Department of Computer Science and Engineering, Parala Maharaja Engineering College, Berhampur 761003, India
Amity School of Engineering and Technology, Amity University, Uttar Pradesh 201303, India
Department of Computer Applications, Sikkim Manipal Institute of Technology, Sikkim Manipal University, Sikkim 737102, India
Directorate of Research, Sikkim Manipal University, Gangtok, Sikkim 737102, India
Show Author Information
Hide Author Information
Abstract
As a huge number of satellites revolve around the earth, a great probability exists to observe and determine the change phenomena on the earth through the analysis of satellite images on a real-time basis. Therefore, classifying satellite images plays strong assistance in remote sensing communities for predicting tropical cyclones. In this article, a classification approach is proposed using Deep Convolutional Neural Network (DCNN), comprising numerous layers, which extract the features through a downsampling process for classifying satellite cloud images. DCNN is trained marvelously on cloud images with an impressive amount of prediction accuracy. Delivery time decreases for testing images, whereas prediction accuracy increases using an appropriate deep convolutional network with a huge number of training dataset instances. The satellite images are taken from the Meteorological & Oceanographic Satellite Data Archival Centre, the organization is responsible for availing satellite cloud images of India and its subcontinent. The proposed cloud image classification shows 94% prediction accuracy with the DCNN framework.
No abstract is available for this article. Click the button above to view the PDF directly.
References
[1]
M.Mizutori and D.Guha-Sapir, Economic Losses, Poverty and Disasters 1998-2017. United Nations Office for Disaster Risk Reduction, United Nations, New York, USA, 2017.
[2]
P.Wallemacq and H.Rowena, Economic Losses, Poverty & Disasters: 1998–2017. Brussels, Belgium: Centre for Research on the Epidemiology of Disasters, 2018.
[3]
F.Thomalla and H.Schmuck, ‘We all knew that a cyclone was coming’: Disaster preparedness and the cyclone of 1999 in Orissa, India, Disasters, vol. 28, no. 4, pp. 373–387, 2004.
L. Z.Jiang, S. M.Fu, and J. H.Sun, New method for detecting extratropical cyclones: The eight-section slope detecting method, Atmos. Ocean. Sci. Lett., vol. 13, no. 5, pp. 436–442, 2020.
U.Neu, M. G.Akperov, N.Bellenbaum, R.Benestad, R.Blender, R.Caballero, A.Cocozza, H. F.Dacre, Y.Feng, K.Fraedrich, et al., IMILAST: A community effort to intercompare extratropical cyclone detection and tracking algorithms, Bull. Am. Meteorol. Soc., vol. 94, no. 4, pp. 529–547, 2013.
A.Murata, S. I. I.Watanabe, H.Sasaki, H.Kawase, and M.Nosaka, The development of a resolution-independent tropical cyclone detection scheme for high-resolution climate model simulations, J. Meteorol. Soc. Japan. Ser. II, vol. 97, no. 2, pp. 519–531, 2019.
D.Matsuoka, M.Nakano, D.Sugiyama, and S.Uchida, Deep learning approach for detecting tropical cyclones and their precursors in the simulation by a cloud-resolving global nonhydrostatic atmospheric model, Prog. Earth Planet. Sci., vol. 5, no. 1, p. 80, 2018.
S.Shakya, S.Kumar, and M.Goswami, Deep learning algorithm for satellite imaging based cyclone detection, IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 13, pp. 827–839, 2020.
J. G.Powers, J. B.Klemp, W. C.Skamarock, C. A.Davis, J.Dudhia, D. O.Gill, J. L.Coen, D. J.Gochis, R.Ahmadov, S. E.Peckham, et al., The weather research and forecasting model: Overview, system efforts, and future directions, Bull. Am. Meteorol. Soc., vol. 98, no. 8, pp. 1717–1737, 2017.
