MEETING TOMORROW S EARTH OBSERVATION CHALLENGES _ WHITE PAPER

MEETING TOMORROW S EARTH OBSERVATION CHALLENGES _ WHITE PAPER

CONTENTS I. INTRODUCTION 3 II. CHALLENGES FOR TOMORROW S EARTH OBSERVATION APPLICATIONS 4 III. CONNECT, VISUALIZE, ANALYZE, ACT: LUCIAD AS A SINGLE EARTH OBSERVATION PLATFORM 4 A. Connect, visualize, analyze, act 4 B. Luciad s Next-Generation Capabilities In Analyzing Eo Data 6 C. Luciad for Space applications besides EO 10 IV. USEFUL REFERENCES 11 Statement of Copyright and Confidentiality Copyright 2016 Luciad NV All rights reserved The descriptive materials and related information in this white paper are proprietary to Luciad. This document must not be reproduced in whole or in part, or used for tendering or manufacturing purposes, except under an agreement or with the consent in writing of Luciad and then only on the condition that this notice is included in any such reproduction. The information contained in this document is subject to change without notice.

I. INTRODUCTION Earth observation (EO) comprises both the collection of the Earth s surface information as well as the analysis and presentation of the gathered data. The Earth s atmosphere, oceans, and landscapes are changing rapidly, with human activities being a major driver. Both the monitoring and the modeling of those changes are critical for governments, the private sector and citizens to make informed decisions about the global challenges our society is facing. Vital information is being gathered by EO systems based on land, at sea, in the air and in space. However, the current process of collecting, storing, analyzing and distributing this information often remains slow and cumbersome, fragmented and incomplete when classical solutions are used. As an important example, climate research is heavily based upon computer models that simulate the Earth s climate for time periods spanning several decades. Those global models discretize the entire Earth surface to a resolution that has recently gone down to a few kilometers, resulting in billions of grid cells. In that framework, millions of weather observations (including EO data) are collected and assimilated on a daily basis, while past observational data is reanalyzed and climate simulations are performed. As a result, massive amounts of data are produced. Climate simulations are performed repeatedly to consider different scenarios. Thus, Big Data and the management of that data are challenges for organizations engaged in climate research. They involve data storage, analysis and visualization tasks for various applications. Other applications of EO face similar challenges: an increasingly large number of data sources is collected and updated at a steadily higher frequency. Terabytes of heterogeneous data sources (including multispectral imagery, video from drones and space and so on) need to be handled, processed, distributed and analyzed on a daily basis. This is often referred to as the 4 Vs: Volume, Velocity, Variety and Veracity. Classical solutions cannot handle the 4 Vs. VOLUME DATA AT REST Systems produce more data than ever before and at a pace unmatched in human history (terabytes to exabytes of existing data to process) VELOCITY DATA IN MOTION Streaming data, milliseconds to seconds to respond VARIETY DATA IN MANY FORMS Data comes in many forms: structured, unstructured, text, multimedia, etc. VERACITY DATA IN DOUBT Data can come from different sources. It is hard to know which information is accurate and which is out of date. Figure 1: The four Vs that characterize Big Data. Classical solutions cannot handle the four Vs. 3

