Assessing the Performance of UAS-Compatible Multispectral and Hyperspectral Sensors for Soil Organic Carbon Prediction

Crucil, Giacomo; Castaldi, Fabio; Aldana-Jague, Emilien; Wesemael, Bas van; Macdonald, A. J.; Oost, Kristof Van

doi:10.3390/su11071889

Cited by 45 publications

(39 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is noted that data acquired in field conditions could be affected by several limitations such as variable illumination, atmospheric conditions, and sensor sampling distance which could affect the accuracy in using such dataset. However, in the work by the authors of [70], the correlation between datasets obtained under laboratory and outdoor conditions was demonstrated. Thus, a neural network could be trained with laboratory data and validated using remote UAV or airborne data.…”

Section: Discussionmentioning

confidence: 99%

Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data

et al. 2019

View full text Add to dashboard Cite

Terrestrial hydrocarbon spills have the potential to cause significant soil degradation across large areas. Identification and remedial measures taken at an early stage are therefore important. Reflectance spectroscopy is a rapid remote sensing method that has proven capable of characterizing hydrocarbon-contaminated soils. In this paper, we develop a deep learning approach to estimate the amount of Hydrocarbon (HC) mixed with different soil samples using a three-term backpropagation algorithm with dropout. The dropout was used to avoid overfitting and reduce computational complexity. A Hyspex SWIR 384 m camera measured the reflectance of the samples obtained by mixing and homogenizing four different soil types with four different HC substances, respectively. The datasets were fed into the proposed deep learning neural network to quantify the amount of HCs in each dataset. Individual validation of all the dataset shows excellent prediction estimation of the HC content with an average mean square error of ~ 2 . 2 × 10 - 4 . The results with remote sensed data captured by an airborne system validate the approach. This demonstrates that a deep learning approach coupled with hyperspectral imaging techniques can be used for rapid identification and estimation of HCs in soils, which could be useful in estimating the quantity of HC spills at an early stage.

show abstract

Section: Discussionmentioning

confidence: 99%

Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Spectral indices can increase flexibility in the detection approach because they use only a handful of wavelengths and can be easily calculated across platforms as long as the same wavelengths are present (Pontius, 2014). Spectroscopic models that require hundreds of wavelengths can have limited applicability across platforms when sensor measurement characteristics vary (Castaldi et al, 2018; Crucil et al, 2019; Nouri et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

Canopy spectral reflectance detects oak wilt at the landscape scale using phylogenetic discrimination

Sapes

Lapadat

Schweiger

et al. 2021

Preprint

View full text Add to dashboard Cite

The oak wilt disease caused by the invasive fungal pathogen Bretziella fagacearum is one of the greatest threats to oak-dominated forests across the Eastern United States. Accurate detection and monitoring over large areas are necessary for management activities to effectively mitigate and prevent the spread of oak wilt. Canopy spectral reflectance contains both phylogenetic and physiological information across the visible near-infrared (VNIR) and short-wave infrared (SWIR) ranges that can be used to identify diseased red oaks. We develop partial least square discriminant analysis (PLS-DA) models using airborne hyperspectral reflectance to detect canopies at early stages of disease development and assess the importance of VNIR, SWIR, phylogeny, and physiology for oak wilt detection. We achieve high accuracy through a three-step phylogenetic process in which we first distinguish oaks from other species (90% accuracy), then red oaks from white oaks (Quercus macrocarpa) (93% accuracy), and, lastly, infected from non-infected trees (80% accuracy). Including SWIR wavelengths increased model accuracy by ca. 20% relative to models based on VIS-NIR wavelengths alone; using a phylogenetic approach also increased model accuracy by ca. 20% over a single-step classification. SWIR wavelengths include spectral information important in differentiating red oaks from other species and in distinguishing diseased red oaks from healthy red oaks. We determined the most important wavelengths to identify oak species, red oaks, and diseased red oaks. We also demonstrated that several multispectral indices associated with physiological decline can detect differences between healthy and diseased trees. The wavelengths in these indices also tended to be among the most important wavelengths for disease detection within PLS-DA models, indicating a convergence of the methods. Indices were most significant for detecting oak wilt during late August, especially those associated with canopy photosynthetic activity and water status. Our study suggests that coupling phylogenetics, physiology, and canopy spectral reflectance provides an interdisciplinary and comprehensive approach that enables detection of forest diseases at large scales even at early disease stages. These results have potential for direct application by forest managers for early detection to initiate actions to mitigate the disease and prevent pathogen spread.

