Table 4.

MLS application in crop management.

Crop	Crop Management Practices	Important Findings	Ref.
Apple	Estimation of tree canopy density	The 2D algorithm identified peak points at the canopy center, correlating with apple variety, tree age, and tall spindle form. Notably, the middle section exhibited higher canopy points than the top and bottom. The 3D algorithm excelled in evaluating tree canopy point density, surpassing the 2D version. Ensuring precise alignment during scanning is vital to avoid experimental errors.	[109]
Wheat	Crop biomass estimation	A proximal active reflectance sensor offering spectral indices and crop height estimates was compared to the LiDAR system. The correlation between LiDAR-derived crop height and crop biomass was 0.79, revealing substantial variability in biomass across the field. This suggested the potential of LiDAR technology for large-scale operations and site-specific management.	[110]
Apple	Leaf area detection	A test platform was constructed to measure orchard tree canopy leaf areas manually. Polynomial regression, BPNN, and PLSR algorithms were employed to analyze the relationship between canopy point clouds and leaf areas. The BP neural network (86.1% test, 73.6% verification accuracy) and PLSR (78.46% test, 60.3% verification accuracy) outperformed the Fourier function in polynomial regression (59.73% accuracy).	[111]
Vineyard	Estimation of canopy size parameters (thickness, height, and volume) and LAI	UAVs, MLSs, and mobile apps (MA) effectively estimated canopy size variations. Strong correlations (R2 > 0.7) were observed, with the highest at R2 = 0.78 (UAV vs. MLS) for canopy volumes. Height data showed robust correlations (R2 = 0.86, MA vs. MLS), while thickness data had weaker correlations (R2 = 0.48, UAV vs. MLS). LAI demonstrated moderate but consistent correlations with canopy volumes, ranging from R2 = 0.69 (LAI vs. UAV) to R2 = 0.74 (LAI vs. V MLS).	[112]
Sorghum	Detection and measurement of estimation of individual sorghum panicles	Panicles were identified with 89.3% overall accuracy, encompassing a 10.7% omission and 14.3% commission rate. Estimated panicle dimensions demonstrated a strong correlation with LiDAR-derived measurements (panicle length: r = 0.88, RMSE = 3.10 cm; panicle width: r = 0.79, RMSE = 1.67 cm; plant height: r = 1.00, RMSE = 0.80 cm). Comparison with harvested panicle data revealed moderate-to-high correlations (panicle length: r = 0.79, RMSE = 2.48 cm; panicle width: r = 0.63, RMSE = 1.49 cm; plant height: r = 0.86, RMSE = 11.4 cm).	[113]
Cabbage, Leek, Potato, Wheat	Canopy estimation	Soil and plant segregation was accomplished by calculating weighted sums, eliminating the need for additional sensor data. This dynamic method extracted vegetation from point clouds in strips with varying coverage and sizes. The resulting vegetation clouds were validated against drone imagery, confirming a precise match with all green areas.	[114]
Ryegrass	Estimation of fresh weight and dry matter yield	R2 between FWY and seasons (winter, spring, summer, and autumn) were 0.81, 0.92, 0.94, and 0.90, respectively. Similarly, the R2 values between DMY and the seasons were 0.87, 0.73, 0.87, and 0.79, respectively. These results suggest that LIDAR estimation of DMY is accurate within seasons for paired-row breeding plots. However, it is sensitive to significant changes in dry matter content (%) among seasons, requiring seasonal algorithms for correction.	[115]
Miscanthus giganteus	Measurement of crop height	The sensor assessed stem densities in static mode, yielding an average error of 5.08% (max 8%, min 1.8%). It also measured crop height in a 5 × 10 m field, showing a 4.2% error compared to manual measurements. The sensor traversed a field edge in dynamic mode, generating a three-dimensional crop structure. An ordinary least-squares surface-fitting algorithm produced top and ground surfaces, resulting in an average crop height. Dynamic measurements showed a 3.8% average error (max 6.5%, min 1.5%).	[116]