G. E.Liston and R. A.Pielke, A climate version of the regional atmospheric modeling system, Theor. Appl. Climatol., vol. 66, nos. 1&2, pp. 29–47, 2000.
R. A.Pielke, W. R.Cotton, R. L.Walko, C. J.Tremback, W. A.Lyons, L. D.Grasso, M. E.Nicholls, M. D.Moran, D. A.Wesley, T. J.Lee, et al., A comprehensive meteorological modeling system-RAMS, Meteorol. Atmos. Phys., vol. 49, nos. 1–4, pp. 69–91, 1992.
M.Mu, W. S.Duan, and B.Wang, Conditional nonlinear optimal perturbation and its applications, Nonlin. Process. Geophys., vol. 10, no. 6, pp. 493–501, 2003.
J. J.Baik and H. S.Hwang, Tropical cyclone intensity prediction using regression method and neural network, J. Meteorol. Soc. Japan. Ser. II, vol. 76, no. 5, pp. 711–717, 1998.
M.Rütgers, S.Lee, and D.You, Typhoon track prediction using satellite images in a generative adversarial network, arXiv preprint arXiv: 1808.05382, 2018.
R.Kovordányi and C.Roy, Cyclone track forecasting based on satellite images using artificial neural networks, ISPRS J. Photogramm. Remote Sens., vol. 64, no. 6, pp. 513–521, 2009.
J.Salat, J.Pascual, M.Flexas, T. M.Chin, and J.Vazquez-Cuervo, Forty-five years of oceanographic and meteorological observations at a coastal station in the NW Mediterranean: A ground truth for satellite observations, Ocean Dynamics, vol. 69, no. 9, pp. 1067–1084, 2019.
A. K.Rai, N.Mandal, A.Singh, and K. K.Singh, Landsat 8 OLI satellite image classification using convolutional neural network, Procedia Comput. Sci., vol. 167, pp. 987–993, 2020.
A.Wimmers, C.Velden, and J. H.Cossuth, Using deep learning to estimate tropical cyclone intensity from satellite passive microwave imagery, Mon. Wea. Rev., vol. 147, no. 6, pp. 2261–2282, 2019.
B.Chen, B. F.Chen, and H. T.Lin, Rotation-blended CNNs on a new open dataset for tropical cyclone image-to-intensity regression, in Proc. 24th ACM SIGKDD Int. Conf. Knowledge Discovery & Data Mining, London, UK, 2018, pp. 90–99.
J. S.Combinido, J. R.Mendoza, and J.Aborot, A convolutional neural network approach for estimating tropical cyclone intensity using satellite-based infrared images, in 2018 24th Int. Conf. Pattern Recognition (ICPR), Beijing, China, 2018, pp. 1474–1480.
C.Kar, A.Kumar, and S.Banerjee, Tropical cyclone intensity detection by geometric features of cyclone images and multilayer perceptron, SN Appl. Sci., vol. 1, no. 9, p. 1099, 2019.
S. M.Chen, Y. M.Wang, and I.Tsou, Using artificial neural network approach for modelling rainfall-runoff due to typhoon, J. Earth Syst. Sci., vol. 122, no. 2, pp. 399–405, 2013.
C. C.Young and W. C.Liu, Prediction and modelling of rainfall-runoff during typhoon events using a physically-based and artificial neural network hybrid model, Hydrol. Sci.J., vol. 60, no. 12, pp. 2102–2116, 2015.
S.Kim, Y.Matsumi, S. Q.Pan, and H.Mase, A real-time forecast model using artificial neural network for after-runner storm surges on the Tottori coast, Japan, Ocean Eng., vol. 122, pp. 44–53, 2016.
Y. F.Wang, W.Zhang, and W.Fu, Back Propogation(BP)-neural network for tropical cyclone track forecast, in Proc. 2011 19th Int. Conf. Geoinformatics, Shanghai, China, 2011, pp. 1–4.
