The Earth Engine data catalog contains a wide variety of popular, curated datasets, including the world's largest online collection of Landsat scenes (> 2.0M), numerous MODIS collections, and many other raster and vector data sets. In order to facilitate multi-source analysis, the platform provides a uniform access mechanism to a variety of raster data types, independent of their bands, projection, bit-depth, and resolution.
Using a just-in-time, distributed computation model, Earth Engine can rapidly process enormous quantities of geo-spatial data. All computation is performed lazily; nothing is computed until it's required either for output or as input to another step. This model allows real-time feedback and preview during algorithm development, which supports rapid algorithm development, testing, and improvement cycle that scales seamlessly to large-scale production data processing.
Through integration with a variety of other services, Earth Engine is able to bring to bear considerable analytic and technical firepower in a transparent fashion, including: AI-based classification via integration with Google's machine learning infrastructure, publishing and distribution at Google scale through integration with the Google Maps API, Maps Engine and Google Earth.
The Earth Engine platform supports two applications programming interfaces (APIs) for accessing datasets and algorithms: a JavaScript API for building dynamic web pages and a Python API for batch mode analysis. This talk will introduce the Python API, give examples of integration with other open source geospatial libraries, and discuss potential applications of Earth Engine to the analysis of climate and weather datasets.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner