Evaluation of Orthomosics and Digital Surface Models Derived from Aerial Imagery for Crop Type Mapping

Identification lodging degree of wheat using point cloud data and convolutional neural network

Wheat is one of the important food crops, and it is often subjected to different stresses during its growth. Lodging is a common disaster in filling and maturity for wheat, which not only affects the quality of wheat grains, but also causes severe yield reduction. Assessing the degree of wheat lodging is of great significance for yield estimation, wheat harvesting and agricultural insurance claims. In particular, point cloud data extracted from unmanned aerial vehicle (UAV) images have provided technical support for accurately assessing the degree of wheat lodging. However, it is difficult to process point cloud data due to the cluttered distribution, which limits the wide application of point cloud data. Therefore, a classification method of wheat lodging degree based on dimensionality reduction images from point cloud data was proposed. Firstly, 2D images were obtained from the 3D point cloud data of the UAV images of wheat field, which were generated by dimensionality reduction based on Hotelling transform and point cloud interpolation method. Then three convolutional neural network (CNN) models were used to realize the classification of different lodging degrees of wheat, including AlexNet, VGG16, and MobileNetV2. Finally, the self-built wheat lodging dataset was used to evaluate the classification model, aiming to improve the universality and scalability of the lodging discrimination method. The results showed that based on MobileNetV2, the dimensionality reduction image from point cloud obtained by the method proposed in this paper has achieved good results in identifying the lodging degree of wheat. The F1-Score of the classification model was 96.7% for filling, and 94.6% for maturity. In conclusion, the point cloud dimensionality reduction method proposed in this study could meet the accurate identification of wheat lodging degree at the field scale.

The Evaluation of Spectral Vegetation Indexes and Redundancy Reduction on the Accuracy of Crop Type Detection

Knowledge about crop type distribution is valuable information for effective management of agricultural productivity, food security estimation, and natural resources protection. Algorithms for automatic crop type detection have great potential to positively influence these aspects as well as speed up the process of crop type mapping in larger areas. In the presented study, we used 14 Sentinel-2 images to calculate 12 widely used spectral vegetation indices. Further, to evaluate the effect of reduced dimensionality on the accuracy of crop type mapping, we utilized principal component analysis (PCA). For this purpose, random forest (RF)-supervised classifications were tested for each index separately, as well as for the combinations of various indices and the four initial PCA components. Additionally, for each RF classification feature importance was assessed, which enabled identification of the most relevant period of the year for the differentiation of crop types. We used 34.6% of the ground truth field data to train the classifier and calculate various accuracy measures such as the overall accuracy (OA) or Kappa index. The study showed a high effectiveness of the Modified Chlorophyll Absorption in Reflectance Index (MCARI) (OA = 86%, Kappa = 0.81), Normalized Difference Index 45 (NDI45) (OA = 85%, Kappa = 0.81), and Weighted Difference Vegetation Index (WDVI) (OA = 85%, Kappa = 0.80) in crop type mapping. However, utilization of all of them together did not increase the classification accuracy (OA = 78%, Kappa = 0.72). Additionally, the application of the initial three components of PCA allowed us to achieve an OA of 78% and Kappa of 0.72, which was unfortunately lower than the single-index classification (e.g., based on only NDVI45). This shows that dimensionality reductions did not increase the classification accuracy. Moreover, feature importance from RF indicated that images captured from June and July are the most relevant for differentiating crop types. This shows that this period of the year is crucial to effectively differentiate crop types and should be undeniably used in crop type mapping.

Nitrogen Estimation for Wheat Using UAV-Based and Satellite Multispectral Imagery, Topographic Metrics, Leaf Area Index, Plant Height, Soil Moisture, and Machine Learning Methods

To improve productivity, reduce production costs, and minimize the environmental impacts of agriculture, the advancement of nitrogen (N) fertilizer management methods is needed. The objective of this study is to compare the use of Unmanned Aerial Vehicle (UAV) multispectral imagery and PlanetScope satellite imagery, together with plant height, leaf area index (LAI), soil moisture, and field topographic metrics to predict the canopy nitrogen weight (g/m2) of wheat fields in southwestern Ontario, Canada. Random Forests (RF) and support vector regression (SVR) models, applied to either UAV imagery or satellite imagery, were evaluated for canopy nitrogen weight prediction. The top-performing UAV imagery-based validation model used SVR with seven selected variables (plant height, LAI, four VIs, and the NIR band) with an R2 of 0.80 and an RMSE of 2.62 g/m2. The best satellite imagery-based validation model was RF, which used 17 variables including plant height, LAI, the four PlanetScope bands, and 11 VIs, resulting in an R2 of 0.92 and an RMSE of 1.75 g/m2. The model information can be used to improve field nitrogen predictions for the effective management of N fertilizer.

