Hyperspectral LiDAR can recover soil spectral information from beneath vegetation canopies and help estimate key soil properties, according to a study published in Remote Sensing of Environment. The findings offer a new approach to monitoring soils in vegetated areas, where exposed ground is often limited or unavailable.
The study was conducted by a research team led by Prof. WANG Li from the State Key Laboratory of Remote Sensing and Digital Earth at the Aerospace Information Research Institute of the Chinese Academy of Sciences (AIRCAS).
Using a three-dimensional radiative transfer model, the researchers compared passive hyperspectral imagery with hyperspectral LiDAR point clouds to assess their ability to identify understory soils and estimate soil organic carbon and total nitrogen.
In Earth observation, important surface targets such as soil, rock, and water are often partly hidden by vegetation. Passive hyperspectral sensors record sunlight reflected from all surfaces within a pixel, producing mixed signals from leaves, branches, and the ground. This makes it difficult to isolate spectral information from the soil beneath a canopy.
LiDAR, by contrast, is an active sensing technology that uses laser pulses to capture the three-dimensional structure of a scene. Hyperspectral LiDAR extends this capability by recording spectral information at multiple wavelengths. Laser pulses passing through gaps in the canopy can therefore provide both the position and spectral characteristics of understory surfaces.
The researchers evaluated the two approaches in terms of radiance intensity, spectral curves, and spectral indices. They also investigated the mechanisms responsible for radiance errors in active and passive hyperspectral observations of understory surfaces.
The results showed that hyperspectral LiDAR detected soils more reliably under dense vegetation and complex multiple-scattering conditions. In passive hyperspectral images, soil signals were easily masked by the canopy as vegetation cover increased. Hyperspectral LiDAR, however, captured three-dimensional point clouds and spectral information across the entire scene.
By filtering the point clouds, the researchers were able to isolate clean ground returns. The resulting soil spectra were less affected by light scattering between the canopy and ground. The recovered spectra also showed potential for estimating soil organic carbon and total nitrogen, with coefficients of determination of 0.332 and 0.485, respectively.
"Our study shows that hyperspectral LiDAR can recover understory soil spectra that are difficult to isolate using passive imagery," said Prof. WANG. "The technique provides a potential pathway toward three-dimensional monitoring of interconnected forest and soil systems, particularly in areas where exposed soil is rarely visible."
Doctoral student HAO Yishuo is the first author of the study. Prof. WANG Li and Assistant Prof. BI Kaiyi from AIRCAS are the corresponding authors.
The research was supported by the National Natural Science Foundation of China, the National Key R&D Program of China, and an independent research project of AIRCAS.
NDVI distributions derived from the hyperspectral imagery, and from the HSL point cloud before and after filtering. (Image by AIRCAS)
Comparison of echo intensity of HSL ground points with and without vegetation cover, under fractional vegetation cover of (a) 0.77, (b) 0.38, and (c) 0.19. (Image by AIRCAS)
Retrieval results of soil organic carbon (SOC) and total nitrogen (TN) using understory spectra derived from the HSL point cloud. (Image by AIRCAS)
Research News
Hyperspectral LiDAR Reveals Soil Properties Beneath Vegetation Canopies
Hyperspectral LiDAR can recover soil spectral information from beneath vegetation canopies and help estimate key soil properties, according to a study published in Remote Sensing of Environment. The findings offer a new approach to monitoring soils in vegetated areas, where exposed ground is often limited or unavailable.
The study was conducted by a research team led by Prof. WANG Li from the State Key Laboratory of Remote Sensing and Digital Earth at the Aerospace Information Research Institute of the Chinese Academy of Sciences (AIRCAS).
Using a three-dimensional radiative transfer model, the researchers compared passive hyperspectral imagery with hyperspectral LiDAR point clouds to assess their ability to identify understory soils and estimate soil organic carbon and total nitrogen.
In Earth observation, important surface targets such as soil, rock, and water are often partly hidden by vegetation. Passive hyperspectral sensors record sunlight reflected from all surfaces within a pixel, producing mixed signals from leaves, branches, and the ground. This makes it difficult to isolate spectral information from the soil beneath a canopy.
LiDAR, by contrast, is an active sensing technology that uses laser pulses to capture the three-dimensional structure of a scene. Hyperspectral LiDAR extends this capability by recording spectral information at multiple wavelengths. Laser pulses passing through gaps in the canopy can therefore provide both the position and spectral characteristics of understory surfaces.
The researchers evaluated the two approaches in terms of radiance intensity, spectral curves, and spectral indices. They also investigated the mechanisms responsible for radiance errors in active and passive hyperspectral observations of understory surfaces.
The results showed that hyperspectral LiDAR detected soils more reliably under dense vegetation and complex multiple-scattering conditions. In passive hyperspectral images, soil signals were easily masked by the canopy as vegetation cover increased. Hyperspectral LiDAR, however, captured three-dimensional point clouds and spectral information across the entire scene.
By filtering the point clouds, the researchers were able to isolate clean ground returns. The resulting soil spectra were less affected by light scattering between the canopy and ground. The recovered spectra also showed potential for estimating soil organic carbon and total nitrogen, with coefficients of determination of 0.332 and 0.485, respectively.
"Our study shows that hyperspectral LiDAR can recover understory soil spectra that are difficult to isolate using passive imagery," said Prof. WANG. "The technique provides a potential pathway toward three-dimensional monitoring of interconnected forest and soil systems, particularly in areas where exposed soil is rarely visible."
Doctoral student HAO Yishuo is the first author of the study. Prof. WANG Li and Assistant Prof. BI Kaiyi from AIRCAS are the corresponding authors.
The research was supported by the National Natural Science Foundation of China, the National Key R&D Program of China, and an independent research project of AIRCAS.
NDVI distributions derived from the hyperspectral imagery, and from the HSL point cloud before and after filtering. (Image by AIRCAS)
Comparison of echo intensity of HSL ground points with and without vegetation cover, under fractional vegetation cover of (a) 0.77, (b) 0.38, and (c) 0.19. (Image by AIRCAS)
Retrieval results of soil organic carbon (SOC) and total nitrogen (TN) using understory spectra derived from the HSL point cloud. (Image by AIRCAS)