Evaluation of Leaf Area Index (LAI) of Broadacre Crops Using UAS-Based LiDAR Point Clouds and Multispectral Imagery

Abstract

Leaf area index (LAI) is an established structural variable that reflects the 3D leaf layering of vegetation in response to environmental inputs. In this context, unmanned aerial system (UAS)-based methods present a new approach to such plant-to field-scale LAI assessment for precision agriculture applications. This study used UAS-based light detection and ranging (LiDAR) data and multispectral imagery (MSI) as two modalities to evaluate the LAI of a snap bean field, toward eventual yield modeling and disease risk assessment. LiDAR-derived and MSIderived metrics were fed to multiple biophysical-based and regression models. The correlation between the derived LAI and field-measured LAI was significant. Six LiDAR-derived metrics were fit in eight models to predict LAI, among which the square root of the laser penetration index (√LPI) achieved the most accurate prediction result ( 𝑹 𝟐 =0.61, nRMSE=19%). The MSIderived models, which contained both structural features and spectral signatures, provided similar predicting effectiveness, with predicted 𝑹 𝟐 ≈0.5 and nRMSE≈22%. We furthermore observed variation in model effectiveness for different cultivars, different cultivar groups, and different UAS flight altitudes, for both the LiDAR and MSI approaches. For data collected at a consistent flight altitude, MSI-derived models could even exceed LiDARderived models, in terms of accuracy. This finding could support the possibility of replacing LiDAR with more cost-effective MSIbased approaches. However, LiDAR remains a viable modality, since a LiDAR-derived 3D model only required a single predictor variable, while an MSI-derived model relied on multiple independent variables in our case.

Potato Leaf Area Index Estimation Using Multi-Sensor Unmanned Aerial Vehicle (UAV) Imagery and Machine Learning

“…From our results, we can find that the richer the spectral information in the data, the higher the prediction accuracy of LAI. Similar findings are present in [58][59][60], showing that the multi-hyperspectral imageryderived model outperforms the LiDAR-derived model. In all four combination experiments, the accuracy of the combination features was higher than the accuracy of their components.…”

Soybean leaf estimation based on RGB images and machine learning methods

“…The findings reveal that, with distribution ranges of 68.14–77.97%, 66.77–79.33%, and 78.74–89.93%, respectively, better predictions of LN, LFW, and LAI were made in terms of ATPA. These benefits of RF models are primarily attributable to the incorporation of multiple ML algorithms (such as bootstrap aggregation and random variable selection), which reduce overfitting and autocorrelation of the input variables [ 42 ] and generally have no negative impact on the model when more input variables are added [ 50 ]. As a result, the RF model has strong generalizability [ 48 ] and the highest prediction accuracy.…”

Soybean leaf estimation based on RGB images and machine learning methods

“…The ndings reveal that, with distribution ranges of 68.14-77.97%, 66.77-79.33%, and 78.74-89.93%, respectively, better predictions of LN, LFW, and LAI were made in terms of ATPA. These bene ts of RF models are primarily attributable to the incorporation of multiple ML algorithms (such as bootstrap aggregation and random variable selection), which reduce over tting and autocorrelation of the input variables [27] and generally have no negative impact on the model when more input variables are added [37]. As a result, the RF model has strong generalizability [35] and the highest prediction accuracy.…”