Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data
<p>Sample preparation of the experiment combining sandy-clay-loam with diesel. Photos show the HC contaminant being increasingly added to the same soil sample until saturated. From left; addition of 5 mL , followed by 10 mL, 15 mL, 20 mL, and 25 mL of the HC.</p> "> Figure 2
<p>Scanning process of the dataset.</p> "> Figure 3
<p>HySpex 384 m line scan acquisition process. The camera (nadir) acquires hyperspectral lines of pixels. The hyperspectral image is obtained by translation of the object under constant illumination.</p> "> Figure 4
<p>(<b>a</b>) A typical network before and (<b>b</b>) after applying dropout (adapted from the work by the authors of [<a href="#B55-remotesensing-11-01938" class="html-bibr">55</a>]).</p> "> Figure 5
<p>Spectral reflectance of different soils and <math display="inline"><semantics> <mrow> <mn>15</mn> <mo>%</mo> </mrow> </semantics></math> hydrocarbon concentration mixtures.Clay mixtures.</p> "> Figure 6
<p>Mean square error of different soils contaminated with different HC contents.</p> "> Figure 7
<p>Neural network estimated output and target output of different soils contaminated with different HC contents.</p> "> Figure 8
<p>Mean square error of the 3 different soils contaminated with different HC contents.</p> ">
Abstract
:1. Introduction
2. Data Acquisition
2.1. Materials
2.2. Sample Preparation
- Each soil type was air-dried, and therefore all samples contained similar levels of moisture.
- Fifty grams of a soil sample type was added to a petri dish (12 cm in diameter)
- The sample was scanned with a Hyspex SWIR 384 m camera under constant illumination.
- In the same sample, initially, 2 mL of the HC were added to the soil using a syringe (to clay and clayloam), which was subsequently changed to 5 mL of the HC to the other soil types.
- A disposable plastic spoon was used to homogenize the mixture and to flatten its surface in order to have even surfaces, except for some soil samples containing clay which tends to be sticky and difficult to flatten due to the characteristics of the soil type, e.g., Figure 1b.
- The sample was scanned with a Hyspex SWIR 384 m camera under constant illumination.
- In the same sample, a further 5 mL of HC was added to the mixture.
- The disposable spoon was used to homogenize the mixture and another scan was taken.
- The procedure was repeated with increments of 5 mL of HCs until the mixture was saturated and formed a shallow local pool (see Figure 1).
2.3. Hyperspectral Imaging
3. Methodology
3.1. Workflow
- Obtaining the dataset via a controlled experiment by mixing and homogenizing different Hydrocarbon (HC) types with soil samples and scanning them with a Hyspex Shortwave Infrared (SWIR) 384 m camera.
- Applying the Deep Learning (DL) model trained using a three-term backpropagation algorithm with dropout for the abundance estimation of the HCs.
- Structuring the DL model with different dropout ratios to determine the most efficient DL setting.
- Testing and validating the performance of the proposed method for abundance estimation of the different HCs by using the same network structure and hyperparameters.
- Comparing the accuracy and performance of the DL model with a hybrid spectral unmixing method [21] and DL models trained using a standard backpropagation algorithm with and without dropout (to prove the generalization ability of dropout), respectively.
3.2. Deep Learning
3.3. Dropout
3.4. Backpropagation
3.5. Three-Term Backpropagation
Algorithm 1: Learning method using the three-term backpropagation with dropout used in training the DNN model. |
Data: , , , , , e DNN weights, , are randomly initialized ... ... ... initialize the learning rate, ; momentum factor, ; and proportional factor, |
3.6. Hyperparameters
3.7. Architecture of the Deep Learning Model
4. Results
4.1. Experiment with Laboratory Data
4.2. Soil Continuity Experiments
4.3. Experiment with Remote Sensed Data
5. Discussion
6. Conclusions
Author Contributions
Acknowledgments
Conflicts of Interest
References
- United Nations Environment Programme (UNEP). Environmental Assessment of Ogoniland; United Nations Environment Programme: Nairobi, Kenya, 2011; pp. 1–262. [Google Scholar]
