| Presentation Title: | Blending AI- and Physics-based Models with Advanced Sensing Techniques to Support Weather Research |
|---|
| Full Name: | Marouane Temimi |
|---|
| Affiliation / Institution: | Stevens Institute of Technology |
|---|
Abstract: | Building resilient communities in urban areas requires a strong understanding and accurate modeling of extreme weather hazards. The complexity of the processes involved in the development of weather extremes suggests the consideration of new modeling and monitoring techniques that overcome the limitations of the standard physics-based approaches. The advent of AI and Machine Learning (AI/ML) has certainly provided the community with new alternatives to achieve weather-ready communities.
AI/ML techniques in conjunction with advancement in sensing techniques augmented the volume of collected datasets. They facilitated at the same time the processing of the increasing volume of observations to serve the modeling and predictions of weather extremes. AI/ML techniques have shown the potential to advance the modeling of predictions of weather extremes when used in a standalone configuration or when blended with physics-based models. In addition, with the abundance of observations from various platforms and sensors, ranging from in situ to space-borne ones, AI/ML techniques were useful to pre-process the collected datasets and facilitate their integration in the modeling and decision-making processes. Furthermore, AI/ML techniques leveraged citizen science and crowdsourcing to infer novel observations from otherwise qualitative posts to generate quantitative information that helps to enhance the impact assessment of extreme events.
We will present studies in which AI/ML significantly enhanced the analysis of vast datasets collected using various sources. We will present examples like those addressing the segmentation of satellite data, the processing of images and videos from on-the-ground cameras, and the analysis of measurement of environmental variables across urban areas. We will demonstrate the usefulness of AI/ML techniques to support the modeling and prediction of extreme weather events and eventually achieve resilient communities. |