II. CHALLENGES FOR TOMORROW S EARTH OBSERVATION APPLICATIONS These are tomorrow s challenges in EO that need to be addressed: Scalability. Not only do we have more and more data, but its quality and resolutions are also increasing to centimeter-scale accuracy, and it is updated daily. Data sets range from multi-spectral to hyperspectral and multi-dimensional, such as weather data in GRIB or NetCDF format. Distributed parallel solutions are required to bring the processing to the data instead of bringing the data to the processing. We still see that classical end-user applications are neither ready, nor designed to be deployed in distributed environments such as Apache Hadoop or cloud solutions such as CloudEO for example. From 2D, to 3D, and 4D. We live in a three-dimensional world. Datasets are more and more accurate, and Lidar or radar acquisition methods allow 3D reconstruction of the world. The F-SAR system from DLR or nanosatellites such as CubeSat already provide data in 3D or 4D. Interoperability through Standards. Programs like GMES/Copernicus or initiatives like GEOSS are promoting the publication and usage of international standards to facilitate exchange of information. Luciad understands the importance of relying on standards. Not only does it build its products on standards, but it also actively participates in standardization bodies such as the OGC SWGs. Integration. Integrating different data sources (earth, weather, social network, real time and so on) in an end-user and business-oriented application is always challenging. Classical GIS solutions have difficulties in combining big EO data sets with real-time weather and location feeds. Multi-temporal Analysis. To understand the evolution of the world, it is essential to compare different data from different time periods. That data has evolved in structure, format and publication architecture. Applications have to manage the temporal variable to provide a decision tool to the end user. Existing applications often have difficulties ingesting new data sources and formats due to their closed architecture. In addition, sensor data sources are becoming dynamic. For example, streaming HD video is now sent from space by the ISS. Interactive Visual Analytics. When dealing with frequently updated and dynamic data sources, interactivity is of key importance. Classical solutions poorly maintain interactivity with big data. Luciad products handle interactivity by performing the analysis operations more efficiently. They only analyze the data present in the view, for example. In addition, the analysis operations are implemented using hardware-accelerated GPGPU algorithms (based on OpenCL, OpenGL, and WebGL). III. CONNECT, VISUALIZE, ANALYZE, ACT: LUCIAD AS A SINGLE EARTH OBSERVATION PLATFORM A. CONNECT, VISUALIZE, ANALYZE, ACT Luciad offers a complete COTS ecosystem with server, cloud, desktop, browser and mobile technology to supply integrated solutions to both current and future EO challenges. A wide variety of data connectors are able to ingest any type of data (satellite imagery in any format such as GeoTIFF or JPEG2000, base elevation and raster maps, vector data such as OpenStreetMap data, real-time radar feeds or streaming video imagery, for example in MISB 601 format with MPEG2 encoding). All that data can be accessed in any form (open or proprietary standards) and from any source, whether it comes directly from the source or from classical GIS systems, and whether it comes in the form of a file, as a database, or exposed as a web service. LuciadLightspeed is Luciad s flagship product. Thanks to its unique in-memory approach and GPU accelerated implementations, it is renowned as the fastest geospatial application in the world, able to handle vast amounts of imagery and vector data. For example, it handles point data from over 10 million static points to over half a million moving objects with a sub-second refresh rate. LuciadLightspeed is designed to maintain a 60-frames-per-second update rate even under the most demanding conditions. 4

ACT _ L UCI A D L IGH T SPE E D _ L UCI A D RIA _ L UCI A D MOBILE OTHER (non Luciad based) VISUALIZE ANALYZE _ L UCI A D L IGH T SPE E D _ L UCI A D FUSI O N CONNECT OTHER (non Luciad based) Multiple data sources, real-time feeds, sensors, simulations, OGC and other services Figure 2: Luciad product diagram LuciadLightspeed is the ideal solution in any EO tool to integrate EO data with real-time data sources. LuciadLightspeed boasts all capabilities required by EO in terms of: Platform independence A single unified API for 2D, 3D and 4D Management of imagery, vector and survey data Multi-spectral image processing, visual analysis, change detection Extreme performance through GPU accelerated algorithms Support for all reference systems and projections LuciadLightspeed is a components-based product that allows users to build next-generation client side applications, but it can also be used on Big Data platforms such as Hadoop for distributed processing. Like all Luciad products, LuciadLightspeed is fully web services enabled. LuciadLightspeed is the ideal solution in any EO tool to integrate EO data with real-time data sources. LuciadFusion is Luciad s high-performance data management and data server solution. LuciadFusion belongs at the heart of any enterprise-wide EO architecture. It is based on LuciadLightspeed s core geospatial components for high-performance processing (transforming, tiling, multi-leveling ). LuciadFusion is a server application equipped with integration components. LuciadFusion is designed to manage, fuse and serve geospatial data. LuciadFusion consists of a Data Connectivity Manager, a Data Server and a Storage Capability. EO analysts can rely on LuciadFusion to access geospatial data from any database, service, or file through an easy-to-use data management application. 5

