When thinking of the Earth’s atmosphere, the troposphere is often considered the closest layer to the surface. However, the planetary boundary layer (PBL), the lowest part of the troposphere, actually has the most significant impact on surface weather. Recognized as an important scientific issue in the 2018 planetary science decadal survey, the PBL has the potential to improve storm forecasting and climate projections.
Adam Milstein, a technical staff member at Lincoln Laboratory’s Applied Space Systems Group, explains, “The PBL is where the surface meets the atmosphere, facilitating exchanges of moisture and heat that contribute to severe weather and climate change. It is also where humans reside, and the movement of aerosols within the PBL is crucial for air quality and human health.”
Despite its importance for weather and climate studies, certain aspects of the PBL, such as its height, are challenging to accurately determine with current technology. Over the past four years, Lincoln Laboratory staff have been conducting research on the PBL, focusing on utilizing machine learning for 3D-scanned atmospheric profiles and enhancing the vertical structure of the atmosphere for improved drought prediction.
This research is a continuation of over a decade of work on neural network algorithms developed by Lincoln Laboratory for NASA missions. These algorithms have significantly enhanced storm monitoring and climate forecasting capabilities through missions like TROPICS and Aqua. By using deep learning techniques, the laboratory aims to further improve the accuracy of PBL details.
Collaborating with NASA, Lincoln Laboratory has shown that these new algorithms can provide more precise information about the PBL, particularly regarding its height. This knowledge is crucial for various applications, including drought prediction, which is essential for addressing global issues related to water scarcity.
By combining deep learning techniques with existing operational approaches, Lincoln Laboratory and NASA’s Jet Propulsion Laboratory are working to enhance drought prediction over the continental United States. The goal is to create a reliable tool for scientists to use in the long term.
The next phase of the project involves comparing the deep learning results with direct measurements collected by instruments like radiosondes flown on weather balloons. This validation process will help quantify the accuracy of the developed techniques and demonstrate their potential for advancing drought prediction capabilities.
Overall, this improved neural network approach holds promise for exceeding current drought prediction capabilities and becoming a valuable tool for scientific research in the future.
