# landsat8.earth 2D/3D WebGL-powered Landsat 8 satellite imagery analysis [](https://landsat8.earth) 3D, False-color Infrared Landsat visualization of the Eastern Sierra ## Overview Landsat8.earth is an application to explore [Landsat 8][landsat8] imagery in the browser using [WebGL][webgl] (via [deck.gl][deck.gl]). The backend runs on AWS Lambda and is powered by [Cloud-Optimized GeoTIFFs][cogeo], a fast cloud-native file format that enables streaming portions of raster images. It draws from the huge repository of Landsat 8 data [publicly available on S3][landsat8-aws], thanks to the AWS Open Data program. - [`landsat-mosaic-tiler`][landsat-mosaic-tiler] is the backend; a Python application running on AWS Lambda, and serves Landsat imagery on demand. - [`landsat-cogeo-mosaic`][landsat-cogeo-mosaic] is used to pregenerate mosaics (collections) of Landsat images, enabling a visualization of many images at once. - [`landsat-mosaic-latest`][landsat-mosaic-latest] keeps an updated Landsat mosaic, allowing for the option of viewing the latest available imagery. - [`deck.gl-raster`][deck.gl-raster] handles the analysis of Landsat data in WebGL. [landsat8]: https://www.usgs.gov/land-resources/nli/landsat/landsat-8?qt-science_support_page_related_con=0#qt-science_support_page_related_con [webgl]: https://developer.mozilla.org/en-US/docs/Web/API/WebGL_API [deck.gl]: https://deck.gl [cogeo]: https://cogeo.org [landsat8-aws]: https://registry.opendata.aws/landsat-8/ [landsat-mosaic-tiler]: https://github.com/kylebarron/landsat-mosaic-tiler [landsat-cogeo-mosaic]: https://github.com/kylebarron/landsat-cogeo-mosaic [landsat-mosaic-latest]: https://github.com/kylebarron/landsat-mosaic-latest [deck.gl-raster]: https://github.com/kylebarron/deck.gl-raster/ ### Credits