WASABI

Pakistan ranks third in water scarcity world-wide, and suffers from a lack of proper water resource management. While drying up on the one hand, the country experiences catastrophic floods on the other. Therefore, to better manage water resources so that to enforce an even distribution of water throughout the year and to predict flooding beforehand, a solution is required to densely supervise the volume of water bodies (including rivers, lakes, and reservoirs) and predict the ratio at which the water resources are varying. To this end, the project WASABI (WAter resource estimation by SAtellite Based Image analysis) aims at laying the foundations for a system that quantitatively monitors Pakistan’s water bodies and helps national authorities with a more proactive water supply management.

To do so, the WASABI system will employ remote sensing satellite imagery of Pakistan’s water bodies. This data will be processed with state-of-the-art machine learning / deep learning techniques to estimate the current quantity of water and - based on historic time series of population and seasonal capacity fluctuation - predict the amount of water available in the near future. WASABI‘s core research challenge lies in the fact that labeled data for Pakistani waterbodies is scarce, which is why unsupervised learning and transfer learning from densely labeled regions such as Europe will be explored.

WASABI brings together computer vision experts from the LAVIS Working Group at RheinMain University of Applied Sciences (HSRM) in Wiesbaden with colleagues from the TUKL-NUST R&D Centre at the National University of Sciences and Technology (NUST) in Islamabad. The project also features an academic exchange program, with interns from NUST visiting Wiesbaden on an annual basis.