AUTHOR=Wu Qiaoli , Yang Shenhui , Jiang Jie 

TITLE=Impacts of clumping effect on canopy reflectance using 3D radiative transfer modeling

JOURNAL=Frontiers in Forests and Global Change

VOLUME=Volume 6 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/forests-and-global-change/articles/10.3389/ffgc.2023.1106773

DOI=10.3389/ffgc.2023.1106773

ISSN=2624-893X

ABSTRACT=The product of leaf area index (LAI) and clumping index (CI) quantifies the effective leaf abundance and distribution across the landscape, and therefore, governs the radiation absorption, evapotranspiration, and carbon assimilation processes in the terrestrial ecosystems. However, a few studies focused on quantifying the sensitivity of canopy bidirectional reflectance distribution function (BRDF) to changes in CI in a forward manner, hampering an accurate understanding of the relationship between CI and BRDF. In this study, we simulated how BRDF responds to changes in CI in Qinghai spruce (Picea crassifolia) forests based on a three-dimensional radiative transfer model LESS and ground-measured data. We found that the LESS model effectively simulated the hot-spot, roof, and bowl-edge characteristics of the canopy BRDF by changing the sun-sensor geometry. We constructed forest scenes with variable CI to investigate the clumping effect on BRDF with different solar and observation angles. The red band bidirectional reflectance factor (BRF) showed higher sensitivity to changes in CI than that in the NIR band. Canopy BRFs in the red band along the principal plane and cross principal planes measured in different seasons showed consistent sensitivity to changes in CI, suggesting that the red band BRF is helpful for CI inversion for forests with different levels of foliage clumping. In the NIR band, canopy BRFs along the principal plane measured in growing seasons (with solar zenith angle (SZA) < 40) and the cross principal plane measured in non-growing seasons (with SZA> 40) were sensitive to changes in CI in highly clumped forests. However, canopy BRF in the NIR band showed low sensitivity to changes in CI in highly clumped forests (CI<0.6), especially along the cross principal plane when SZA was approximately 10. The simulated BRFs in the red  and NIR bands showed relatively low sensitivity to changes in SZAs at a VZA of 40 and 0 , respectively. We highly recommend including the red band BRF for CI retrievals, and using a VZA of 40 in the red band and 0 in the NIR band may help reduce the CI estimation uncertainty caused by changes in SZA.