J. Y.Zhou, J.Xiang, and S. X.Huang, Classification and prediction of typhoon levels by satellite cloud pictures through GC-LSTM deep learning model, Sensors, vol. 20, no. 18, p. 5132, 2020.
S. S.Roy, V.Lakshmanan, S. K. R.Bhowmik, and S. B.Thampi, Doppler weather radar based nowcasting of cyclone Ogni, J. Earth Syst. Sci., vol. 119, no. 2, pp. 183–199, 2010.
S. N. K. B.Amit and Y.Aoki, Disaster detection from aerial imagery with convolutional neural network, in Proc. 2017 Int. Electronics Symp. Knowledge Creation and Intelligent Computing (IES-KCIC), Surabaya, Indonesia, 2017, pp. 239–245.
M.Kim, M. S.Park, J.Im, S.Park, and M. I.Lee, Machine learning approaches for detecting tropical cyclone formation using satellite data, Remote Sens., vol. 11, no. 10, p. 1195, 2019.
I.Kouretas and V.Paliouras, Simplified hardware implementation of the softmax activation function, in Proc. 2019 8th Int. Conf. Modern Circuits and Systems Technologies (MOCAST), Thessaloniki, Greece, 2019, pp. 1–4.
X. Y.Deng, Q.Liu, Y.Deng, and S.Mahadevan, An improved method to construct basic probability assignment based on the confusion matrix for classification problem, Inf. Sci., vols. 340&341, pp. 250–261, 2016.
R.Susmaga, Confusion matrix visualization, in Intelligent Information Processing and Web Mining, M. A.Kłpotek, S. T.WierzchońK.Trojanowski, eds. Berlin, Germany: Springer, 2004, pp. 107–116.
Jena KK, Bhoi SK, Nayak SR, et al. Deep Convolutional Network Based Machine Intelligence Model for Satellite Cloud Image Classification. Big Data Mining and Analytics, 2023, 6(1): 32-43. https://doi.org/10.26599/BDMA.2021.9020017
The articles published in this open access journal are distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/).
10.26599/BDMA.2021.9020017.F001
CNN model.
10.26599/BDMA.2021.9020017.F002
DCNN-based machine intelligence model for cyclone detection.
Experimental Evaluation
The deep convolutional network has been implemented on Python with the help of keras API as the front end and TensorFlow as the backend. The keras and TensorFlow provide ease of implementing the deep convolutional network, as it contains predefined layers, models, and optimizers[
19
]. The implementation platform is GPU NVIDIA MX250, type GDDR5, and has 1518 MHz and a RAM of 8 GB, which enhances the speed of training the model on a 64-bit Windows platform. A sequential model, which contains layers (Conv2D, Max-pooling, Flattening, Dropout) with
the input of 224 pixel 224 pixel image, is illustrated in
Fig. 3
. All training and validation images after processing (loading, resizing, and labeling) pass through it. The models are trained and validated in every epoch (once a total dataset passes through forward and backward in the network or sequential model). Some sample images for cyclone and non-cyclone are displayed in
Fig. 4
and
Fig. 5
, respectively. An image of 224 pixel224 pixel is given as the input to the sequential model where two convolutional 2D networks are applied, taking 64 filters, defining the kernel of 33, and keeping the padding the “same”. Each convolutional feature extraction takes place in a feature matrix, and the matrix is followed to the next layer for further processing.
10.26599/BDMA.2021.9020017.F003
Proposed sequential model using ReLU.
10.26599/BDMA.2021.9020017.F004
Bulbul cyclone images in the dataset (sample images)[
20
,
21
].
10.26599/BDMA.2021.9020017.F005
Normal images in the dataset (sample images)[
20
,
21
].
As previously mentioned, convolution is applied to three segments, and the filters double in each segment. Therefore, feature detection gives high accuracy to predict the testing images. Dropout is mainly introduced for avoiding overfitting. The model performance is evaluated using a confusion matrix[
43
,
44
].
