Yu-Feng Ho
I am a research assistant / geoinformatician focusing on lidar processing, machine learning on terrain and vegetation height modeling. I am also working on cloud-native vector data format and functionality development.
Sessions
Accurate global forest aboveground biomass (AGB) mapping is important for reducing the uncertainties in terrestrial carbon sink assessments. Current global approaches for AGB mapping, such as ESA CCI Biomass maps, rely on methods that are globally optimized; thus, they may result in lower performance when evaluated in specific regions. In this workshop, we aim to present the cloud-optimized and quality-filtered GEDI point observations and analyze how they can contribute to improved global biomass mapping. The outcomes of the workshop will help identify potential improvements to the datasets and define future experimentation to assess their contribution to AGB mapping.
The workshop will start with a presentation of the OEMC high-quality GEDI point dataset. Then, the OEMC Stakeholder, Gamma Remote Sensing, will present their global and regional AGB activities and requirements for the satellite LiDAR data. The workshop will conclude with a discussion about the presented OEMC GEDI dataset and define future directions of the experimentation.
GEDTM30 (github.com/openlandmap/GEDTM30) is a open source global digital terrain model at 1 arc second. It is the first permissive license 1 arc-second terrain model of the world (under CC-BY license). Upon this model, we are presenting a framework to merge national, state-based or individual digital terrain model to improve GEDTM30 data quality locally. Due to permissive license, GEDTM30 can be used as a base layer to create derive products. By merging local lidar DTMs with GEDTM30, it opens the gate to federation of data, sharing the common goods but keeps the best interests for individuals. This framework will be tested by GEDTM30 with lidar data from several countries (e.g. Romania, Italy, the Netherlands, Brazil, etc), and the land surface variables are included to assess the merged GEDTM30 quality. In the end, the code and framework will be open to serve any stakeholders to improve DTM quality of their area of interest and have freedom to decide for contribution.
Spaceborne Lidar, such as ICESat-2 and GEDI, is global missions for land surface monitor tools across terrain, vegetation and ice monitoring. The huge terabytes volume of data and non-cloud-optimized format thwarts the usage and access for the dataset. In the workshop, we are presenting an algorithm to reorganize spaceborne lidar and a STAC visualization solution. The functionalities will be demonstrated in Jupyter notebook. covering accessing STAC collections of ICESat-2 and GEDI respectively, spatial and temporal lazy-loading from DuckDB, and data exportation to desired format. The workshop will be in Python to connect various software APIs.
This workshop provides a hands-on tutorial to merge to improve global digital terrain models (DTMs) by high-quality local LiDAR data while maintaining consistency with a standardized global framework.
The session complements the oral presentation "A Framework of Federal Global Ensemble Terrain Model" and offers participants a practical workflow to empower GEDTM30 data users to improve GEDTM30 for local application.
The workshop is structured in three parts:
(1) Introduction and Data Access: Participants will learn to access and visualize GEDTM30 elevation data through STAC-compliant endpoints and understand its spatial structure and metadata.
(2) Local LiDAR Integration: This section focuses on preprocessing local high-quality LiDAR-derived DTMs, resampling them to match the GEDTM30 30-meter grid, and incorporating them into the ensemble model to enhance local accuracy while preserving global consistency.
(3) Land Surface variable Derivation and Validation: Participants will derive surface parameters (e.g., slope, aspect) from the enhanced terrain model and perform validation analyses to quantify improvements and ensure consistency with the global baseline.
This workshop is intended for researchers, data scientists, and GIS professionals interested in terrain modeling, geospatial data fusion, and scalable environmental data processing workflows.