Near real-time sensor data is growing exponentially faster than our ability to sense and make sense of it as the current classical computing and supercomputing approaches require an enormous volume of computational resources, storage, and training time. Quantum computing is posed to exponentially outperforms today's high-performance computers and accelerate the evolution of information occurring in classical systems and sensor networks. The Lightning Imaging Sensor (LIS) mounted on the International Space Station (ISS) locates, senses, and detects lightning activities from low Earth orbit and measures radiant energy at millisecond timing over a broad regional spectrum. In this paper, we are introducing a space-to-ground hybrid quantum-classical machine learning architecture to demonstrate the application potential and the feasibility of Hybrid Quantum Neural Networks using the ISS LIS lightning dataset and a corresponding background dataset as inputs (1) to train a classical deep neural network model (2) where the feature extraction outputs are encoded as quantum states using multiple calibrated quantum encoding patterns (3) then used as part of a quantum feature mapping process, (4) allowing for enhanced hybrid QNN training, optimization, and auto-differentiation, setting the path towards polynomial advantage. The following is how the rest of the paper is structured: The introduction is covered in section I, and the motivation and our contribution are covered in section II. Section III presents related works. Section IV describes the NRT lightning imaging sensor. Section V describes the space- to-ground hybrid quantum architecture. Section VI. Space-to-ground hybrid quantum-classical machine learning architecture. Section VII describes quantum machine learning. Section VIII presents a performance analysis. Finally, section XI concludes the research paper.