show abstract

“…These studies have determined that SOC exhibits diagnostic absorption features in VNIR and SWIR that can be used for estimating SOC concentration in soil by exploiting the relationship between reflectance and organic carbon spectral features [7]. Although laboratory-based spectroscopy has resulted in robust and accurate estimates of soil properties, this technique only provides an estimate at the sample point location and geostatistical techniques have to be used to infer continuous spatial information at large scale [8]. The use of hyperspectral remote sensing sensors onboard drone-based platforms (also known as Unmanned Aerial Vehicle, UAV, or Unmanned Aerial Systems, UAS) has introduced new opportunities for providing detailed spatially explicit spectral information of several soil properties, including SOC content [8].…”

Section: Introductionmentioning

confidence: 99%

“…Although laboratory-based spectroscopy has resulted in robust and accurate estimates of soil properties, this technique only provides an estimate at the sample point location and geostatistical techniques have to be used to infer continuous spatial information at large scale [8]. The use of hyperspectral remote sensing sensors onboard drone-based platforms (also known as Unmanned Aerial Vehicle, UAV, or Unmanned Aerial Systems, UAS) has introduced new opportunities for providing detailed spatially explicit spectral information of several soil properties, including SOC content [8]. These hyperspectral sensors have the potential to provide detailed information on the reflectance characteristics of soil in several hundred wavelength bands at the field or even landscape scale.…”

Section: Introductionmentioning

confidence: 99%

“…There are many benefits to such selection: (1) To use the selected bands in the improvement of the fit of statistical models [18,19]. (2) UAVs are quickly becoming the "go-to" platform for airborne hyperspectral remote sensing sensors, but to our knowledge, no previous study has used hyperspectral sensors adapted for onboard UAVs to assess SOC contents in agricultural land and selected specific bands or spectral regions that are closely associated with SOC levels [6,8]. (3) The selection of specific spectral bands or regions may help to clarify the relationship between the spectral information and SOC levels.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Ensemble Identification of Spectral Bands Related to Soil Organic Carbon Levels over an Agricultural Field in Southern Ontario, Canada

et al. 2019

View full text Add to dashboard Cite

The recent use of hyperspectral remote sensing imagery has introduced new opportunities for soil organic carbon (SOC) assessment and monitoring. These data enable monitoring of a wide variety of soil properties but pose important methodological challenges. Highly correlated hyperspectral spectral bands can affect the prediction and accuracy as well as the interpretability of the retrieval model. Therefore, the spectral dimension needs to be reduced through a selection of specific spectral bands or regions that are most helpful to describing SOC. This study evaluates the efficiency of visible near-infrared (VNIR) and shortwave near-infrared (SWIR) hyperspectral data to identify the most informative hyperspectral bands responding to SOC content in agricultural soils. Soil samples (111) were collected over an agricultural field in southern Ontario, Canada and analyzed against two hyperspectral datasets: An airborne Nano-Hyperspec imaging sensor with 270 bands (400–1000 nm) and a laboratory hyperspectral dataset (ASD FieldSpec 3) along the 1000–2500 nm range (NIR-SWIR). In parallel, a multimethod modeling approach consisting of random forest, support vector machine, and partial least squares regression models was used to conduct band selections and to assess the validity of the selected bands. The multimethod model resulted in a selection of optimal band or regions over the VNIR and SWIR sensitive to SOC and potentially for mapping. The bands that achieved the highest respective importance values were 711–715, 727, 986–998, and 433–435 nm regions (VNIR); and 2365–2373, 2481–2500, and 2198–2206 nm (NIR-SWIR). Some of these bands are in agreement with the absorption features of SOC reported in the literature, whereas others have not been reported before. Ultimately, the selection of optimal band and regions is of importance for quantification of agricultural SOC and would provide a new framework for creating optimized SOC-specific sensors.

show abstract

Assessing the Performance of UAS-Compatible Multispectral and Hyperspectral Sensors for Soil Organic Carbon Prediction

Cited by 45 publications

References 47 publications

Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data

Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data

Canopy spectral reflectance detects oak wilt at the landscape scale using phylogenetic discrimination

Ensemble Identification of Spectral Bands Related to Soil Organic Carbon Levels over an Agricultural Field in Southern Ontario, Canada

Contact Info

Product

Resources

About