SIParCS 2022 - Fletcher Wadsworth
Collecting and Processing Point Cloud Data for Snow Depth and Density Measurement
Obtaining high-quality hydrometeorological measurements are essential for climate modeling and water management. Snow in particular is challenging and expensive to measure. Snowpack is highly variable in space and difficult to predict due to its dependence on wind, which is also dependent on wind. A measurement technique which yields higher accuracy at low cost would be of great use to Earth Systems Science and water management entities. Here we adapt an automotive LiDAR sensor and the Raspberry Pi, a low-cost single board computer, to develop a system which significantly improves snow measurement capabilities. LiDAR point cloud data is large and unwieldy, leading to concerns regarding deployment of such a measurement system in remote locations where data cannot be retrieved frequently. In this project, we explore process level parallelism to simultaneously interface with the sensor to collect data and process the data to extract key observations with reduced memory requirements. This allows for the system to operate uninterrupted for longer periods of time while retaining most important data. The motivation and efficacy of this approach is discussed. Preliminary point cloud processing routines are also developed: one to estimate the depth of snowpack over an area with far more granularity than individual measurements, and another to determine the density of snowflakes in the air. The performance and output of these algorithms are discussed and future improvements are suggested.
Mentors: Ethan Gutmann, Agbeli Ameko
Slides and poster