Abstract:Remote sensing retrieval of foliar biochemical traits via inversion of radiative transfer models is hampered by the uncertainties arising from the ill-posed problem of insufficient remote sensing information for the inversion. The use of leaf transmittance in addition to leaf reflectance in retrieving leaf traits could be a key step to alleviate this issue. To achieve this objective, our study introduces a novel theoretical framework for the simultaneous retrieval of leaf directional-hemispherical reflectance and transmittance, leveraging multi-angular optical observations at the canopy level. This framework partitions the canopy's spectral reflectance in the range of 400 nm to 2500 nm into two distinct components, each aligning with the contributions of 1st-order and multiple-order scatterings of radiation in the canopy. At a specific illumination-observation (IO) angle, these two scattering components are assumed to be determined by leaf hemispherical reflectance and scattering albedo (i.e., the sum of leaf reflectance and transmittance). The average leaf directional-hemispherical reflectance and transmittance can be estimated from combined canopy reflectance observed under different IO angles. The proposed framework was implemented using a single site data collected on August 19th, 2018, in Saihanba National Forest Park, Hebei, China. A comprehensive dataset was compiled, including hyperspectral images of a dense white birch canopy acquired at 8 different illumination-observation (IO) angles. This dataset was augmented with corresponding canopy structural traits, including leaf area index, diameter at breast height, crown radius, tree number, and domain size. Additionally, terrain features such as slope and aspect, along with leaf spectra and background spectra, were included in this extensive dataset. The geometrical-optical model GOST2 was utilized to establish the relationships between canopy reflectance and leaf reflectance and transmittance. Initially, the performance of GOST2 was evaluated, followed by the construction of a leaf spectral dataset generated by PROSPECT-D and input into GOST2 to construct the LUT. Subsequently, leaf reflectance and transmittance were estimated for different combinations of IO angles. The results demonstrated a close agreement between the canopy reflectance factors modelled by GOST2 and the observations obtained from UAV. Overall, leaf reflectance was more accurately retrieved than leaf transmittance across the entire spectrum of interest, primarily due to the significant influence of the 1st-order scattering in canopy reflectance. In the near-infrared region, where the magnitudes of 1st- and multiple-order scatterings were similar, the proposed framework produced satisfactory leaf transmittance with a relative error of <10% when compared to the measured reference data. Furthermore, the results indicated that the absolute error of model simulations should be less than half of the multiple-order scattering, which is particularly stringent for wavelengths characterized by strong absorption features. The proposed framework holds potential for reducing uncertainties in various fields, including global mappings of leaf biochemical traits.

Inverting a canopy reflectance model using a neural network

Inverting a canopy re ̄ ectance model using a neural network

A Method for Inverting Shallow Sea Acoustic Parameters Based on the Backward Feedback Neural Network Model

Inversion of Rice Canopy Chlorophyll Content and Leaf Area Index Based on Coupling of Radiative Transfer and Bayesian Network Models

Retrieval of the Leaf Area Index from MODIS Top-of-Atmosphere Reflectance Data Using a Neural Network Supported by Simulation Data

Simultaneous estimation of leaf directional-hemispherical reflectance and transmittance from multi-angular canopy reflectance

Estimating Forest fAPAR from Multispectral Landsat-8 Data Using the Invertible Forest Reflectance Model INFORM

Inversion of Suspended Particulate Matter Concentration in Maozhou River Based on Band Selection of Hyperspectral Data

Using Artificial Neural Networks to Estimate Cloud-Base Height From AERI Measurement Data

A New Directional Canopy Emissivity Model Based on Spectral Invariants

LAI RETRIEVAL OF REED CANOPY USING THE NEURAL NETWORK METHOD

Synergetic inversion of leaf area index and leaf chlorophyll content using multi-spectral remote sensing data

Simultaneous Estimation of Land Surface and Atmospheric Parameters From Thermal Hyperspectral Data Using a LSTM-CNN Combined Deep Neural Network

A MWCMLAI-Net method for LAI inversion in maize and rice using GF-3 and Lutan radar data

Retrieval of Subtropical Forest Canopy Closure from Remote Sensing by Using Akaike Information Criterion and Artificial Neural Network Model

Study on the LAI and FPAR inversion of maize from airborne LiDAR and hyperspectral data

Identification of Robust Hybrid Inversion Models on the Crop Fraction of Absorbed Photosynthetically Active Radiation Using PROSAIL Model Simulated and Field Multispectral Data

Estimation of Crop LAI Using Hyperspectral Vegetation Indices and a Hybrid Inversion Method

Quantitative Inversion of Long Sequential Leaf Area Index Using Remote Sensing Based on BP Neural Network from 1988 to 2013

Layered Soil Remote Sensing with Multichannel Passive Microwave Observations Using a Physics-Embedded Artificial Intelligence Framework: A Theoretical Study

A Quick Method for Predicting Reflectance Spectra of Nanophotonic Devices via Artificial Neural Network