Environmental studies are characterized by the high number of involved disciplines, which are characterized by different methods for measuring and analysing. This feature has as a consequence the production of a large amount of data, which are often heterogeneous. Such data, when jointly analyzed, may constitute a suitable device for the quantitative analysis of environmental phenomena. Complexity and uncertainty intervene in any environmental analysis: the construction of synthetic measures for environmental quality, the study of the relationship between human health and environment, study of the spatial and/or temporal evolution of environmental phenomena. In each of these domains the quantitative support for analysis consist in many, and often heterogeneous, data. For this reason complexity and uncertainty are two peculiar and joint features of the environmental studies. Statistical methods are a powerful tool in this field, since they manage uncertainty directly and propose flexible techinques for data reduction or for treating complexity through a model. The research project is constituted by four joint research issues: study of environmental indicators; methods for the integration of heterogeneous data; study of multivariate spatio-temporal models; study of the relationship between environment and health. For each issue the Research Units will contribute on the basis of their experience and the research interests of their members. As regards the first joint research issue, methods for constructing indicators for air and water quality will be proposed. Moreover, environmental sampling issues are strictly connected with the construction of environmental indicators, in particular for the evaluation of their uncertainty. The project proposes to develop methods for optimal spatial sampling and methods for the estimation of species abundance at small and large scale. As regards the integration of heterogeneous data issue, the problem of the statistical calibration of different measures of the same phenomenon will be examined, methods for the harmonisation of data collected on different spatial grids will be studied and methods for uncertainty evaluation and for the statistical calibration in deterministic model will be proposed. For multivariate and spatio-temporal models the effects of introducing a spatial component in models that do not consider that feature will be studied. Examples are non linear and non gaussian models, extreme values models, statistical models for monitoring, models for short multivariate temporal or spatial series, typical of ecological studies. Specific analyses for spatial or spatio-temporal models will be deepened, as: methods for dealing with spatial heterogeneity; models for the analysis of multivariate spatio-temporal series based on hidden Markov chains; estimation, forecasting and hypothesis testing in spatio-temporal models. About the relationships between health and environment the main topics to be developed are: the evaluation and quantification of the health risk of human populations exposed to potentially hazardous environmental pollutants; methods for health impact assessment; relationships between health and meteorological variables; the heterogeneity of the health-environment relationship in macro-areas by means of new for this area statistical techniques. The research project touches various fields of application. The study of air pollution, which is one of the most studied topics in environmental statistics, is proposed along with the analysis of other environment and ecosystems alterations.
Trattamento statistico della complessità e dell'incertezza negli studi ambientali
COCCHI, DANIELA
2004
Abstract
Environmental studies are characterized by the high number of involved disciplines, which are characterized by different methods for measuring and analysing. This feature has as a consequence the production of a large amount of data, which are often heterogeneous. Such data, when jointly analyzed, may constitute a suitable device for the quantitative analysis of environmental phenomena. Complexity and uncertainty intervene in any environmental analysis: the construction of synthetic measures for environmental quality, the study of the relationship between human health and environment, study of the spatial and/or temporal evolution of environmental phenomena. In each of these domains the quantitative support for analysis consist in many, and often heterogeneous, data. For this reason complexity and uncertainty are two peculiar and joint features of the environmental studies. Statistical methods are a powerful tool in this field, since they manage uncertainty directly and propose flexible techinques for data reduction or for treating complexity through a model. The research project is constituted by four joint research issues: study of environmental indicators; methods for the integration of heterogeneous data; study of multivariate spatio-temporal models; study of the relationship between environment and health. For each issue the Research Units will contribute on the basis of their experience and the research interests of their members. As regards the first joint research issue, methods for constructing indicators for air and water quality will be proposed. Moreover, environmental sampling issues are strictly connected with the construction of environmental indicators, in particular for the evaluation of their uncertainty. The project proposes to develop methods for optimal spatial sampling and methods for the estimation of species abundance at small and large scale. As regards the integration of heterogeneous data issue, the problem of the statistical calibration of different measures of the same phenomenon will be examined, methods for the harmonisation of data collected on different spatial grids will be studied and methods for uncertainty evaluation and for the statistical calibration in deterministic model will be proposed. For multivariate and spatio-temporal models the effects of introducing a spatial component in models that do not consider that feature will be studied. Examples are non linear and non gaussian models, extreme values models, statistical models for monitoring, models for short multivariate temporal or spatial series, typical of ecological studies. Specific analyses for spatial or spatio-temporal models will be deepened, as: methods for dealing with spatial heterogeneity; models for the analysis of multivariate spatio-temporal series based on hidden Markov chains; estimation, forecasting and hypothesis testing in spatio-temporal models. About the relationships between health and environment the main topics to be developed are: the evaluation and quantification of the health risk of human populations exposed to potentially hazardous environmental pollutants; methods for health impact assessment; relationships between health and meteorological variables; the heterogeneity of the health-environment relationship in macro-areas by means of new for this area statistical techniques. The research project touches various fields of application. The study of air pollution, which is one of the most studied topics in environmental statistics, is proposed along with the analysis of other environment and ecosystems alterations.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