LuciadFusion Data Connectivity Manager (DCM) The DCM can be used to prepare authoritative data and data products ready for dissemination. The data can be managed in different information products or Themes a combination of different datasets as required by each specific user group. Using the appropriate Themes, the required data is then ready to be easily served to one or more functional applications. Fusion Engine The Fusion Engine optimizes raster and vector data by bringing different data sources together in themes (as defined through the DCM). This involves operations such as tiling, multi-leveling, re-projecting, vector feature simplification, styling and feature filtering. It can also involve custom manipulations. This is key for handling the huge datasets images and pictures typical for the EO sector. Storage Capability Through efficient and advanced caching and storing mechanisms, LuciadFusion ensures fast retrieval and guarantees that the data is readily available at all times. Benchmark tests carried out against server solutions of traditional GIS vendors have revealed that LuciadFusion has data processing and loading performance behavior that is up to 400% faster. Data Server The LuciadFusion Data Server serves the data and relevant meta-data directly into clients by means of a high performance Tile Service protocol. Alternatively, applications can access the same information through standard OGC protocols. Customers select LuciadFusion for the benefits of intelligent data management, high-volume data processing and storage, and its ability to serve data to multiple applications from a central source. Additionally, LuciadFusion can now fuse, optimize and distribute multi-spectral imagery in its original format, preserving all bands and bit depths. LuciadRIA is Luciad s fully web product, following HTML5 and WebGL standards and operating without any plug-ins. Similarly to LuciadLightspeed it provides a single unified API (JavaScript) for 2D, 3D, and 4D. Web apps built with LuciadRIA allow tens or even hundreds of thousands of users to publicly access EO data directly from the browser, in 2D and 3D. LuciadRIA uses GPU hardware acceleration to obtain unprecedented performance. B. LUCIAD S NEXT-GENERATION CAPABILITIES IN ANALYZING EO DATA In this section we highlight how Luciad products handle tomorrow s EO challenges. Scalability Luciad products are designed from the ground up to be scalable and handle high-resolution, globe-covering data sets. For example, LuciadFusion tiles, multi-levels and simplifies all your vector, (multi-spectral) raster, weather and elevation data so that it can be provided to your application at the desired resolution. As another example of scalability, LuciadLightspeed offers geospatial developers a complete and refined GPU accelerated toolset for multi-spectral image processing. You can apply filters to single-level and multi-level, uniform and non-uniform or sparse image data sets. Examples are multi-level uniform images decoded from GeoTIFF or BigTIFF files and multi-level sparse image mosaics such as the ones provided by LuciadFusion. LuciadLightspeed maintains the original data and efficiently applies the image filters you composed. For example, when image operators are applied during visualization, they are only executed for the visible pixels, leading to immediate updates. As a result, users can interactively modify and tweak the operator parameters and combinations and see the results on a map in real-time, in any projection in 2-D or 3-D. By contrast, classical solutions often require you to apply the image operator to all your pixels before they are able to visualize the result. 6

_M E E TI NG TOMORROW S EART H OBS E RVATI O N C H ALL E NGE S Figure 3: GPU accelerated image processing here applying a sharpening filter (right) to a blurred image (left). Luciad products can handle Big Data sources such as Satellite AIS and provide hardware-accelerated components for interactive filtering and analysis of the data. The following images show millions of recorded Satellite AIS vessel plots in a LuciadLightspeed view. Figure 4: Visualizing millions of vessel plot positions obtained from AIS satellite data. Luciad products are built as components with a cleanly designed API. By relying on cross-platform technologies such as Java and Javascript, those components are ready to be used in Big Data platforms such as Apache Hadoop for example. Luciad customers seamlessly use the same components to build client-side applications and distributed Big Data processing algorithms. This makes Luciad technology ready to handle the scalability challenge. From 2D to 3D, and 4D Luciad products provide a single API for 2D and 3D, and are highly optimized for temporal data visualization and analysis. For example, LuciadLightspeed uses hardware-accelerated algorithms to orthorectify and drape geo-referenced video footage from drones and satellites in real-time. 7

