Lifemapper comprises two primary goals: the construction and maintenance of an extensive predicted species habitat map archive, and the exposure of spatial data and analysis services based on this archive. We achieve these goals with a variety of open source software and standards.
The bulk of the project is written in Python. The database underlying the project is PostgresSQL spatially enabled with PostGIS. Spatial data analysis is done with GDAL (raster data) and OGR (vector data). Mapserver renders spatial data to our website and web services via Open Geospatial Consortium (OGC) standards, Web Mapping Service (WMS) for map images and Web Coverage Service (WCS) for raster data. Vector data is available for download as standard KML or zipped shapefiles.
The architecture of the Lifemapper project consists of three independent elements. Lifemapper implements the openModeller species niche modeling platform on a cluster of 64 Intel computer nodes with 128 processors and a museum data pipeline to build a global geospatial data archive of predicted species distributions.
The first element, openModeller, is running as a REST web service on our compute cluster. This web service creates species niche models and projects them onto environmental scenarios. The openModeller project provides a number of fundamental niche modeling algorithms as plug-ins, including GARP, Climate Space Model, Bioclimatic Envelopes, and others. Additional algorithms are planned for the future. It is currently being developed by Centro de Referência em Informação Ambiental (CRIA), Escola Politécnica da USP (Poli), and Instituto Nacional de Pesquisas Espaciais (INPE).
The compute cluster has 10 nodes and 250 processors with 40 terabytes of local storage. The cluster is built with Rocks, an open-source Linux cluster distribution. Sun Grid Engine (SGE) accepts, schedules, and manages remote execution of Lifemapper niche modeling experiments on the nodes, plus manages and schedules allocation of distributed resources.
The second element of Lifemapper is the workhorse of the project - the data pipeline. The pipeline assembles niche modeling experiments, dispatches them to the openModeller webservice, retrieves the results, and catalogs them. It is the data pipeline that builds and maintains our archive.
The third element of Lifemapper is the Spatial Data Library (SDL). This is not only an archive of all the input spatial data used in creating the habitat maps, but also a catalog of the resulting niche model maps. Data in the SDL is publicly available via REST web services for the metadata and OGC services for the spatial data. The website provides a mechanism for browsing the archive and exploring environmental data, species occurrence points, and niche model maps while web services built on the archive are targeted at researchers who would like to programmatically query, analyze, and download the data produced.
These elements make up the Lifemapper project. They can operate in tandem or independently: each element could be replaced by a comparable service or application for a similar output, or incorporated into a new application with unique objectives.