Deep Learning Framework for High-Resolution Large-Scale Vegetation Optical Depth Mapping Using Airborne LiDAR and Mobile GNSS-T Data
Open Access
- 18 March 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19 (19391404) , 12087-12100
Abstract
Accurate, high-resolution mapping of Vegetation Optical Depth (VOD) is critical for assessing forest biomass, ecosystem health yet current satellite products lack the spatial detail needed to capture fine-scale heterogeneity. Ground-based GNSS Transmissometry (GNSS-T) offers meter-scale VOD mea surements but is limited in spatial coverage due to operational constraints. This paper presents a novel data fusion framework that combines mobile GNSS-T observations with airborne LiDAR to produce large-scale VOD maps at 5 m × 5 m resolution blocks. The method leverages a Cone-Based Point Distribution Imaging (CPDI) approach, which transforms LiDAR point clouds within GNSS-aligned conical sampling zones into 28 × 30 × 3 angu lar–height histograms, encoding vegetation structure relevant to microwave attenuation. These CPDI images are used to train a convolutional neural network (CNN) to predict block-level VOD values, which are then mapped across entire LiDAR domain over a 400 m × 400 m region. The framework was validated using data from the SMAPVEX22 campaign at Harvard Forest, MA and Millbrook, NY, achieving strong agreement between predicted and measured VOD (R2 = 0.93 training, R2 = 0.74 testing) and demonstrating consistent correlations with in-situ basal area (R2 = 0.72) and biomass (R2 = 0.46) at the regional level. The results show that integrating GNSS-T and LiDAR enables scalable, high-resolution VOD mapping across heteroge neous forested landscapes, offering new capabilities for drought monitoring, forest management, and wildfire risk assessment.Keywords
Funding Information
- National Aeronautics and Space Administration
- Future Investigators in NASA Earth and Space Science and Technology program (80NSSC25K0460)
This publication has 36 references indexed in Scilit:
- A review of forward modelling and retrieval approaches for forest soil moisture and vegetation optical depth using L-band radiometryRemote Sensing of Environment, 2025
- Divergence between in situ and satellite-based estimates of forest canopy water contentRemote Sensing of Environment, 2025
- Estimation of mangrove heights and aboveground biomass using UAV-LiDAR, Sentinel-1 and ZY-3 stereo imagesEcological Informatics, 2025
- Synergistic retrieval of mangrove vital functional traits using field hyperspectral and satellite dataInternational Journal of Applied Earth Observation and Geoinformation, 2024
- Utility of L-band and X-band vegetation optical depth to examine vegetation response to soil moisture droughts in South AsiaRemote Sensing of Environment, 2023
- A new global C-band vegetation optical depth product from ASCAT: Description, evaluation, and inter-comparisonRemote Sensing of Environment, 2023
- SMAP Detects Soil Moisture Under Temperate Forest CanopiesGeophysical Research Letters, 2020
- Overview of SMOS performance in terms of global soil moisture monitoring after six years in operationRemote Sensing of Environment, 2016
- Modelling vegetation and the carbon cycle as interactive elements of the climate system2002
- Lidar Remote Sensing of the Canopy Structure and Biophysical Properties of Douglas-Fir Western Hemlock ForestsRemote Sensing of Environment, 1999