Figure 5: LuciadLightspeed visualizing 3D data sets together with aerial imagery. Figure 6: LuciadLightspeed s hardware accelerated rendering engine can easily deal with Lidar. LuciadLightspeed offers full support for 3D data sources such as LIDAR (LAS), OBJ, Collada, and elevation data from GeoTIFF, DTED and other sources. As another example, LuciadFusion allows the fusion and distribution of multiple GRIB and NetCDF weather files, maintaining all dimensions and temporal slices and providing the client information with the data in its native format. By contrast, classical solutions need to convert the data to RGB images on the server and can only provide the processed imagery to the client. Interoperability through Standards Luciad products are built on top of open standards such as OGC WMS, WMTS, WFS, WCS, KML, GML, ISO 19115, to name a few. Luciad products also connect to de facto standards such as GeoTIFF, Erdas IMG, ESRI SHP and more. Luciad actively participates in defining these standards. For example, Luciad was a leading force behind the definition and reference implementation of OGC GeoPackage and is actively participating in Interoperability Experiments. Luciad is a member of the OGC Architecture Board. 8

Integration Luciad products integrate any data source on-the-fly. This includes elevation data such as DTED or DEM, raster data such as GeoTIFF and JPEG2000, vector data such as SHP, KML, weather data such as GRIB and NetCDF and real-time data such as radar and video feeds. LuciadLightspeed takes any data in any format and coordinate reference system and transforms the data on the fly to a common map reference in 2D or 3D. LuciadLightspeed and LuciadRIA use a proprietary draping algorithm that obtains high quality draping of any data source on any terrain. This includes other raster imagery, vector data such as streets, as well as dynamic data sources such as video. Figure 7: Real-time orthorectified video replay in LuciadLightspeed. Multi-temporal Analysis The challenge of integrating multiple temporal data sets provided in different forms and publication formats is addressed through Luciad s commitment to allow for the integration of any data source. In addition, Luciad products are designed with interfaces that can be implemented and extended. Those interfaces allow them to literally ingest any (future) data format, and result in a sustainable An important EO challenge is to integrate multiple temporal data sets in different forms and publication formats application that can be extended and adapted to future demands. A clear model-view-controller separation enables the re-use of components. For example, adding support for a new image format only requires the implementation of a new connector (model), while the existing visualization components can be re-used (view). Next to this Luciad products are designed for temporal changes and dynamic big data sources. This makes Luciad products the ideal choice for building interactive visual EO applications. 9

Interactive Visual Analytics With the growing size and update frequency of available EO data, it becomes increasingly important to provide interactive and visual analysis tools. For example, detecting variation or correlation between images becomes easier when users can compare the images interactively side-by-side and in quick succession. LuciadLightspeed offers ready-made visual change detection controllers for the Lightspeed view. They allow users to quickly compare sets of multiple layers: The swipe controller allows users to swipe between two layer sets that cover the same area. With the flicker controller, users can toggle the visibility of layer sets, swapping one layer set out for another with a single mouse click. The porthole controller allows users to peek through and compare layers. These controllers work with any data set, be it vector or imagery data sets, in 2D and 3D. Figure 8: Visually comparing two multispectral SWIR images with the swipe controller. Such interactive tools will become increasingly important in our ever-changing world to analyze dynamically updated data sources. As LuciadLightspeed lets users implement and apply image operators (such as band combinations, sharpening, and so on) on the visible pixels only, it allows users to build extremely interactive visual analysis tools for EO data. As a result, users can change the parameters, curves and so on in real-time and get an immediate update, whether users are analyzing a 100MB GeoTIFF image covering a small region, or a 100TB multi-level tile set pyramid covering the entire Earth. C. LUCIAD FOR SPACE APPLICATIONS BESIDES EO Luciad s state-of-the-art GPU accelerated technology can be used far beyond EO applications in Space such as mission planning and operational tools. Luciad products have been used for over a decade in Defense and Aviation systems where warfighters and pilots rely on Luciad for their mission critical systems that require sub-second response time and maintained data accuracy through any number of data transformations, reprojections or calculations. 10

IV. USEFUL REFERENCES Please contact Luciad for a selection of references that have implemented a Luciad-based solution for handling Earth Observation data. MORE INFORMATION For more technical information or to find out more about our other products and services, please contact us at info@luciad.com. www.luciad.com U.S. & Canada +1 202 507 5895 Europe & Rest of World +32 16 23 95 91 11