09-10, 14:00–14:30 (America/Chicago), Grand C
This presentation illustrates a workflow for visualizing and analyzing Cloud-optimized GeoTIFFs and Cloud-Optimized Point Clouds using TiTiler and Lonboard - dynamic tiling and interactive visualization libraries, allowing 2.5D interactive map exploration in Python Jupyter notebooks.
With the increase in cloud computing proximate to data we needed a way to do large data visualization in the notebook based browser environments. For this talk, we will focus on a use-case for Above Ground Biomass (AGB) analysis for visualizing LiDAR derived AGB reference maps along with forest trails for a study region in BCI, Panama. This use case combines LiDAR with high resolution raster layers in an interactive visualization inside a web-based coding notebook. The notebook is developed with an aim of end-to-end processing in the cloud for the NASA-ESA Multi-mission Algorithm and Analysis Platform (MAAP) [1]. MAAP is an open-source scalable and efficient platform for large-scale processing of NASA and ESA datasets along with user generated datasets related to biomass estimation. In this context, a lot of MAAP user data products are published in cloud-optimized formats and accessible via MAAP Spatio-Temporal Asset Catalog (STAC) endpoints [2].
From the perspective of biomass analysis, raw LIDAR data and derived products play a key role. However, raw LiDAR data, typically consisting of billions of points in LAS/LAZ files are not efficient for cloud-based processing attributed to large data egress cost. Hence, Cloud-Optimized Point Cloud (COPC), a compressed and efficient representation of lidar data is used for publishing LiDAR data to STAC, making it suitable for large-scale geospatial applications. In this presentation, we explore a detailed explanation of the process of converting COPC data into a GeoDataFrame. This involves extracting relevant information from the COPC file, such as point coordinates, attributes, and metadata. We will demonstrate how to perform this conversion using open-source tools and provide code examples for reproducibility. Subsequently, we will present Lonboard [3] - a Python library that allows for interactive visualization of geospatial data. We will showcase how to use Lonboard to visualize the GeoDataFrame containing COPC data. This will include creating a map, adding markers for point features, and customizing the visualization with interactive features such as zooming, panning, and highlighting.
To provide context and geographic reference, we will integrate a base map into our visualization using Titiler [4], a dynamic raster tile service. We will demonstrate how to use Titiler to add a base map layer to our Lonboard visualization, enabling users to explore the COPC data in relation to real-world geographic features.
Join this talk if you want to -
- Understand a science use-case for accessing big EO data from STAC
- Explore best practices for converting LiDAR LAS files to COPCs
- Understand visualizing COPCs and rasters in a interactive maps using Lonboard and TiTilers
- Integrate line/polygon geometry visualization to the above maps
References:
- MAAP - https://maap-project.github.io/
- MAAP STAC - https://stac-browser.maap-project.org/
- Titiler - https://developmentseed.org/titiler/
- Lonboard - https://developmentseed.org/lonboard/latest/
