Develops process-level methods for global and regional data: covariance modeling on spheres, nonstationary multivariate spatio-temporal inference, and data assimilation. Applications include climate and air quality evaluation, wind forecasting, and environmental contamination assessment.