- Raksuntorn, N.; Du, Q. A new linear mixture model for hyperspectral image analysis. In Proceedings of the 2008 IEEE International Geoscience and Remote Sensing Symposium, Boston, MA, USA, 7–11 July 2008; Volume 3, pp. III-258–III-261. [Google Scholar]
- Eweje, G. Environmental costs and responsibilities resulting from oil exploitation in developing countries: The case of the Niger Delta of Nigeria. J. Bus. Ethics 2006, 69, 27–56. [Google Scholar] [CrossRef]
- Latimer, J.S.; Zheng, J. The sources, transport, and fate of PAHs in the marine environment. PAHs Ecotoxicol. Perspect. 2003, 9. [Google Scholar] [CrossRef]
- Okparanma, R.N.; Mouazen, A.M. Determination of total petroleum hydrocarbon (TPH) and polycyclic aromatic hydrocarbon (PAH) in soils: A review of spectroscopic and nonspectroscopic techniques. Appl. Spectrosc. Rev. 2013, 48, 458–486. [Google Scholar] [CrossRef]
- Scafutto, R.D.M.; de Souza Filho, C.R.; de Oliveira, W.J. Hyperspectral remote sensing detection of petroleum hydrocarbons in mixtures with mineral substrates: Implications for onshore exploration and monitoring. ISPRS J. Photogramm. Remote Sens. 2017, 128, 146–157. [Google Scholar] [CrossRef]
- Salem, F.; Kafatos, M. Hyperspectral image analysis for oil spill mitigation. Paper presented at the 22nd Asian Conference on Remote Sensing, Taipei, Taiwan, 3–7 October 2011; Volume 5, p. 9. [Google Scholar]
- Okparanma, R.N.; Mouazen, A.M. Visible and near-infrared spectroscopy analysis of a polycyclic aromatic hydrocarbon in soils. Sci. World J. 2013, 2013, 160360. [Google Scholar] [CrossRef]
- Chakraborty, S.; Weindorf, D.C.; Li, B.; Ali, M.N.; Majumdar, K.; Ray, D. Analysis of petroleum contaminated soils by spectral modeling and pure response profile recovery of n-hexane. Environ. Pollut. 2014, 190, 10–18. [Google Scholar] [CrossRef]
- Lammoglia, T.; de Souza Filho, C.R. Spectroscopic characterization of oils yielded from Brazilian offshore basins: Potential applications of remote sensing. Remote Sens. Environ. 2011, 115, 2525–2535. [Google Scholar] [CrossRef]
- Fingas, M.; Brown, C. Oil spill remote sensing. In Earth System Monitoring; Springer: Berlin/Heidelberg, Germany, 2013; pp. 337–388. [Google Scholar]
- Clasen, A.; Somers, B.; Pipkins, K.; Tits, L.; Segl, K.; Brell, M.; Kleinschmit, B.; Spengler, D.; Lausch, A.; Förster, M. Spectral unmixing of forest crown components at close range, airborne and simulated Sentinel-2 and EnMAP spectral imaging scale. Remote Sens. 2015, 7, 15361–15387. [Google Scholar] [CrossRef]
- Scafutto, R.D.M.; de Souza Filho, C.R. Quantitative characterization of crude oils and fuels in mineral substrates using reflectance spectroscopy: Implications for remote sensing. Int. J. Appl. Earth Obs. Geoinf. 2016, 50, 221–242. [Google Scholar] [CrossRef]
- Pelta, R.; Ben-Dor, E. Assessing the detection limit of petroleum hydrocarbon in soils using hyperspectral remote-sensing. Remote Sens. Environ. 2019, 224, 145–153. [Google Scholar] [CrossRef]
- Webster, G.T.; Soriano-Disla, J.M.; Kirk, J.; Janik, L.J.; Forrester, S.T.; McLaughlin, M.J.; Stewart, R.J. Rapid prediction of total petroleum hydrocarbons in soil using a hand-held mid-infrared field instrument. Talanta 2016, 160, 410–416. [Google Scholar] [CrossRef]
- Pelta, R.; Ben-Dor, E. An Exploratory Study on the Effect of Petroleum Hydrocarbon on Soils Using Hyperspectral Longwave Infrared Imagery. Remote Sens. 2019, 11, 569. [Google Scholar] [CrossRef]
- Okparanma, R.N.; Coulon, F.; Mayr, T.; Mouazen, A.M. Mapping polycyclic aromatic hydrocarbon and total toxicity equivalent soil concentrations by visible and near-infrared spectroscopy. Environ. Pollut. 2014, 192, 162–170. [Google Scholar] [CrossRef] [Green Version]
- Schwartz, G.; Ben-Dor, E.; Eshel, G. Quantitative analysis of total petroleum hydrocarbons in soils: Comparison between reflectance spectroscopy and solvent extraction by 3 certified laboratories. Appl. Environ. Soil Sci. 2012, 2012, 751956. [Google Scholar] [CrossRef]
- Gholizadeh, A.; Saberioon, M.; Ben-Dor, E.; Boruuvka, L. Monitoring of selected soil contaminants using proximal and remote sensing techniques: Background, state-of-the-art and future perspectives. Crit. Rev. Environ. Sci. Technol. 2018, 48, 243–278. [Google Scholar] [CrossRef]