As mentioned above, the model starts with an input image where the first convolutional layer extracts 64 feature matrices taking a kernel size of 33. The input of the second convolutional layer is the output of the first convolutional layer. Here the convolution again extracts 64 features from the feature matrices of the previous layer, taking a kernel of 33. Down-sampling images is the main objective of the CNN model, and therefore, a MaxPool2D, which is also called a max-pooling layer, reduces images without reducing their features. The max-pooling layer retrieves all the features into a feature map/matrix. The pool size of 22 reduces the feature matrices to half of the original size. The output of the max-pooling layer is the input of the third convolutional layer. In the second segment of the model, which contains the third and fourth convolutional layers, 128 feature matrices are extracted, followed by a dropout of 0.5, which prevents the model from overfitting. The second max-pooling layer again down-samples the feature matrices into half of the previous size by using a pool size of 22. The third segment of the model, which contains three convolutional layers with 256 feature matrices each and a kernel size of 33, again extracts the appropriate features for classifying the problem. The max-pooling layer again comes into the role of down-sampling. The input feature matrices, which are the outputs of the previous convolutional layer, are finally reduced into many feature matrices. The model at this phase deals with the matrices of size 2828 and 256 feature matrices or filters. Thus, converting these 2D arrays or matrices into a single vector is carried by the flattening layer whose output is 2828256, which is equal to 200 704. The model now deals with 200 704 parameters, which are being distinguished by a dense layer of two units, as the model is a binary classifier. The SoftMax classifies the parameters into a probability of a cyclone or not. The epochs “” defined in this article are 10, and the batch size is 10. The percentages of accuracy, loss, validation loss, and validation accuracy are presented in
Table 1
.
10.26599/BDMA.2021.9020017.T001
Epochs and corresponding results while training and validating the proposed model. (%)
Epochs
Training loss
Training accuracy
Validation loss
Validation accuracy
-1/10
86.64
93.72
11.72
99.15
-2/10
3.86
98.80
10.64
98.95
-3/10
3.13
99.19
4.15
99.84
-4/10
2.47
99.39
4.81
99.96
-5/10
2.58
99.43
12.56
98.43
-6/10
1.66
99.51
2.37
100.00
-7/10
0.35
99.90
1.93
100.00
-8/10
0.00
100.00
1.55
100.00
-9/10
0.01
100.00
1.09
100.00
-10/10
13.68
98.74
7.07
99.80
Average
11.44
98.87
5.79
99.61
Standard deviation
26.72
1.86
4.43
0.56
We can also define model loss and accuracy with reference to the image shown in
Fig. 6
. Here, accuracy and loss are optimal with respect to per epoch. From the first epoch, the training data show a condition called underfitting. However, while the epoch increases, the training data tend to fit the model. In -1, the accuracy is nearly 93%, and the loss is considered nearly 8%. It suggests that during the first batch of the dataset with 494 images, 93% are trained accurately; meanwhile, 7% are not trained accurately. In -2, the accuracy is approximately 98%, and the loss percentage is approximately 2.0. The model has significantly trained many images accurately from the batch size. As soon as the model is showing a good result at a point exactly at -8, the model starts to overfit and thus signifies that the training accuracy is more, whereas the validation accuracy is less. However, in -9, the accuracy reduces to 1%, and the model perfectly fits. In the last epoch, the accuracy level of model training is nearly 98%, and the accuracy level of validation is 99%. With these results, we can easily confirm that the model fits the data. The underfit problem is also eradicated.
10.26599/BDMA.2021.9020017.F006
Training and validation accuracy with respect to per epoch.
Figure 6
displays the graph between validation accuracy and training accuracy. The deviation between the two lines is evident at the beginning of epochs. However, the deep neural network during the training increases its accuracy, and the deviation between the two lines decreases. The lines shown in
Fig. 6
tend to merge at the ninth epoch, justifying the fitting of the proposed model to the training and validating datasets.
