The aim of the study is to propose a technique for the retrieval of point sources of atmospheric trace species from concentration measurements. The inverse problem of identifying the parameters of a point source is addressed within the assimilative framework of renormalization recently proposed for the identification of distributed emissions. This theory has been extended for the point sources based on the property that these are associated with the maximum of the renormalized estimate computed from the observations. This approach along with an analytic dispersion model is used for point source identification, and the sensitivity of the samplers is described by the same model in backward mode. The proposed technique is illustrated not only with synthetic measurements but also with seven sets of observations, corresponding to convective conditions, taken from the lowwind tracer diffusion experiment conducted at the Indian Institute of Technology Delhi in 1991. The position and intensity of the source are retrieved exactly with the synthetic measurements in all the sets validating the technique. The position of the source is retrieved with an average error of 17 m, mostly along wind; its intensity is estimated within a factor 2 for all the sets of real observations. From a theoretical point of view, the link established between point and distributed sources clarifies new concepts for the exploitation of monitoring networks. In particular, the influence of the noise on the identification of a source is related to the relative visibility of the various regions described with a renormalizing weight function. The geometry of the environment modified according to the weights is interpreted as an apparent geometry. It is analogous to the apparent flatness of the starry sky in eye's view, usually considered an impression rather than a scientific fact.

A compartmental model is formulated for oxygen transport in the cerebrovascular bed of the brain. The model considers the arteriolar, capillary and venular vessels. The vascular bed is represented as a series of compartments on the basis of blood vessel diameter. The formulation takes into account such parameters as hematocrit, vascular diameter, blood viscosity, blood flow, metabolic rate, the nonlinear oxygen dissociation curve, arterial PO2, P50 (oxygen tension at 50% hemoglobin saturation with O2) and carbon monoxide concentration. The countercurrent diffusional exchange between paired arterioles and venules is incorporated into the model. The model predicts significant longitudinal PO2 gradients in the precapillary vessels. However, gradients of hemoglobin saturation with oxygen remain fairly small. The longitudinal PO2 gradients in the postcapillary vessels are found to be very small. The effect of the following variables on tissue PO2 is studied: blood flow, PO2 in the arterial blood, hematocrit, P50, concentration of carbon monoxide, metabolic rate, arterial diameter, and the number of perfused capillaries. The qualitative features of PO2 distribution in the vascular network are not altered with moderate variation of these parameters. Finally, the various types of hypoxia, namely hypoxic, anemic and carbon monoxide hypoxia, are discussed in light of the above sensitivity analysis.

No abstract

An increasing number of satellites are being launched to observe the atmospheric concentrations of a variety of trace species. They cover a wide area at once and are expected to provide more extensive information than the rare ground-based concentration measurements. The paper introduces an adjoint technique to retrieve the emissions based on a recent concept of renormalization. This technique is used with a set of synthetic column-averaged measurements for an idealized satellite corresponding to a prescribed ground-level source. The Indian region is considered with two contrast meteorological conditions in the months of January and July, corresponding to winter and monsoon season. Since it is not feasible to handle a large volume of satellite data in the inversion due to the time involved in the computation of the matrices, a preprocessing is suggested to extract the manageable data set as a representative of the whole data. Considering a limited number of observations, it is shown that the emissions are underestimated without and with the renormalization procedure. The degree of underestimation is relatively more with non-renormalized estimates. The non-renormalized estimate is degraded further by a refined resolution of the model, whereas the renormalized estimate is not altered appreciably. The preprocessing based on aggregation of data is found to retrieve the prescribed emissions up to 75% in the month of January and 90% in the month of July. The relatively computationally expensive renormalization may be avoided except in the case of partial visibility of the area of interest, due to cloud cover or a technical constraint. A simple criterion for the optimum design of a monitoring network is suggested.