- Pabón, R.E.C.; de Souza Filho, C.R. Spectroscopic characterization of red latosols contaminated by petroleum-hydrocarbon and empirical model to estimate pollutant content and type. Remote Sens. Environ. 2016, 175, 323–336. [Google Scholar] [CrossRef]
- Ahmed, A.M.; Duran, O.; Zweiri, Y.; Smith, M. Hybrid spectral unmixing: Using artificial neural networks for linear/nonlinear switching. Remote Sens. 2017, 9, 775. [Google Scholar] [CrossRef]
- Xu, M.; Zhang, L.; Du, B.; Zhang, L.; Fan, Y.; Song, D. A mutation operator accelerated quantum-behaved particle swarm optimization algorithm for hyperspectral endmember extraction. Remote Sens. 2017, 9, 197. [Google Scholar] [CrossRef]
- Uezato, T.; Murphy, R.J.; Melkumyan, A.; Chlingaryan, A. A novel spectral unmixing method incorporating spectral variability within endmember classes. IEEE Trans. Geosci. Remote Sens. 2016, 54, 2812–2831. [Google Scholar] [CrossRef]
- Weeks, A.R. Fundamentals of Electronic Image Processing; SPIE Optical Engineering Press: Bellingham, WA, USA, 1996. [Google Scholar]
- Keshava, N.; Mustard, J.F. Spectral unmixing. IEEE Signal Process. Mag. 2002, 19, 44–57. [Google Scholar] [CrossRef]
- Drumetz, L.; Tochon, G.; Chanussot, J.; Jutten, C. Estimating the number of endmembers to use in spectral unmixing of hyperspectral data with collaborative sparsity. In Proceedings of the Submitted to the 13th International Conference on Latent Variable Analysis and Signal Separation (LVA-ICA), Grenoble, France, 21–23 February 2017; pp. 1–10. [Google Scholar]
- Hapke, B. Bidirectional reflectance spectroscopy: 1. Theory. J. Geophys. Res. Solid Earth 1981, 86, 3039–3054. [Google Scholar] [CrossRef]
- Dobigeon, N.; Tourneret, J.Y.; Richard, C.; Bermudez, J.; McLaughlin, S.; Hero, A.O. Nonlinear unmixing of hyperspectral images: Models and algorithms. Signal Process. Mag. IEEE 2014, 31, 82–94. [Google Scholar] [CrossRef]
- Halimi, A.; Altmann, Y.; Buller, G.S.; McLaughlin, S.; Oxford, W.; Clarke, D.; Piper, J. Robust unmixing algorithms for hyperspectral imagery. In Proceedings of the Sensor Signal Processing for Defence (SSPD), Edinburgh, UK, 22–23 September 2016; pp. 1–5. [Google Scholar]
- Liu, J.; Luo, B.D.S.C.J. Exploration of planetary hyperspectral images with unsupervised spectral unmixing: A case study of planet Mars. Remote Sens. 2018, 10, 737. [Google Scholar] [CrossRef]
- Zhang, X.; Li, C.; Zhang, J.; Chen, Q.; Feng, J.; Jiao, L.; Zhou, H. Hyperspectral Unmixing via Low-Rank Representation with Space Consistency Constraint and Spectral Library Pruning. Remote Sens. 2018, 10, 339. [Google Scholar] [CrossRef]
- Kang, S.; Xiurui, G.; Hairong, T. A new target detection method using nonlinear PCA for hyperspectral imagery. Bull. Surv. Mapp. 2015, 1, 105–108. [Google Scholar]
- Feng, D.C.; Chen, F.; Wen-Li, X. Detecting local manifold structure for unsupervised feature selection. Acta Autom. Sin. 2014, 40, 2253–2261. [Google Scholar] [CrossRef]
- Ahmed, A.; Duran, O.; Zweiri, Y.; Smith, M. Application of hybrid switch method to quantify oil spills. In Proceedings of the Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS), Amsterdam, The Netherlands, 23–26 September 2018. [Google Scholar]
- Iordache, M.D.; Bioucas-Dias, J.M.; Plaza, A. Sparse unmixing of hyperspectral data. IEEE Trans. Geosci. Remote Sens. 2011, 49, 2014–2039. [Google Scholar] [CrossRef]
- Feng, J.; Rogge, D.; Rivard, B. Comparison of lithological mapping results from airborne hyperspectral VNIR-SWIR, LWIR and combined data. Int. J. Appl. Earth Obs. Geoinf. 2018, 64, 340–353. [Google Scholar] [CrossRef]
- Harris, J.; Rogge, D.; Hitchcock, R.; Ijewliw, O.; Wright, D. Mapping lithology in Canada’s Arctic: application of hyperspectral data using the minimum noise fraction transformation and matched filtering. Can. J. Earth Sci. 2005, 42, 2173–2193. [Google Scholar] [CrossRef]
- Rogge, D.; Rivard, B.; Segl, K.; Grant, B.; Feng, J. Mapping of NiCu–PGE ore hosting ultramafic rocks using airborne and simulated EnMAP hyperspectral imagery, Nunavik, Canada. Remote Sens. Environ. 2014, 152, 302–317. [Google Scholar] [CrossRef]
- Khamehchiyan, M.; Charkhabi, A.H.; Tajik, M. Effects of crude oil contamination on geotechnical properties of clayey and sandy soils. Eng. Geol. 2007, 89, 220–229. [Google Scholar] [CrossRef]
- Soil Survey Division Staff. Soil Survey Manual; Number 18; Government Printing Office: Washington, DC, USA, 1993.
- Mathieu, M.; Roy, R.; Launeau, P.; Cathelineau, M.; Quirt, D. Alteration mapping on drill cores using a HySpex SWIR-320m hyperspectral camera: Application to the exploration of an unconformity-related uranium deposit (Saskatchewan, Canada). J. Geochem. Explor. 2017, 172, 71–88. [Google Scholar] [CrossRef]
- Guo, Y.; Liu, Y.; Oerlemans, A.; Lao, S.; Wu, S.; Lew, M.S. Deep learning for visual understanding: A review. Neurocomputing 2016, 187, 27–48. [Google Scholar] [CrossRef]
- Chen, Y.; Lin, Z.; Zhao, X.; Wang, G.; Gu, Y. Deep learning-based classification of hyperspectral data. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 2014, 7, 2094–2107. [Google Scholar] [CrossRef]
- LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436–444. [Google Scholar] [CrossRef]
- Ayrey, E.; Hayes, D.J. The Use of Three-Dimensional Convolutional Neural Networks to Interpret LiDAR for Forest Inventory. Remote Sens. 2018, 10, 649. [Google Scholar] [CrossRef]
- Kim, P. Deep Learning. In MATLAB Deep Learning: With Machine Learning, Neural Networks and Artificial Intelligence; Springer: New York, NY, USA, 2017; pp. 103–120. [Google Scholar]
- Gallego, A.J.; Pertusa, A.; Gil, P. Automatic Ship Classification from Optical Aerial Images with Convolutional Neural Networks. Remote Sens. 2018, 10, 511. [Google Scholar] [CrossRef]
- Jin, X.; Jie, L.; Wang, S.; Qi, H.J.; Li, S.W. Classifying Wheat Hyperspectral Pixels of Healthy Heads and Fusarium Head Blight Disease Using a Deep Neural Network in the Wild Field. Remote Sens. 2018, 10, 395. [Google Scholar] [CrossRef]
- Li, J.; Xi, B.; Li, Y.; Du, Q.; Wang, K. Hyperspectral Classification Based on Texture Feature Enhancement and Deep Belief Networks. Remote Sens. 2018, 10, 396. [Google Scholar] [CrossRef]
- Xu, Y.; Wu, L.; Xie, Z.; Chen, Z. Building extraction in very high resolution remote sensing imagery using deep learning and guided filters. Remote Sens. 2018, 10, 144. [Google Scholar] [CrossRef]
- Mei, S.; Yuan, X.; Ji, J.; Zhang, Y.; Wan, S.; Du, Q. Hyperspectral image spatial super-resolution via 3D full convolutional neural network. Remote Sens. 2017, 9, 1139. [Google Scholar] [CrossRef]
- Huang, Z.; Pan, Z.; Lei, B. Transfer learning with deep convolutional neural network for SAR target classification with limited labeled data. Remote Sens. 2017, 9, 907. [Google Scholar] [CrossRef]
- Wu, H.; Prasad, S. Convolutional recurrent neural networks for hyperspectral data classification. Remote Sens. 2017, 9, 298. [Google Scholar] [CrossRef]
- Fu, G.; Liu, C.; Zhou, R.; Sun, T.; Zhang, Q. Classification for high resolution remote sensing imagery using a fully convolutional network. Remote Sens. 2017, 9, 498. [Google Scholar] [CrossRef]
- Srivastava, N.; Hinton, G.; Krizhevsky, A.; Sutskever, I.; Salakhutdinov, R. Dropout: A simple way to prevent neural networks from overfitting. J. Mach. Learn. Res. 2014, 15, 1929–1958. [Google Scholar]
- Brownlee, J. Deep Learning with Python: Develop Deep Learning Models on Theano and TensorFlow Using Keras; Machine Learning Mastery: Vermont Victoria, Australia, 2016. [Google Scholar]
- Zweiri, Y.H.; Whidborne, J.F.; Althoefer, K.; Seneviratne, L.D. A new three-term backpropagation algorithm with convergence analysis. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA’02), Washington, DC, USA, 11–15 May 2002; Volume 4, pp. 3882–3887. [Google Scholar]
- Zweiri, Y.H.; Seneviratne, L.D.; Althoefer, K. Stability analysis of a three-term backpropagation algorithm. Neural Netw. 2005, 18, 1341–1347. [Google Scholar] [CrossRef]
- Zweiri, Y.H.; Whidborne, J.F.; Seneviratne, L.D. Optimization and stability of a three-term backpropagation algorithm. In Proceedings of the International Conference of Neural Networks (CI’2000), Como, Italy, 23–27 July 2000. [Google Scholar]
- Dahl, G.E.; Sainath, T.N.; Hinton, G.E. Improving deep neural networks for LVCSR using rectified linear units and dropout. In Proceedings of the 2013 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Vancouver, BC, Canada, 26–31 May 2013; pp. 8609–8613. [Google Scholar]
- Maas, A.L.; Hannun, A.Y.; Ng, A.Y. Rectifier nonlinearities improve neural network acoustic models. In Proceedings of the International Conference on Machine Learning (ICML 2013), Atlanta, GA, USA, 16–21 June 2013; Volume 30, p. 3. [Google Scholar]
- Riedmiller, M.; Braun, H. A direct adaptive method for faster backpropagation learning: The RPROP algorithm. In Proceedings of the IEEE International Conference on Neural Networks, San Francisco, CA, USA, 28 March–1 April 1993; pp. 586–591. [Google Scholar]
- Abdulkadir, S.J.; Shamsuddin, S.M.; Sallehuddin, R. Three term back propagation network for moisture prediction. In Proceedings of the International Conference on Clean and Green Energy, Hong Kong, China, 5 January 2012; pp. 103–107. [Google Scholar]
- Arlot, S.; Celisse, A. A survey of cross-validation procedures for model selection. Stat. Surv. 2010, 4, 40–79. [Google Scholar] [CrossRef]
- Nascimento, J.M.; Dias, J.M.B. Vertex component analysis: A fast algorithm to unmix hyperspectral data. IEEE Trans. Geosci. Remote Sens. 2005, 43, 898–910. [Google Scholar] [CrossRef]
- Heinz, D.C. Fully constrained least squares linear spectral mixture analysis method for material quantification in hyperspectral imagery. IEEE Trans. Geosci. Remote Sens. 2001, 39, 529–545. [Google Scholar] [CrossRef] [Green Version]
- Altmann, Y. Nonlinear spectral unmixing of hyperspectral images. Ph.D. Thesis, l’Institut National Polytechnique de Toulouse (INP Toulouse), Toulouse, France, 2013. [Google Scholar]
- Altmann, Y.; Halimi, A.; Dobigeon, N.; Tourneret, J.Y. Supervised nonlinear spectral unmixing using a postnonlinear mixing model for hyperspectral imagery. IEEE Trans. Image Process. 2012, 21, 3017–3025. [Google Scholar] [CrossRef]
- Krizhevsky, A.; Sutskever, I.; Hinton, G.E. Imagenet classification with deep convolutional neural networks. In Proceedings of the Advances in Neural Information Processing Systems, Lake Tahoe, CA, USA, 3–6 December 2012; pp. 1097–1105. [Google Scholar]
- Crucil, G.; Castaldi, F.; Aldana-Jague, E.; van Wesemael, B.; Macdonald, A.; Van Oost, K. Assessing the Performance of UAS-Compatible Multispectral and Hyperspectral Sensors for Soil Organic Carbon Prediction. Sustainability 2019, 11, 1889. [Google Scholar] [CrossRef]
Sample Combination | HC (mL) | Soil (gr) | Sample Combination | HC (mL) | Soil (gr) |
---|---|---|---|---|---|
Clay - Diesel 0 | 0 | 50 | Clay - Bio- diesel 0 | 0 | 50 |
Clay - Diesel 1 | 2 | 50 | Clay - Bio- diesel 1 | 2 | 50 |
Clay - Diesel 2 | 4 | 50 | Clay - Bio- diesel 2 | 4 | 50 |
Clay - Diesel 3 | 5 | 50 | Clay - Bio- diesel 3 | 5 | 50 |
Clay - Diesel 4 | 10 | 50 | Clay - Bio- diesel 4 | 10 | 50 |
Clay - Diesel 5 | 15 | 50 | Clay - Bio- diesel 5 | 15 | 50 |
Clay - Diesel 6 | 20 | 50 | Clay - Bio- diesel 6 | 20 | 50 |
Clay - Diesel 7 | 25 | 50 | Clay - Bio- diesel 7 | 25 | 50 |
Clay - Ethanol 0 | 0 | 50 | Clay Loam - Ethanol 0 | 0 | 50 |
Clay - Ethanol 1 | 2 | 50 | Clay Loam - Ethanol 1 | 2 | 50 |
Clay - Ethanol 2 | 4 | 50 | Clay Loam - Ethanol 2 | 4 | 50 |
Clay - Ethanol 3 | 5 | 50 | Clay Loam - Ethanol 3 | 5 | 50 |
Clay - Ethanol 4 | 10 | 50 | Clay Loam - Ethanol 4 | 10 | 50 |
Clay - Ethanol 5 | 15 | 50 | Clay Loam - Ethanol 5 | 15 | 50 |
Clay - Ethanol 6 | 20 | 50 | Clay Loam - Ethanol 6 | 20 | 50 |
Clay - Ethanol 7 | 25 | 50 | Clay Loam - Ethanol 7 | 25 | 50 |
Clay Loam - Diesel 0 | 0 | 50 | Clay Loam - Bio- diesel 0 | 0 | 50 |
Clay Loam - Diesel 1 | 2 | 50 | Clay Loam - Bio- diesel 1 | 2 | 50 |
Clay Loam - Diesel 2 | 4 | 50 | Clay Loam - Bio- diesel 2 | 4 | 50 |
Clay Loam - Diesel 3 | 5 | 50 | Clay Loam - Bio- diesel 3 | 5 | 50 |
Clay Loam - Diesel 4 | 10 | 50 | Clay Loam - Bio- diesel 4 | 10 | 50 |
Clay Loam - Diesel 5 | 15 | 50 | Clay Loam - Bio- diesel 5 | 15 | 50 |
Clay Loam - Diesel 6 | 20 | 50 | Clay Loam - Bio- diesel 6 | 20 | 50 |
Clay Loam - Petrol 0 | 0 | 50 | |||
Clay Loam - Petrol 1 | 2 | 50 | |||
Clay Loam - Petrol 2 | 4 | 50 | Sand Loam - Petrol 0 | 0 | 50 |
Clay Loam - Petrol 3 | 5 | 50 | Sand Loam - Petrol 1 | 5 | 50 |
Clay Loam - Petrol 4 | 10 | 50 | Sand Loam - Petrol 2 | 10 | 50 |
Clay Loam - Petrol 5 | 15 | 50 | Sand Loam - Petrol 3 | 15 | 50 |
Clay Loam - Petrol 6 | 20 | 50 | Sand Loam - Petrol 4 | 20 | 50 |
Clay Loam - Petrol 7 | 25 | 50 | Sand Loam - Petrol 5 | 25 | 50 |
Clay Loam - Petrol 8 | 30 | 50 | Sand Loam - Petrol 6 | 30 | 50 |
Clay Loam - Petrol 9 | 35 | 50 | Sand Loam - Petrol 7 | 35 | 50 |
Clay Loam - Petrol 10 | 40 | 50 | Sand Loam - Petrol 8 | 40 | 50 |
Clay Loam - Petrol 11 | 45 | 50 | Sand Loam - Petrol 9 | 45 | 50 |
Sand Clay Loam - Diesel 0 | 0 | 50 | Sand Clay Loam - Bio- diesel 0 | 0 | 50 |
Sand Clay Loam - Diesel 1 | 5 | 50 | Sand Clay Loam - Bio- diesel 1 | 5 | 50 |
Sand Clay Loam - Diesel 2 | 10 | 50 | Sand Clay Loam - Bio- diesel 2 | 10 | 50 |
Sand Clay Loam - Diesel 3 | 15 | 50 | Sand Clay Loam - Bio- diesel 3 | 15 | 50 |
Sand Clay Loam - Diesel 4 | 20 | 50 | Sand Clay Loam - Bio- diesel 4 | 20 | 50 |
Sand Clay Loam - Diesel 5 | 25 | 50 | Sand Clay Loam - Bio- diesel 5 | 25 | 50 |
Sand Clay Loam - Ethanol 0 | 0 | 50 | Sand Clay Loam - Petrol 0 | 0 | 50 |
Sand Clay Loam - Ethanol 1 | 5 | 50 | Sand Clay Loam - Petrol 1 | 5 | 50 |
Sand Clay Loam - Ethanol 2 | 10 | 50 | Sand Clay Loam - Petrol 2 | 10 | 50 |
Sand Clay Loam - Ethanol 3 | 15 | 50 | Sand Clay Loam - Petrol 3 | 15 | 50 |
Sand Clay Loam - Ethanol 4 | 20 | 50 | Sand Clay Loam - Petrol 4 | 20 | 50 |
Sand Clay Loam - Ethanol 5 | 25 | 50 | Sand Clay Loam - Petrol 5 | 25 | 50 |
Sand Clay Loam - Ethanol 6 | 30 | 50 | Sand Clay Loam - Petrol 6 | 30 | 50 |
Sand Clay Loam - Petrol 7 | 35 | 50 | |||
Sand Loam - Diesel 0 | 0 | 50 | Sand Loam - Bio- diesel 0 | 0 | 50 |
Sand Loam - Diesel 1 | 5 | 50 | Sand Loam - Bio- diesel 1 | 5 | 50 |
Sand Loam - Diesel 2 | 10 | 50 | Sand Loam - Bio- diesel 2 | 10 | 50 |
Sand Loam - Diesel 3 | 15 | 50 | Sand Loam - Bio- diesel 3 | 15 | 50 |
Sand Loam - Diesel 4 | 20 | 50 | Sand Loam - Bio- diesel 4 | 20 | 50 |
Sand Loam - Ethanol 0 | 0 | 50 | |||
Sand Loam - Ethanol 1 | 5 | 50 | |||
Sand Loam - Ethanol 2 | 10 | 50 | |||
Sand Loam - Ethanol 3 | 15 | 50 | |||
Sand Loam - Ethanol 4 | 20 | 50 |
Specification | HySpex SWIR-384 m |
---|---|
Spectral Range (nm) | 930–2500 |
Spatial Pixels (pixels) | 384 |
Spectral Channels | 288 |
Spectral Sampling (nm) | 5.45 |
FOV (degrees) | 16° |
Pixel FOV across/along (mrad) | 0.73/0.73 |
Bit resolution (raw data)/Digitization | 16 |
Noise floor () | 150 |
Dynamic range | 7500 |
Peak SNR (at full resolution) | >1100 |
Max speed (at full resolution)(fps) | 400 |
Full Width Half Maximum | ∼1 pixel |
Power consumption (W) | 30 |
Dimensions (l-w-h) (cm) | 38-12-17.5 |
Weight (kg) | 5.7 |
Dataset | Size | Number of Mixtures |
---|---|---|
Clay biodiesel | 8 | |
Clay diesel | 8 | |
Clay ethanol | 8 | |
Clay loam biodiesel | 7 | |
Clay loam diesel | 7 | |
Clay loam ethanol | 8 | |
Clay loam petrol | 12 | |
Sandy loam biodiesel | 5 | |
Sandy loam diesel | 5 | |
Sandy loam ethanol | 5 | |
Sandy loam petrol | 10 | |
Sandy clay loam biodiesel | 6 | |
Sandy clay loam diesel | 6 | |
Sandy clay loam ethanol | 7 | |
Sandy clay loam petrol | 8 |
Corresponding mixtures (mL) | Petrol | Diesel | Biodiesel | Ethanol |
---|---|---|---|---|
2.0 | 0.02 | 0.023 | 0.034 | 0.029 |
4.0 | 0.055 | 0.063 | 0.065 | 0.059 |
5.0 | 0.068 | 0.08 | 0.08 | 0.073 |
10.0 | 0.128 | 0.148 | 0.149 | 0.136 |
15.0 | 0.181 | 0.206 | 0.208 | 0.191 |
20.0 | 0.227 | 0.258 | 0.260 | 0.240 |
25.0 | 0.269 | 0.303 | 0.305 | 0.283 |
30.0 | 0.340 | 0.342 | 0.345 | 0.321 |
35.0 | 38.1 | – | – | – |
40.0 | 42.0 | – | – | – |
45.0 | 46.2 | – | – | – |
Dataset | Test Set with Dropout | Test Set without Dropout |
---|---|---|
Clay biodiesel | ||
Clay diesel | ||
Clay ethanol | ||
Clay loam biodiesel | ||
Clay loam diesel | ||
Clay loam ethanol | ||
Clay loam petrol | ||
Sandy loam biodiesel | ||
Sandy loam diesel | ||
Sandy loam ethanol | ||
Sandy loam petrol | ||
Sandy clay loam biodiesel | ||
Sandy clay loam diesel | ||
Sandy clay loam ethanol | ||
Sandy clay loam petrol |
HC Types | DO 10% | DO 20% | DO 30% | DO 40% | DO 50% |
---|---|---|---|---|---|
Bio-diesel | |||||
MSE | |||||
Diesel | |||||
MSE | |||||
Ethanol | |||||
MSE | |||||
Petrol | |||||
MSE |
HC Types | DO 10% | DO 20% | DO 30% | DO 40% | DO 50% |
---|---|---|---|---|---|
Bio-diesel | |||||
MSE | |||||
Diesel | |||||
MSE | |||||
Ethanol | |||||
MSE |
HC Types | DO 10% | DO 20% | DO 30% | DO 40% | DO 50% |
---|---|---|---|---|---|
Bio-diesel | |||||
MSE | |||||
Diesel | |||||
MSE | |||||
Ethanol | |||||
MSE | |||||
Petrol | |||||
MSE |
HC Types | DO 10% | DO 20% | DO 30% | DO 40% | DO 50% |
---|---|---|---|---|---|
Bio-diesel | |||||
MSE | |||||
Diesel | |||||
MSE | |||||
Ethanol | |||||
MSE | |||||
Petrol | |||||
MSE |
Mixtures | Reference | Proposed Method | Hybrid Switch Method | Conventionally Trained NN with DO (0.2) | Conventionally Trained NN |
---|---|---|---|---|---|
Clay–biodiesel | 0 | 0.002 | 0.03 | 0.61 | 0.69 |
Clay–biodiesel | 3 | 3.4 | 3.7 | 3.9 | 4.5 |
Clay–biodiesel | 6 | 6.5 | 6.9 | 7.3 | 7.9 |
Clay–biodiesel | 8 | 8.4 | 9.4 | 9.8 | 9.9 |
Clay–biodiesel | 14.9 | 15.3 | 16.9 | 18.0 | 18.9 |
Clay–biodiesel | 20.8 | 21.4 | 22.6 | 17.3 | 17.9 |
Clay–biodiesel | 26.0 | 26.8 | 28.3 | 29.9 | 30.3 |
Clay–biodiesel | 30.5 | 31.9 | 32.2 | 35.1 | 38.6 |
Average error (%) | 2 | 10 | 17 | 20 | |
Clay–diesel | 0 | 0.003 | 0.04 | 1.06 | 1.79 |
Clay–diesel | 3 | 3.3 | 3.7 | 4.6 | 4.9 |
Clay–diesel | 6 | 6.5 | 6.3 | 6.9 | 7.21 |
Clay–diesel | 8 | 8.1 | 7.6 | 4.8 | 3.7 |
Clay–diesel | 14.8 | 15.3 | 13.0 | 19.1 | 19.8 |