Now, the confusion matrix helps us determine the accuracy of the total model. The formula of the confusion matrix is determined in Eq. (
3
). It can be defined by the ratio of the addition of true positive and false positive with respect to the total number of instances. The total number of testing images is 1008, out of which 408 are random cyclone images of AMPHAN and OCKHI. These images are not considered before in the training or validation dataset, and the model has predicted 352 images to be cyclones[
20
,
21
]. Thus, it has also predicted 587 images to be non-cyclones from 600 images. The accuracy can be defined by the following formula:
where is denoted as a predicted “cyclone” of a cyclone class, is denoted as predicted “not a cyclone” but belongs to the cyclone class, is denoted as predicted “not a cyclone” of the no_cyclone class, is denoted as a predicted “cyclone” which belongs to the no_cyclone class. Putting the values of , , and , the model obtains an accuracy of 94.14%, and an error rate of less than 6%. The confusion matrix of the proposed model is illustrated in
Fig. 7
. The corresponding experimental results are outlined in
Table 2
.
10.26599/BDMA.2021.9020017.T002
Testing results obtained through the proposed model.
Result
Value
Sensitivity
86.3%
Specificity
0.995
Precision
0.992
Negative predictive value
0.914
False positive rate
0.005
False discovery rate
0.009
False negative rate
0.137
Accuracy
94.2%
F1 score
0.923
Matthews correlation coefficient
0.881
10.26599/BDMA.2021.9020017.F007
Confusion matrix of the proposed model.
From the summarized results in
Table 2
, the proposed model shows an appealing detection accuracy of 94.2%. The model reveals not only detection accuracy but also a sensitivity of 86.3%, which appears to be unique in the field of cyclone detection. Another point of observation is that the false positive rate is quite impressively low, with only 0.005. It shows the real power of our proposed system, as it seems to be stable in all the performance measures.
In the final stage of analysis, we compare our proposed cyclone detection model with other state-of-the-art cyclone detection models. Given that our model is based on a deep learning technique, we shortlist a few deep learning based models for comparison in
Table 3
. In this regard, DLR–FH, Deep CNN, and DAV–T cyclone detection models implemented by Rajesh et al.[
7
], Deep CNN (2D) model proposed by Matsuoka et al.[
8
], Multilayer CNN proposed by Kovordányi and Roy[
17
], and PCA-based CNN model proposed by Rai et al.[
22
] are considered. The proposed model exhibits at par performance with other models in terms of detection accuracy. The proposed multilayer CNN model shows 12.20% and 5.1% better detection accuracy than the deep learning approach proposed by Matsuoka et al.[
8
] and the DAV-T approach implemented by Rajesh et al.[
7
], respectively. Our system also slightly lacks less than DLR-FH, Multilayer CNN, and PCA-CNN approaches.
10.26599/BDMA.2021.9020017.T003
Comparison of the proposed model with other state-of-the-art models.
Author
Method
Accuracy (%)
Rajesh et al.[
7
]
DLR–FH
96.00
Deep CNN
90.00
DAV–T
82.00
Matsuoka et al.[
8
]
Deep CNN (2D)
89.10
Kovordányi and Roy[
17
]
Multilayer CNN
98.70
Rai et al.[
22
]
Principal Component Analysis-CNN (PCA-CNN)
94.50
Proposed model
Multilayer CNN
94.20
10.26599/BDMA.2021.9020017.F003
Proposed sequential model using ReLU.
10.26599/BDMA.2021.9020017.F004
Bulbul cyclone images in the dataset (sample images)[
20
,
21
].
10.26599/BDMA.2021.9020017.F005
Normal images in the dataset (sample images)[
20
,
21
].
10.26599/BDMA.2021.9020017.F006
Training and validation accuracy with respect to per epoch.
10.26599/BDMA.2021.9020017.F007
Confusion matrix of the proposed model.