Different fields have different ways of approaching data analysis, thus exchanging data brings up a whole host of issues ranging from data file formats and terminology to data structures and interpretation of the numbers themselves.

Here we bring together data with distinctly different structures: standard multi-dimensional climate data (gridded longitude, latitude and time), Geographical Information Systems (GIS) geometries (e.g. country outlines), GIS image collections, and a database of events whoses columns include country and time.

We map all the data into a common data model which has dependent variables that are functions of multiple independent variables. Thus our software, while it reads data from diverse sources, hides the details of data access, and the user sees uniformly-structured data objects (e.g. data with associated metadata). These objects are then manipulated and viewed to analyze the data. This mapping also allows us to translate between many different file formats and data server protocols.

In our example,we create both a set of maps with the data overlaid (GIS-type analysis with the added convenience of time and disaster-type indices), and time series as a function of country and disaster-type (here overlaid, but could have done any time-series analysis).