Remote Sensing & Next-Generation Satellite LiDAR

Overview

We develop computational LiDAR pipelines that push height-resolved sensing from airborne to spaceborne platforms. By combining compressive sampling with machine learning, we reconstruct forest structure, topography, and canopy height from photon-limited measurements that conventional processing cannot resolve. Transformer models, hyperspectral side-information, and emulation frameworks bridge high-altitude and satellite regimes.

A central focus is NASA’s CASALS (Concurrent Artificially-intelligent Spectrometry and Adaptive Lidar System), a spaceflight-prototype lidar for the Surface Topography and Vegetation (STV) observing system. We develop the reconstruction methods for its sparse, photon-starved measurements, most recently a diffusion-based Bayesian framework that jointly performs compressed-sensing recovery, denoising, and super-resolution of HyperHeight data cubes.