We find (1) that for aggregate-level data, multiplicative models are more likely than additive models to introduce bias into the estimation of rates, an effect not found with individual-level data; and (2) that under many circumstances multiplicative models reduce the precision of the estimates, an effect also not found in individual-level models. For both additive and multiplicative models of aggregate-level data, we find that, in the presence of covariates, narrow confidence interval are obtained only when two or more antecedent factors are strongly related to the measured covariate and/or the exposure of primary substantive interest. We conclude that the equivalency of fitting additive versus multiplicative models in studies with individual-level binary data does not carry over to studies that analyze aggregate-level information. For aggregate data, we strongly recommend use of additive models. Supported by Grant #1 U19 EH000102 from the National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA.     

Eurocode 8-compatible synthetic time-series as input to dynamic analysis

Nonlinear dynamic analysis of existing or planned structures often requires the use of accelerograms that match a target design spectrum. Here, our main concern is to generate a set of motions with a good level of fit to the Eurocode 8 design spectra for France. Synthetic time series are generated by means of a non-stationary stochastic method. To calibrate the input parameters in the stochastic approach, we select a reference set of accelerograms for a Eurocode 8 type B site category from the PEER Ground-Motion Database, which are then adjusted to the target spectrum through wavelet addition. Then, we compute nonlinear seismic responses of a soil column, including pore pressure effects, and brittle and ductile structures to the stochastic time-series, the natural accelerograms and time-series generated using stationary stochastic approaches. The results of these calculations reveal considerable variability in response despite the similarities in terms of spectral acceleration.     

Dispersion of surface plumes in the Southern North Sea

Summary Continuous releases of fluorescent dye were made in the southern North Sea and the power law dependence of the plume width, represented by _y , against the diffusion time determined. The spreading was non-Fickian and could be represented in the form _y ² =B ² t whereB was a diffusion velocity of magnitude 1.4×10^–2 m/s. Such spreading was reproduced in both particle tracking and finite difference plume models by allowing the horizontal diffusivity,K _H, to depend linearly on diffusion time. The weakness of this method is that it is not clear how the diffusion parameters can be extrapolated to weather conditions that are different to those prevailing at the time of the experiments. However, comparable spreading rates were obtained by combining a Fickian diffusion model with an advective field that represented the near-surface current shears due to wind and waves. The resulting shear diffusion effect produced realistic simulation of the observed dispersion rates. An advantage of this approach is that it enables predictions to be made over a range of weather conditions provided that the wind and wave shears can be accurately parameterized.

---

Ausbreitung von Tracerwolken an der Oberfläche in der südlichen Nordsee

Zusammenfassung Fluoreszierender Farbstoff wurde in der südlichen Nordsee kontinuierlich freigesetzt und die Breite der Wolke, repräsentiert durch _y , in Abhägigkeit von der Ausbreitungszeit bestimmt. Die Ausbreitung folgte nicht dem Fickschen Ansatz und konnte in der Form _y ² =B ² t dargestellt werden, wobeiB eine Ausbreitungsgeschwindigkeit von der Größenordnung 1.4×10^–2 m/s darstellt. Eine derartige Ausbreitung wurde sowohl mit Modellen des Partikel-Tracking als auch mit Modellen der finiten Differenzen reproduziert, wobei die horizontale AusbreitungsgrößeK _H linear von der Diffusionszeit abhängig war. Die Schwäche dieser Methode ist, daß es nicht klar ist, wie die Diffusions-Parameter für andere Wetterbedingungen als zur Zeit des Experiments zu extrapolieren wären. Vergleichbare Ausbreitungsverhältnisse werden durch die Kombination eines Fickschen Diffusionsmodells mit einem advektiven Feld erreicht, welches die oberflächennahen durch Wind und Wellen verursachten Stromscherungen berücksichtigt. Der so resultierende Scher-Diffusions-Effekt erbrachte ein realistische Simulation der beobachteten Ausbreitung. Ein Vorteil dieses Vorgehens ist, daß es Vorhersagen über einen Bereich von Wetterbedingungen ermöglicht, vorausgesetzt, daß die durch Wind und Wellen verurschte Scherung zutreffend parameterisiert werden kann.

