Maritime data are wide-ranging and originate from multiple sources, including vessel locations, administrative records, and geographic information. Extracting and managing this knowledge can be useful for monitoring fisheries, trade, illegal acts, etc. The paper presents a geospatial data processing architecture dedicated to multisource maritime data. Our aim is to provide a solution to process these versatile data in order to provide analysis and interpretation capabilities of maritime mobilities, behaviours and situations. This paper describes the processing chain, highlighting the technical challenges associated with big data issues, data analysis and visualization. The work- flow of data, the design of databases and main data processing and algorithms are presented.