Google Code will be shut down early next year (2016). Please follow the new updates and download D-STEM from GitHub.

D-STEM (Distributed Space Time Expectation Maximization) is a statistical software for the analysis of space-time data sets and the prediction, uncertainty included, of the observed variables by means of hierarchical multivariate space-time models. D-STEM is developed in the MATLAB language and it requires MATLAB to run.

D-STEM is able to handle the following data sets:

• Univariate data sets (one variable observed at multiple spatial locations and regular time steps)
• Multivariate data sets (multiple variable observed at multiple and possibly different spatial locations and regular time steps)
• Univariate and multivariate data sets with missing data
• Multivariate data sets of point and pixel data on regular grids (the same variables observerd as both point and pixel data)
• Large data sets

D-STEM provides the following features:

• Model parameters estimation
• Model parameter uncertainty estimation
• Model log-likelihood estimation
• Model cross-validation
• Data fusion
• Space-time mapping of the observed variables
• Space-time mapping of uncertainty

D-STEM is developed and maintained by Francesco Finazzi of the University of Bergamo, Italy and it is based on the hierarchical space-time models introduced by Alessandro Fassò and Francesco Finazzi for the analysis of environmental data sets. Francesco Finazzi and Alessandro Fassò are members of the GRASPA research group on environmental statistics. Francesco Finazzi takes part in the research project StEPhI - Statistics for Environmental Phenomena and their Interactions.

See the Downloads page to download the latest version of D-STEM with demos.

D-STEM - Mapping

Example of maps produced by D-STEM.

- Nitrogen dioxide mapping using ground level data from monitoring networks

- Observed remote sensing data and reconstructed data

- Time lapse of nitrogen dioxide concentration over Europe for 2009