Understanding Growth Dynamics and Yield Prediction of Sorghum Using High Temporal Resolution UAV Imagery Time Series and Machine Learning

Unmanned aerial vehicles (UAV) carrying multispectral cameras are increasingly being used for high-throughput phenotyping (HTP) of above-ground traits of crops to study genetic diversity, resource use efficiency and responses to abiotic or biotic stresses. There is significant unexplored potential for repeated data collection through a field season to reveal information on the rates of growth and provide predictions of the final yield. Generating such information early in the season would create opportunities for more efficient in-depth phenotyping and germplasm selection. This study tested the use of high-resolution time-series imagery (5 or 10 sampling dates) to understand the relationships between growth dynamics, temporal resolution and end-of-season above-ground biomass (AGB) in 869 diverse accessions of highly productive (mean AGB = 23.4 Mg/Ha), photoperiod sensitive sorghum. Canopy surface height (CSM), ground cover (GC), and five common spectral indices were considered as features of the crop phenotype. Spline curve fitting was used to integrate data from single flights into continuous time courses. Random Forest was used to predict end-of-season AGB from aerial imagery, and to identify the most informative variables driving predictions. Improved prediction of end-of-season AGB (RMSE reduction of 0.24 Mg/Ha) was achieved earlier in the growing season (10 to 20 days) by leveraging early- and mid-season measurement of the rate of change of geometric and spectral features. Early in the season, dynamic traits describing the rates of change of CSM and GC predicted end-of-season AGB best. Late in the season, CSM on a given date was the most influential predictor of end-of-season AGB. The power to predict end-of-season AGB was greatest at 50 days after planting, accounting for 63% of variance across this very diverse germplasm collection with modest error (RMSE 1.8 Mg/ha). End-of-season AGB could be predicted equally well when spline fitting was performed on data collected from five flights versus 10 flights over the growing season. This demonstrates a more valuable and efficient approach to using UAVs for HTP, while also proposing strategies to add further value.

Intra-Season Crop Height Variability at Commercial Farm Scales Using a Fixed-Wing UAV

Monitoring the development of vegetation height through time provides a key indicator of crop health and overall condition. Traditional manual approaches for monitoring crop height are generally time consuming, labor intensive and impractical for large-scale operations. Dynamic crop heights collected through the season allow for the identification of within-field problems at critical stages of the growth cycle, providing a mechanism for remedial action to be taken against end of season yield losses. With advances in unmanned aerial vehicle (UAV) technologies, routine monitoring of height is now feasible at any time throughout the growth cycle. To demonstrate this capability, five digital surface maps (DSM) were reconstructed from high-resolution RGB imagery collected over a field of maize during the course of a single growing season. The UAV retrievals were compared against LiDAR scans for the purpose of evaluating the derived point clouds capacity to capture ground surface variability and spatially variable crop height. A strong correlation was observed between structure-from-motion (SfM) derived heights and pixel-to-pixel comparison against LiDAR scan data for the intra-season bare-ground surface (R2 = 0.77 − 0.99, rRMSE = 0.44% − 0.85%), while there was reasonable agreement between canopy comparisons (R2 = 0.57 − 0.65, rRMSE = 37% − 50%). To examine the effect of resolution on retrieval accuracy and processing time, an evaluation of several ground sampling distances (GSD) was also performed. Our results indicate that a 10 cm resolution retrieval delivers a reliable product that provides a compromise between computational cost and spatial fidelity. Overall, UAV retrievals were able to accurately reproduce the observed spatial variability of crop heights within the maize field through the growing season and provide a valuable source of information with which to inform precision agricultural management in an operational context.

Crop Classification in a Heterogeneous Arable Landscape Using Uncalibrated UAV Data

Land cover maps are indispensable for decision making, monitoring, and management in agricultural areas, but they are often only available after harvesting. To obtain a timely crop map of a small-scale arable landscape in the Swiss Plateau, we acquired uncalibrated, very high-resolution data, with a spatial resolution of 0.05 m and four spectral bands, using a consumer-grade camera on an unmanned aerial vehicle (UAV) in June 2015. We resampled the data to different spatial and spectral resolutions, and evaluated the method using textural features (first order statistics and mathematical morphology), a random forest classifier for best performance, as well as number and size of the structuring elements. Our main findings suggest the overall best performing data consisting of a spatial resolution of 0.5 m, three spectral bands (RGB—red, green, and blue), and five different sizes of the structuring elements. The overall accuracy (OA) for the full set of crop classes based on a pixel-based classification is 66.7%. In case of a merged set of crops, the OA increases by ~7% (74.0%). For an object-based classification based on individual field parcels, the OA increases by ~20% (OA of 86.3% for the full set of crop classes, and 94.6% for the merged set, respectively). We conclude the use of UAV to be most relevant at 0.5 m spatial resolution in heterogeneous arable landscapes when used for crop classification.