Supercomputers Make Bluesky's National Geographic Datasets a Reality

by | May 7, 2015

Leicestershire, May 7, 2015 ” Aerial  mapping company Bluesky is exploring the power of supercomputers to process and  deliver 3D maps, comprising trillions of data points, for a range of  environmental applications.

Working with HPC Midlands and experts from the  Karlsruhe Institute of Technology (KIT) in Germany, Bluesky has already been  able to devise workflows to create region-wide maps of sun shadow “ of  particular interest when considering the effectiveness of solar panel  installations. Bluesky is also exploring the use of high performance computers  to scale up existing workflows to help create and maintain other key datasets  such as the National Tree Map, air pollution models and thermal heat loss  surveys.

Having invested in the very latest survey equipment, we  are now generating more detailed data covering larger areas than at any other  time in the history of aerial surveying, commented James Eddy, Technical  Director of Leicestershire based Bluesky International. Our nationwide annual  programme of data capture results in around hundreds of terabytes of raw data  every year.

Processing this amount of data on conventional  computers is simply not time or cost effective, added Simon Schuffert, Research  Associate at KIT. However, by partnering with HPC Midlands and Bluesky we have  proven scalable workflows through a ˜divide and conquer' approach made possible  by parallel programming.

New sensors such as those employed by  Bluesky and the analysis techniques devised by KIT are generating larger volumes  of data than ever before, added James Earl-Fraser, Business Development  Associate at HPC Midlands. In order to derive value there is a growing  requirement for automatic and efficient processing such as that offered by high  performance computers.

The Bluesky partnership initially developed  an open source shadow analysis programme that calculated the amount of shade a  3D surface structure is subject to over a day, month or year. With ground  sampling distances of 25 square centimetres this level of detail would, for a  country the size of Great Britain, mean processing over 3.5 trillion elevation  points! The dataset created by the new shadow analysis could theoretically be  used to accurately predict effectiveness of solar panels as small as those  attached to parking ticket machines, for example, or monthly and annual sun  exposure for agricultural areas.

Having proved  the power of high performance computing for solar energy mapping projects, the  partnership is exploring other applications, including the update and  maintenance of Bluesky's National Tree Map and projects to map air pollution and  heat loss from buildings.