---

La dispersion de la panache de surface dans la partie sud de la Mer du Nord

Résumé A partir de déversements continus de produits fluorescents effectués dans la partie sud de la Mer du Nord, on a pu déterminer l'expression de la largeur du panache _y , par son carré en fonction du temps de diffusion. La dispersion n'était pas du type Fickian et était représentée sous la forme _y ² =B ² t aB était une vitesse de diffusion de grandeur égale 1.4×10^–2 m/s. Une telle dispersion peut être reproduite par une modélisation du panache à la fois par suivi de particules et aux différences finies en admettant que le coefficient de diffusion horizontaleK _H dépendait linéairement du temps. L'inconvénient de cette méthode est qu'il n'est pas facile de savoir comment les paramètres de diffusion peuvent être extrapolés à des conditions météorologiques différentes de celles prévalant au moment des expériences. Quoi qu'il en soit, des taux de dispersion comparables ont été obtenus en combinant un modèle de diffusion Fickian avec un champ d'advection qui représentait les cisaillements de courant de surface dus au vent et aux vagues. L'effet de diffusion du cisaillement résultant produisit une simulation réaliste des taux de dispersion observés. Un avantage de cette approche est de rendre possible des prédictions réalisées à travers un éventail de conditions météorologiques à condition que les cisaillements liés au vent et aux vagues puissent être paramétrisés avec précision.

---

     

Heat flow in the Ozark Plateau, Arkansas and Missouri: relationship to groundwater flow

Heat flow values were calculated from direct measurements of temperature and thermal conductivity at thirteen sites in the Arkansas-Missouri Ozark Plateau region. These thirteen values are augmented by 101 estimates of heat flow, based on thermal conductivity measurements and temperature gradients extrapolated from bottom-hole temperatures. The regional heat flow profile ranges from 9 mW m⁻² to over 80 mW m⁻², but at least two distinct thermal regimes have been identified. Seven new heat flow determinations are combined with three previously published values for the St. Francois Mountains (SFM), a Precambrian exposure of granitic and rhyolitic basement rocks, average 47 mW m⁻². Radioactive heat production of 76 samples of the exposed rocks in the SFM averages 2.4 μW m⁻² and a typical continental basement contribution of 14 mW m⁻² is implied. Conversely, the sedimentary rock sequence of the plateau is characterized by an anomalously low heat flow, averaging approximately 27 mW m⁻². Groundwater transmissivity values that are based on data from 153 wells in deep regional aquifers demonstrate an inverse relationship to the observed heat flow patterns. The areas of high transmissivity that correspond to areas of low total heat flux suggest that the non-conservative vertical heat flow within the Ozark sedimentary sequence can be attributed to the effects of groundwater flow.     

Elevation adjustments of paired natural levees during flooding of the Saskatchewan River

Natural levees control the exchange of water between an alluvial channel and its floodplain, but little is known about the spatial distribution and evolution of levee heights. The summer 2005 flood of the Saskatchewan River (Cumberland Marshes, east‐central Saskatchewan) inundated large areas of floodplain for up to seven weeks, forming prominent new deposits on natural levees along main‐stem channels. Measurements of flood‐deposit thickness and crest heights of 61 levee pairs show that the thickest deposits occur on the lower pre‐flood levee in 80% of the sites, though no clear relationship exists between deposit thickness and magnitude of height difference. Only 16% of the pairs displayed thicker deposits on the higher levee, half of which occurred at sites where relatively clear floodbasin waters re‐entered turbid channels during general flooding. Difference in crest elevation (ΔE) between paired levees is approximately log‐normally distributed, both before and after the flood, though with different mean values. Supplemental observations from tank experiments indicate that during near‐bankfull flows, temporally and spatially variable deposition and erosion occur on levees due to backwater effects associated with nearby channel bars and irregular rises of the channel bed forced by channel extension. During floods, preferential deposition in lows tends to even out crest heights. Copyright © 2009 John Wiley & Sons, Ltd.