The Hydrostreamer is a scientific computation library for R-software, aimed at improving the calculation of runoff products to proper watershed. The package works by downscaling the distributed runoff data by spatial relationship between areal units in runoff data layer and an explicitly represented river network. For large areas, Hydrostreamer calculation can be time-consuming, and therefore Aalto University is currently working on integrating the system into the CSC Taito supercluster in order to improve the processing time for large hydrological datasets.
The focus of the work is in using the Spark environment in the CSC cloud for accomplishing the task. However, the work aims to test different methods of parallelization in super computing environment, and thus see how Hyrdostreamer calculation can best be improved. As a side-effect, the work will also provide knowledge of how the R-software can be used to solve spatial problems in supercomputing environments.
Hydrostreamer on GitHub: https://github.com/mkkallio/hydrostreamer