How we do it


Whiffle is devoted to incorporating the latest computing innovations to boost the performance of our products. This has enabled us to develop a technology that overcomes the otherwise vastly prohibitive computational requirements associated with high-resolution weather forecasting. Our atmospheric simulations are crafted to run on massively parallel supercomputing platforms and are fully geared to exploit the spectacular acceleration offered by modern Graphical Processing Units (GPUs).


The ultra-high resolution Whiffle uses in the weather forecasting services allows for an accurate representation of small-scale processes like turbulence, surface interactions, cloud formation and precipitation. The strength of our forcasting model is that the actual calculations remain very close to the laws of physics that dictate the processes in the real atmosphere. We therefore do not have to rely on modeling assumptions that are common in traditional weather models.


Our integrated fine-scale weather forcasting system first uses a pre-processing engine to convert large-scale weather data into customized input fields. The massive computer power of the multi-core CPU/GPU engine is then unleashed to simulate the weather over the desired area at a resolution of 100m or finer. As a final step, results are ported to the post-processing engine that applies additional model output statistics and prepares data for transfer or visualization.


Schematic representation of whiffle’s integrated weather forecasting system