The effective provision of geodata collections and products plays a central role in scientific projects, public agencies and commercial enterprises. The same requirements are always in the foreground.

Existing data sources must be organized and connected.
Basic operations for selecting and preparing relevant data extracts must be provided.
An intuitive and interactive graphical representation of the data and operations should be easily provided to a wide range of users.

The Geo-Engine platform dramatically simplifies geospatial data portal creation. Our data connectors already tie in most data sources and can be easily extended to include new protocols. Based on our UI component library, portals can be realized with little front-end development effort. For processing the data e.g. for spatial filtering, temporal aggregation or blending of multiple sources the full functionality of the engine can be accessed.

Processing geodata in analyses is prohibitively complex. The selection and procurement of the right data is already costly. The preparation and processing of the data then costs 80% of the total effort, as studies have shown. This means that only 20% of the time can be spent on the actual analysis work.

With Geo Engine, we aim to reverse this ratio so that data scientists can focus on the actual analyses. For this purpose, we offer an existing repository with analysis-ready data. We hold particularly popular data sets directly, and we connect many other data providers in such a way that the data can be provided automatically on request. Own data can also be brought into the system with little effort by means of a semi-automatic import process. In the analysis work, our engine relieves users of the task of standardizing data sets and correctly processing time series with different resolutions or regularities. Through our web-based UI, analyses can be quickly created via exploratory workflows and the results interactively provided. Using our open interfaces, other tools such as GIS software or Python scripts can also be integrated into the analysis process. Overall, we can thus greatly increase the efficiency of data science work in the geospatial environment, as analyses can be created and reused more quickly.

AI methods of machine learning such as Deep Learning enable radically new applications – also for geodata. Satellite images can be classified and segmented so that clouds can be detected, for example, or defective solar panels on drone images. Beyond such detection, predictions based on current and historical data are extremely valuable. The basic prerequisite for all AI applications is high-quality training data.

Here, Geo Engine helps to create training data by connecting and preparing the raw data. For the actual training process, we offer a connection to frameworks such as Tensorflow and corresponding compute capacities in our cloud. The application of the resulting model is also supported by Geo Engine by providing it as an operator that can be applied in the system. Thus, the Geo Engine offers easy access to AI methods and the possibility to significantly improve analyses and predictions.