Man-Yau (Joseph) Chan
Assistant Professor,
Department of Geography,
The Ohio State University
"Advancing the Geospatial Big Data Fusion of Observations and Forecast Models"
Wednesday, Aug 28, 2024, Schedule:
- Nespresso & Teatime - 417 DSL Commons
- 03:00 to 03:30 PM Eastern Time (US and Canada)
- Colloquium - 499 DSL Seminar Room
- 03:30 to 04:30 PM Eastern Time (US and Canada)
In-person attendance is requested.
499 DSL Seminar Room
Zoom access is intended for external (non-departmental) participants only.
Meeting # 942 7359 5552
Abstract:
Numerical weather predictions (NWPs) are essential for a wide range of socioeconomic activities (e.g., green energy production and disaster warnings). Advancements in the accuracy, precision and spatial resolution of NWP will thus benefit society. Since NWP generates predictions by evolving inputted initial atmospheric states (i.e., initial conditions), the precision and accuracy of NWP depends on the accuracy and precision of those initial conditions. These initial conditions are generated by the Geospatial Big Data Fusion of observations and past forecasts. This Fusion process is known as data assimilation (DA). As such, advances in DA will improve those socioeconomically important NWPs.
One of the most popular classes of DA methods for thunderstorms are the ensemble Kalman filters (EnKFs). EnKFs convert point observations into multivariate geospatial forecast corrections through multivariate geospatial forecast statistics. This conversion employs Bayesian inference under the assumption of multivariate Gaussian prior statistics. While EnKFs have been remarkably successful at improving thunderstorm NWP, that Gaussian prior assumption often breaks down in thunderstorms. That breakdown can generate forecast-degrading statistical artifacts. As such, to advance thunderstorm NWP, it is essential to (1) identify those artifacts and their impacts and (2) advance DA methods that avoid that Gaussian prior assumption.
In this seminar, I will present on
- Forecast-wrecking artifacts that arise from the violation of EnKFs’ Gaussian prior assumption,
- A novel, efficient and scalable extension of the EnKF (bi-Gaussian EnKFs) that potentially produces better forecasts than the EnKF, and,
- A simple, efficient, scalable and flexible method to suppress sampling errors in Monte Carlo DA methods.
No prior knowledge of DA or meteorology is needed to understand this seminar.