Clay–diesel | 20.6 | 21.4 | 18.5 | 23.8 | 25.5 |
Clay–diesel | 25.8 | 25.6 | 23.6 | 29.1 | 29.4 |
Clay–diesel | 30.3 | 30.6 | 32.2 | 36.1 | 37.7 |
Average error (%) | 2 | 14 | 20 | 24 | |
Clay–ethanol | 0 | 0.004 | 0.05 | 1.41 | 2.04 |
Clay–ethanol | 2 | 2.7 | 2.3 | 3.4 | 3.9 |
Clay–ethanol | 5 | 5.5 | 5.3 | 6.01 | 6.63 |
Clay–ethanol | 7.3 | 7.6 | 8.2 | 4.9 | 4.1 |
Clay–ethanol | 13.6 | 14.1 | 15.6 | 10.4 | 9.9 |
Clay–ethanol | 19.1 | 19.6 | 18.1 | 22.7 | 22.9 |
Clay–ethanol | 24 | 24.9 | 22.9 | 27.7 | 28.3 |
Clay–ethanol | 28.3 | 28.9 | 27.7 | 32.8 | 33.7 |
Average error (%) | 3 | 9 | 19 | 23 | |
CL–biodiesel | 0 | 0.002 | 0.16 | 0.90 | 1.63 |
CL–biodiesel | 3 | 3.2 | 3.6 | 4.5 | 4.9 |
CL–biodiesel | 6 | 6.4 | 6.6 | 7.1 | 7.8 |
CL-biodiesel | 8 | 8.5 | 7.7 | 9.9 | 10.8 |
CL-biodiesel | 14.9 | 15.5 | 13.6 | 17.9 | 18.3 |
CL-biodiesel | 20.8 | 21.6 | 18.7 | 24.6 | 24.9 |
CL-biodiesel | 26.0 | 26.4 | 24.1 | 29.6 | 29.9 |
Average error (%) | 2 | 11 | 17 | 21 | |
CL-diesel | 0 | 0.003 | 0.001 | 0.76 | 1.88 |
CL-diesel | 3 | 3.3 | 3.1 | 3.7 | 4.3 |
CL-diesel | 6 | 5.8 | 6.6 | 7.2 | 7.8 |
CL-diesel | 8 | 8.4 | 10.4 | 11.6 | 11.9 |
CL-diesel | 14.8 | 14.3 | 12.8 | 18.7 | 19.0 |
CL-diesel | 20.6 | 21.1 | 22.1 | 24.4 | 25.3 |
CL-diesel | 25.8 | 26.8 | 27.7 | 28.2 | 30.7 |
Average error (%) | 3 | 13 | 26 | 31 | |
CL-ethanol | 0 | 0.002 | 0.05 | 1.22 | 1.79 |
CL-ethanol | 2 | 2.4 | 2.7 | 3.6 | 3.9 |
CL-ethanol | 5 | 5.2 | 5.4 | 5.9 | 6.4 |
CL-ethanol | 7.3 | 7.6 | 6.7 | 9.6 | 9.9 |
CL-ethanol | 13.6 | 14.4 | 11.6 | 16.7 | 17.0 |
CL-ethanol | 19.1 | 19.7 | 18.5 | 22.6 | 22.9 |
CL-ethanol | 24 | 24.6 | 25.9 | 26.7 | 27.5 |
CL-ethanol | 28.3 | 28.7 | 29.9 | 30.7 | 31.2 |
Average error (%) | 2 | 11 | 17 | 21 | |
CL-petrol | 0 | 0.002 | 0.006 | 0.36 | 1.40 |
CL-petrol | 2 | 2.4 | 2.2 | 2.9 | 3.3 |
CL-petrol | 5 | 5.1 | 5.5 | 6.3 | 6.9 |
CL-petrol | 6.8 | 5.9 | 7.9 | 8.4 | 8.9 |
CL-petrol | 12.8 | 12.9 | 13.0 | 11.0 | 9.9 |
CL-petrol | 18.1 | 18.4 | 17.4 | 16.8 | 15.6 |
CL-petrol | 22.7 | 23.1 | 24.4 | 19.8 | 18.5 |
CL-petrol | 26.9 | 27.4 | 27.8 | 24.6 | 23.9 |
CL-petrol | 34 | 34.9 | 35.6 | 29.7 | 29.8 |
CL-petrol | 38.1 | 38.5 | 39.6 | 36.6 | 36.9 |
CL-petrol | 42 | 42.4 | 44.4 | 38.3 | 36.1 |
CL-petrol | 46.2 | 46.9 | 45.2 | 43.5 | 43.7 |
Average error (%) | 1 | 5 | 13 | 16 |
Dataset | Traning Data | Test Data |
---|---|---|
Biodiesel with generic model | 7.2238 () | () |
Biodiesel with added noise | ||
SNR (dB) | Traning data | Test data |
40 | ||
30 | ||
20 | 0.001 | |
10 | 0.001 | |
Biodiesel with added noise on testing data | ||
SNR (dB) | Training data | Test data |
40 | ||
30 | ||
20 | ||
10 |
Soil Type | MSE on Training Set | MSE on the Test Set |
---|---|---|
Hamra | ||
Evona | ||
Kokhav |
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Share and Cite
Ahmed, A.M.; Duran, O.; Zweiri, Y.; Smith, M. Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data. Remote Sens. 2019, 11, 1938. https://doi.org/10.3390/rs11161938
Ahmed AM, Duran O, Zweiri Y, Smith M. Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data. Remote Sensing. 2019; 11(16):1938. https://doi.org/10.3390/rs11161938
Chicago/Turabian StyleAhmed, Asmau M., Olga Duran, Yahya Zweiri, and Mike Smith. 2019. "Quantification of Hydrocarbon Abundance in Soils Using Deep Learning with Dropout and Hyperspectral Data" Remote Sensing 11, no. 16: 1938. https://doi.org/10.3390/rs11161938