ABSTRACT
As satellite data become more and more available and popular, it is now possible not only to better understand how the environment shifts, but also to make use of this knowledge to enhance decision-making, direct policy, and promote better governance. Artificial intelligence and computer education can allow users to analyze large quantities of earth observation data quickly and efficiently. Revolutionary neural networks can be used for automating image recognition and classification tasks based on remotely sensed imagery with correct in situ observations. Earth observation data can therefore be analyzed in real time, reducing the required duration and effort of human analysts. The World Bank Group worked in recent years with partners to develop satellite measurement and tools through open source collaboration. The funding of the Big Data Innovation system and the World Bank’s Development Economics Technical Support Group were the incubators of those developments, aiming to analyze the value of street-level imagery in producing relevant information and recognize photos that could better view land plots, taking this novel data source to advance the urgent need for knowledge in the fields of food security, product markets, monitoring of the environment, and support for the production and implementation of sustainable agricultural policies. The World Bank has formed a long-standing collaboration with the European Space Agency (ESA), currently part of the Earth Sustainable Development Observation (EO4SD) initiative, in the context of a widespread use of this technology and eventually fostering information transfers to country clients. Wide geospatial data storage and management continues to be of high importance, including the optimization of numerous conventional cloud environment systems (e.g., MySQL and PostgreSQL) and modern database management systems (e.g., NoSQL, HDFS, SPARK, and HIVE). In order to leverage the spatiotemporal wide data mining system, information extraction and automation require real-time data processing and data extraction. More efficient methods of space time mining should be built to take advantage of the Cloud platforms’ elastic store and machine resources. For the prcoessing of large geospatial data using spatiotemporal methodologies are crucial. Processed data could be further developed and formalized for the optimization of cloud computing. Further work is required to recognize and prevent attacks on the cloud platform to monitor and maintain development policy.
