A Lidar-Based 3-D Photosynthetically Active Radiation Model Reveals the Spatiotemporal Variations of Forest Sunlit and Shaded Leaves

Abstract

Accurately identifying sunlit and shaded leaves using process-based ecological models can improve the simulation accuracy of forest photosynthetic rates and potential carbon sequestration capacity. However, it is still challenging to characterize their three dimensional (3-D) spatiotemporal distributions due to the complex structure. In this study, we developed a light detection and ranging (lidar)-based approach to map the spatiotemporal distribution patterns of photosynthetically active radiation (PAR) and sunlit and shaded leaves within forest canopies. By using both terrestrial laser scanning (TLS) and unmanned aerial vehicle-based lidar system (UAV-LS), we analyzed the influences of different scanning geometries and associated point densities on the separation of sunlit and shaded leaves. Moreover, we further investigated the effects of woody materials and penumbra sizes on identifying sunlit and shaded leaves by separating the foliage and woody materials and estimating the penumbras of sunlit leaves. Our results showed that: (1) The proposed lidar-based PAR model could well capture the variations of field-based pyranometer measurements using fused point data by combining UAV-LS and TLS data (mean R-square = 0.88, mean root mean square error (RMSE) = 155.5 μmol·m−2·s−1, p < 0.01). The separate UAV-LS and TLS-based fractions of sunlit leaves were averagely overestimated by 34.3% and 21.6% when compared to the fused point data due to their different coverages and comprehensiveness. (2) The woody materials showed different effects on sunlit leaf fraction estimations for forest overstory and understory due to the variations of solar zenith angle and tree spatial distribution patterns. The most noticeable differences (i.e., −36.4%) between the sunlit leaf fraction before and after removing woody materials were observed around noon, with a small solar zenith angle and low-density forest stand. (3) The penumbra effects were seen to increase the sunlit leaf fraction in the lower canopy by introducing direct solar radiation, and it should be considered when using 3-D structural information from lidar to identify sunlit and shaded leaves.

Uncrewed aircraft system spherical photography for the vertical characterization of canopy structural traits

“…An alternative approach is laser scanning technology, conducted either airborne (ALS; Korhonen et al ., 2011 ; Korpela et al ., 2014 ) or from terrestrial platforms (TLS; Calders et al ., 2020 ). TLS (Hosoi & Omasa, 2006 ) or ALS (Lim et al ., 2003 ; Lovell et al ., 2003 ) can be used to retrieve LAI and to model canopy light interception (S. Tian et al ., 2021 ) in three dimensions. However, the cost and complexity of laser scanning represent two major drawbacks that prevent its widespread adoption.…”

3D Point Clouds in Forest Remote Sensing

“…Among these, only one paper exclusively uses passive RS data [21], while 29 papers use at least one LiDAR dataset in the analysis [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][22][23][24][25][26][27][28][29][30]. Ten papers exclusively use airborne laser scanning (ALS) data [4,6,7,10,11,13,18,23,26,27], nine papers exclusively use terrestrial laser scanning (TLS) data in the analysis [3,9,15,16,20,22,24,25,30], two papers exclusively use mobile laser scanning (MLS) data …”

“…Finally, five papers use combined active and passive remote sensing data sets [2,14,17,19,28]. Regarding the scale of the analysis, 18 of the studies perform individual tree level (ITL) analysis [1][2][3][8][9][10][11][12][14][15][16]19,20,[23][24][25][26]30], eight papers report stand level (SL) analysis [6,7,17,18,21,22,27,29] and four report a combination of ITL and SL [4,5,13,28]. Tree position, diameter at breast height (DBH) and individual tree height (h) are the most common variables of interest, analyzed in nine, six and six papers, respectively, while the most commonly used methods are 3D reconstruction, point filtering and statistical modelling, which are used in eight, five and five papers, respectively (see Table 1).…”

“…The SM include 30 published manuscripts (see Table 1): 28 research papers [1,2,[4][5][6][7][8][9][10][11][12][13][14][15][17][18][19][20][21][22][23][24][25][26][27][28][29][30], one letter [16] and one technical note [3]. Among these, only one paper exclusively uses passive RS data [21], while 29 papers use at least one LiDAR dataset in the analysis [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]…”

“…Cont. SM include 30 published manuscripts (see Table1): 28 research papers[1,2,[4][5][6][7][8][9][10][11][12][13][14][15][17][18][19][20][21][22][23][24][25][26][27][28][29][30], one letter…”