The group’s data scientists’ main responsibilities include downloading global model data from various sources, running and developing in-house models and converting incoming data to formats used by clients and services. As such the work falls into the behind-the-scenes category; as long as there are not any problems with the data flow, all the continuously running metocean computer programs remain invisible to StormGeo employees and clients, and yet, they form a foundation most of the company’s services depend on.
StormGeo’s metocean data handling framework is never in a frozen state but is continuously developed and modified to comport with the ever changing data requirements and data source details. Possible improvements in the delivery time to the clients and operations are also sought upon. Just recently, the whole StormGeo Shipping in-house WaveWatch3 wave model running and post-processing framework was rewritten resulting in data delivery to BVS7 clients an hour earlier than before.
In addition to developing and maintaining the existing metocean data download and processing systems at StormGeo, the metocean data management and acquisition group continuously researches new possible sources of data to be incorporated into StormGeo products. As technological advances have led to a rapidly increasing number of available metocean data types, it is important to keep up with different data options and assess which ones are most suited for StormGeo’s needs. An example of a future addition to StormGeo data products is multi-model ensemble-based probability zones for tropical cyclone proximity created by StormGeo Houston. Such an approach enables the depiction of track-related tropical cyclone uncertainty based on the prevailing weather conditions rather than historical forecast track errors, as is the case with the traditional tropical cyclone uncertainty cones. This product will be a valuable addition to the tools used by StomGeo’s Shipping Route Advisory Services and will also be included in BVS™ and Fleet Performance FDSS in the future.
As previously mentioned, we have recently updated our WaveWatch3 wave model data processing resulting in a 1 hour gain in StormGeo’s total processing time.