Satellite data validation using the WRF Model and High Performance Computing paradigms
TimeTuesday, July 246:30pm - 8:30pm
DescriptionWeather forecast has been a great challenge for the National Weather Service (NWS) offices (and many other local weather laboratories around the world), due the atmosphere complexity, computational constraints, models accuracy, historical records availability, and several other issues, that affect the final result either in a direct fashion or in an indirect one. Hence, lots of money have been put at the disposal of the NWSs and other NOAA affiliates to work in the improvement of these models.
My proposed investigation it is mostly motivated by the great volume of weather data available at this moment, moreover the recently launched satellite GOES-R (or GOES-16) features new instruments with a higher spatial resolution which allows a more accurate forecast. The key idea will be to work over a particular region of the Earth (most likely over North American territory), gather data measured by fixed stations over the surface, lookup at the satellite data, use both of those inputs as a "calibration" data for the WRF model and also use it as "validation" data (when historical), to approximate errors between the numerical model and the actual atmosphere behavior at a particular time. Then, based on the initial results a calibration of the model will be executed in an attempt to improve the final results (which is short-term forecast), all of this process executed over a cluster taking advantage of HPC paradigms.