Satellite remote sensing is a regular provider of raster data used in many applications such as agricultural, weather and environmental monitoring, and natural resources management. Because of their large size and inadequate database support for image manipulation, file-based storage and management is the norm; typically in a distributed environment involving several file-servers. A variety of software packages are needed to locate and process this data; these lack data management tools, seldom provide adequate time-series data processing, and are not user friendly. This makes sharing, searching and manipulating time-series images tedious and error-prone.

This paper proposes a framework for sharing time-series raster data on the Web. We present a metadata schema, identify the functional requirements and describe an architecture for Web-based discovery and processing of such data. This includes metadata search and query, image viewing, basic image analysis and automation of routine tasks such as creation of averages and comparisons with the mean for a given time period. It also facilitates management of the images by providing suitable tools, a metadatabase that extends existing standards to meet time-series needs, and separation of the static elements from the dynamic or temporal properties. We describe an implementation of the architecture at the Southern African Development Community (SADC) Regional Remote Sensing Unit (RRSU) using free and open source software (Postgres and PostGIS for spatial data, Java MapScript for the web map server, MapBuilder for the mapping client, and JSP and Java Advanced Imaging for the Web-based processing). Evaluation at the RRSU and elsewhere indicates that the system is usable with minimal assistance and saves data managers and end-users from 60% to 85% of their time compared to the tools they currently employ.