Photogrammetric UAV Mapping of Terrain under Dense Coastal Vegetation: An Object-Oriented Classification Ensemble Algorithm for Classification and Terrain Correction

Meng, Xuelian; Shang, Nan; Zhang, Xukai; Li, Chunyan; Zhao, Kaiguang; Qiu, Xiaomin; Weeks, Eddie

doi:10.3390/rs9111187

Cited by 46 publications

(28 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, mapping vegetation height using photogrammetry is challenging in dense coastal wetlands as it is difficult to obtain sufficient numbers of ground points required to calculate vegetation height. For example, errors of up to 80% have been documented in elevation models created for dense cordgrass habitats [41].…”

Section: Discussionmentioning

confidence: 99%

Mapping Coastal Wetland Biomass from High Resolution Unmanned Aerial Vehicle (UAV) Imagery

2019

View full text Add to dashboard Cite

Salt marsh productivity is an important control of resiliency to sea level rise. However, our understanding of how marsh biomass and productivity vary across fine spatial and temporal scales is limited. Remote sensing provides a means for characterizing spatial and temporal variability in marsh aboveground biomass, but most satellite and airborne sensors have limited spatial and/or temporal resolution. Imagery from unmanned aerial vehicles (UAVs) can be used to address this data gap. We combined seasonal field surveys and multispectral UAV imagery collected using a DJI Matrice 100 and Micasense Rededge sensor from the Carpinteria Salt Marsh Reserve in California, USA to develop a method for high-resolution mapping of aboveground saltmarsh biomass. UAV imagery was used to test a suite of vegetation indices in their ability to predict aboveground biomass (AGB). The normalized difference vegetation index (NDVI) provided the strongest correlation to aboveground biomass for each season and when seasonal data were pooled, though seasonal models (e.g., spring, r2 = 0.67; RMSE = 344 g m−2) were more robust than the annual model (r2 = 0.36; RMSE = 496 g m−2). The NDVI aboveground biomass estimation model (AGB = 2428.2 × NDVI + 120.1) was then used to create maps of biomass for each season. Total site-wide aboveground biomass ranged from 147 Mg to 205 Mg and was highest in the spring, with an average of 1222.9 g m−2. Analysis of spatial patterns in AGB demonstrated that AGB was highest in intermediate elevations that ranged from 1.6–1.8 m NAVD88. This UAV-based approach can be used aid the investigation of biomass dynamics in wetlands across a range of spatial scales.

show abstract

Section: Discussionmentioning

confidence: 99%

Mapping Coastal Wetland Biomass from High Resolution Unmanned Aerial Vehicle (UAV) Imagery

2019

View full text Add to dashboard Cite

show abstract

“…The advantage of using patch-based learning for orthophoto classification is sourced from the benefits of spectral and spatial information of the data that can improve the accuracy compared to just using the individual pixels (only spectral information). To understand this parameter and find its suboptimal value, several experiments were conducted with different patch sizes (n = 3, 5, 7,9,11,13 ). The statistical analysis in terms of model accuracy indicates that using larger n yields higher accuracy ( Figure 8).…”

Section: Sensitivitymentioning

confidence: 99%

“…They have tested their algorithm on two datasets, and the results showed the employed methodology to be effective with accuracies of 90% and 96% for the two study areas, respectively. On the other hand, a novel model was presented by Meng et al [9], where they applied OBIA to improve vegetation classification based on aerial photos and global positioning systems. Results illustrated a significant improvement in classification accuracy that increased from 83.98% to 96.12% in overall accuracy and from 0.7806 to 0.947 in the Kappa value.…”

Section: Introductionmentioning

confidence: 99%

Classification of Very High Resolution Aerial Photos Using Spectral-Spatial Convolutional Neural Networks

