Multisensor Agile Adaptive Sampling of the Atmosphere Driven by Real-time Analytics

Abstract:

Atmospheric processes are complex and often involve interactions between the clear and cloudy parts of the atmosphere. In remote sensing, it is common to rely on multiple sensors to observe these different parts of the atmosphere, and to gain insights regarding their interactions using multi-sensor retrieval techniques. Traditionally, the value of multi-sensor observations emerges long after their collection, during a post-processing phase. Unfortunately, any knowledge gained at that stage cannot be used to adjust the observing strategy, often leaving an incomplete picture of the atmosphere.

Today, advancements in communications, computational resources and sensor capabilities enable us to use real-time analytics to retrieve the current state and predict the short-term future state of the atmosphere. In this way, multi-sensor observations can be used in real time to optimize the spatiotemporal sampling of atmospheric processes. We demonstrate the value of this new observing paradigm by adapting the sampling strategy of a phased-array radar and a polarimetric scanning cloud radar, two different yet uniquely complementary systems, using real-time observations from a geostationary satellite, a surface camera and the radars themselves. The tailored pointing and increase in sensitivity realized through this framework, which we call MAAS (Mutlisensor Agile Adaptive Sampling), enables the steered radars to sample a diverse set of atmospheric phenomena such as shallow cumuli, lightning-induced ice crystal orientation and a series of waterspouts.