Big Data in Soil Science – Pandora’s Box or Treasure Chest?

Abstract:

Over the last two decades, new developments in sensing technologies resulted in a massive increase of available data in environmental science and the increase is still ongoing. In the very near future, new satellite missions will provide data of global coverage in a hitherto unknown degree of detail. Wireless sensing technologies offer today the possibility to observe dynamics of soil-related state variables in highest spatial and temporal resolution. Mobile sensing technologies allow the exploration of larger areas in shorter times. On the other hand, the Open Data movement removes permission barriers and data become more and more freely available, opening up new possibilities for data integration.

All this fosters new ambitions but also new challenges for environmental science. Bigger data allows us to ask bigger questions. But, a larger amount of data on its own is worthless. New approaches and methods to handle, structure, analyze, visualize, and to integrate the data into models are needed. And, many of the potential answers are no longer accessible analytically or experimentally. Big Data mining in soil science has specific challenges on its own. The data are often highly heterogeneous, resulting from observations at different spatial and temporal scales, representing the result of highly complex interactions and feedbacks between different environmental compartments. In view of this complexity, asking the “right” question can become a very challenging task.

Big Data mining does not replace the need for profound understanding of processes. Collaboration across different scientific disciplines might be crucial to include all relevant knowledge. From this perspective, Big Data science is a supplement but not a substitute for traditional process-related studies and data-collection. The talk will attempt to shed some light on challenges and potentials of Big Data science from the perspective of traditional soil hydrological science.