2018

View full text Add to dashboard Cite

Classification of aerial photographs relying purely on spectral content is a challenging topic in remote sensing. A convolutional neural network (CNN) was developed to classify aerial photographs into seven land cover classes such as building, grassland, dense vegetation, waterbody, barren land, road, and shadow. The classifier utilized spectral and spatial contents of the data to maximize the accuracy of the classification process. CNN was trained from scratch with manually created ground truth samples. The architecture of the network comprised of a single convolution layer of 32 filters and a kernel size of 3 × 3, pooling size of 2 × 2, batch normalization, dropout, and a dense layer with Softmax activation. The design of the architecture and its hyperparameters were selected via sensitivity analysis and validation accuracy. The results showed that the proposed model could be effective for classifying the aerial photographs. The overall accuracy and Kappa coefficient of the best model were 0.973 and 0.967, respectively. In addition, the sensitivity analysis suggested that the use of dropout and batch normalization technique in CNN is essential to improve the generalization performance of the model. The CNN model without the techniques above achieved the worse performance, with an overall accuracy and Kappa of 0.932 and 0.922, respectively. This research shows that CNN-based models are robust for land cover classification using aerial photographs. However, the architecture and hyperparameters of these models should be carefully selected and optimized.

show abstract

“…Therefore, in this research, we reproduced the target wind turbine site using a 3D topography model constructed by photographs taken from an unmanned aerial vehicle (UAV, commonly known as a drone) with a spatial resolution of 1 m in the horizontal direction [15,16] and, by using it as input data into an LES turbulent model, we conducted a high-resolution 3D numerical flow simulation. Two types of data-a digital elevation model (DEM), as the digital expression of topography, and digital surface model (DSM), as the digital expression of ground surface, including trees-can be obtained by the method used in this research, based on UAV imagery.…”

Section: Introductionmentioning

confidence: 99%

Numerical Investigation of Terrain-Induced Turbulence in Complex Terrain Using High-Resolution Elevation Data and Surface Roughness Data Constructed with a Drone

Uchida

2019

Energies

View full text Add to dashboard Cite

Using the method based on unmanned aerial vehicle (UAV) imagery, two kinds of data can be obtained: the digital elevation model (DEM) for the digital expression of terrain, and the digital surface model (DSM) for the digital expression of the surface of the ground, including trees. In this research, a 3D topography model with a horizontal spatial resolution of 1 m was reproduced using DEM. In addition, using the differences between the DEM and DSM data, we were able to obtain further detailed information, such as the heights of trees covering the surface of the ground and their spatial distribution. Therefore, the surface roughness model and the UAV imagery data were directly linked. Based on the above data as input data, a high-resolution 3D numerical flow simulation was conducted. By using the numerical results obtained, we discussed the effect of the existence of surface roughness on the wind speed at the height of the hub of the wind turbine. We also discussed the effect of the differences in the spatial resolution in the horizontal direction of the computational grid on the reproductive precision of terrain-induced turbulence. As a result, the existence and the vortex structure of terrain-induced turbulence occurring near the target wind turbine was clearly revealed. It was shown that a horizontal grid resolution of about 5 m was required to reproduce terrain-induced turbulence formed from topography with an altitude of about 127 m. By the simulation using the surface roughness model, turbulence intensity higher than class A in the International Electrotechnical Commission (IEC) turbulence category was confirmed at the present study site, as well as the measured data.

show abstract

Photogrammetric UAV Mapping of Terrain under Dense Coastal Vegetation: An Object-Oriented Classification Ensemble Algorithm for Classification and Terrain Correction

Cited by 46 publications

References 54 publications

Mapping Coastal Wetland Biomass from High Resolution Unmanned Aerial Vehicle (UAV) Imagery

Mapping Coastal Wetland Biomass from High Resolution Unmanned Aerial Vehicle (UAV) Imagery

Classification of Very High Resolution Aerial Photos Using Spectral-Spatial Convolutional Neural Networks

Numerical Investigation of Terrain-Induced Turbulence in Complex Terrain Using High-Resolution Elevation Data and Surface Roughness Data Constructed with a Drone

Contact Info

Product

Resources

About