Finanziamento UE – NextGenerationEU PRIN 2022 "Complex Environmental Data and Modeling - CoEnv" PNRR M4C2 investimento 1.1 Avviso 104/2022
Progetto The project Complex Environmental Data and Modeling (CoEnv) proposes new statistical models that favor the integration of complex forms of spatial and spatio-temporal data, to solve data analytic issues in public health and environmental and ecological monitoring and management. With a well blended team of statisticians and environmental economists, with expertise covering theory and applications, we consider some relevant open problems in environmental research, featured in 3 of the 5 Mission areas of the Horizon Europe program and also targeted by the Recovery and Resilience Plan (Partenariati Estesi, research lines 3 & 9), to accelerate the achievement of some of the 17 Sustainable Development Goals of the UN 2030 Agenda. Within this scenario, we wish to support environmental researchers and institutions responsible for environmental management, by developing advanced statistical models capable of analyzing data with increasingly complex structures, that are nowadays encountered in environmental problems. In pursuing this objective, we build a framework for data integration, to facilitate visualization and statistical analysis of
spatio-temporal data that are multivariate, multi-source, multi-temporal, multi-scale, high dimensional and observed on complex spatial domains. The complexity of the data structures here considered poses challenging problems, for which the LITERATURE IS VERY SPARSE and satisfactory statistical solutions are either NOT AVAILABLE or ONLY POORLY IMPLEMENTED in the available software. In particular, we propose new statistical approaches and original methods to face two important problems and their interaction, often encountered in environmental applications. The first problem concerns the difficulties of predicting complex forms of data having different spatial supports and scales. This will be addressed by developing innovative MULTIVARIATE SPATIO-TEMPORAL ECOLOGICAL REGRESSION MODELS FOR MULTIVARIATE DATA, FUNCTIONAL DATA AND MORE COMPLEX OBJECT DATA. The second problem concerns the modeling of multivariate data, functional data and more complex object data that feature complex interactions among the variables or the geographical units, and are observed over complex spatial domains, such as water basins with complicated conformations or river networks. This problem will be tackled by developing FUNCTIONAL GRAPHS and SPATIAL REGRESSION WITH DIFFERENTIAL REGULARIZATION. This enables us to account for sources of anisotropy and
nonstationarities, due for instance to the presence of water currents and streams. The project will be focused on three environmental relevant problems concerning the adverse effects of pollution on health, the ecological and chemical status of water bodies and bioacoustics. The different expertises of team members, across statistics and environmental economics, and their documented collaborations with environmental researchers and policy makers ensure that CoEnv will have a clear societal impact.