# A Hybrid Deep Learning Model for the Bias Correction of SST Numerical Forecast Products Using Satellite Data

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

- We propose a new hybrid model for SST correction, which uses satellite remote sensing observation data and spatio-temporal data of sea surface variables. The performance of our model is then evaluated;
- The attention mechanism is used to assign weights to the information in the dataset, which reflect the influence of spatio-temporal information on the SST correction, so that the key information is highlighted and thus we obtain better correction results;
- Taking the South China Sea area (10°N–15°N, 125°E–130°E) as an example, the accuracy rate was improved by 41.9% after the correction. We analyze the influence of input sequence with different time steps, different model parameters and other variables on the correction effect through the experiments. Experiments on the dataset of the South China Sea show that our new hybrid model is more effective than existing methods, including some classical machine learning methods.

## 2. Related Work

## 3. Problem Definitions

## 4. Method

#### 4.1. The Framework of the New Hybrid SST Correction Model

#### 4.2. Spatial Feature Extraction with 3D-CBAM

#### 4.3. Time Feature Extraction with Attention Mechanism

_{t}represents the moment’s input, ht−1 represents last time’s hidden state, w is the weight matrix, and b is the offset from the input gate to the output gate, which are the characteristics that the ConvLSTM model must learn during training.

_{t}of the ConvLSTM, the calculation formula is shown in (16).

## 5. Experiments and Results

#### 5.1. Data Preparation and Evaluation Metrics

#### 5.2. Comparison of Correction Methods

#### 5.3. Complexity and Training Time Analysis

#### 5.4. Parameters Analysis

#### 5.4.1. Time Step Analysis

#### 5.4.2. Learning Rate Analysis

#### 5.4.3. Epochs Analysis

## 6. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Conflicts of Interest

## References

- Funk, C.C.; Hoell, A. The leading mode of observed and cmip5 enso-residual sea surface temperatures and associated changes in indo-pacific climate. J. Clim.
**2015**, 28, 150202132719008. [Google Scholar] [CrossRef] - Solanki, H.U.; Bhatpuria, D.; Chauhan, P. Integrative analysis of altika-ssha, modis-sst, and ocm-chlorophyll signatures for fisheries applications. Mar. Geod.
**2015**, 38 (Suppl. 1), 672–683. [Google Scholar] [CrossRef] - Yang, Y.; Dong, J.; Sun, X.; Lima, E.; Mu, Q.; Wang, X. A cfcc-lstm model for sea surface temperature prediction. IEEE Geosci. Remote Sens. Lett.
**2017**, 15, 207–211. [Google Scholar] [CrossRef] - Stockdale, T.N.; Balmaseda, M.A.; Vidard, A. Tropical atlantic sst prediction with coupled ocean-atmosphere gcms. J. Clim.
**2006**, 19, 6047. [Google Scholar] [CrossRef] - Song, Z.; Qiao, F.; Yang, Y.; Yuan, Y. An improvement of the too cold tongue in the tropical pacific with the development of an ocean-wave-atmosphere coupled numerical model. Prog. Nat. Sci.
**2007**, 17, 576–583. [Google Scholar] - Xu, Z.; Li, M.; Patricola, C.M.; Ping, C. Oceanic origin of southeast tropical atlantic biases. Clim. Dyn.
**2014**, 43, 2915–2930. [Google Scholar] [CrossRef] - Peng, S.Q.; Xie, L. Effect of determining initial conditions by four-dimensional variational data assimilation on storm surge forecasting. Ocean Model.
**2006**, 14, 1–18. [Google Scholar] [CrossRef] [Green Version] - Li, X.; Wang, Q.; Mu, M. Optimal initial error growth in the prediction of the kuroshio large meander based on a high-resolution regional ocean model. Adv. Atmos. Sci.
**2018**, 35, 1362–1371. [Google Scholar] [CrossRef] - Hemri, S.; Scheuerer, M.; Pappenberger, F.; Bogner, K.; Haiden, T. Trends in the predictive performance of raw ensemble weather forecasts. Geophys. Res. Lett.
**2014**, 41, 9197–9205. [Google Scholar] [CrossRef] - Vannitsem, S. Dynamical properties of mos forecasts: Analysis of the ecmwf operational forecasting system. Weather Forecast.
**2010**, 23, 1032–1043. [Google Scholar] [CrossRef] - Tian, D.; Martinez, C.J.; Graham, W.D.; Hwang, S. Statistical downscaling multimodel forecasts for seasonal precipitation and surface temperature over the southeastern united states. J. Clim.
**2014**, 27, 8384–8411. [Google Scholar] [CrossRef] - Libonati, R.; Trigo, I.; Dacamara, C.C. Correction of 2m-temperature forecasts using kalman filtering technique. Atmos. Res.
**2008**, 87, 183–197. [Google Scholar] [CrossRef] - Pelosi, A.; Medina, H.; Bergh, J.V.D.; Vannitsem, S.; Chirico, G.B. Adaptive kalman filtering for postprocessing ensemble numerical weather predictions. Mon. Weather Rev.
**2017**, 145, 4837–4854. [Google Scholar] [CrossRef] - Wang, J.; Chen, C.; Long, K.; Feng, L. Temporal and spatial distribution of short-time heavy rain of Sichuan Basin in summer. Plateau Mt. Meteorol. Res.
**2015**, 35, 16–20. [Google Scholar] - Zhang, Q.; Yu, Y.; Zhang, W.; Luo, T.; Wang, X. Cloud detection from fy-4a’s geostationary interferometric infrared sounder using machine learning approaches. Remote Sens.
**2019**, 11, 3035. [Google Scholar] [CrossRef] [Green Version] - Zeng, J.; Zhang, C.; Wang, H.; Chu, H. Correction model for the temperature of numerical weather prediction by SVM. Second Target Recognit. Artif. Intell. Summit Forum
**2020**, 11427, 114270Z. [Google Scholar] - Zhang, R.; Yu, Z.H.; Jiang, Q.R. Neural network bp model approximation and prediction of complicated weather systems. Acta Meteorol. Sin.
**2001**, 15, 105–115. [Google Scholar] - Sayeed, A.; Choi, Y.; Jung, J.; Lops, Y.; Eslami, E.; Salman, A.K. A deep convolutional neural network model for improving WRF forecasts. Atmos. Environ.
**2020**, 253, 118376. [Google Scholar] [CrossRef] - Kupilik, M.; Witmer, F.D.W.; MacLeod, E.-A.; Wang, C.; Ravens, T. Gaussian Process Regression for Arctic Coastal Erosion Forecasting. IEEE Trans. Geosci. Remote Sens.
**2019**, 57, 1256–1264. [Google Scholar] [CrossRef] - Brekke, C.; Solberg, A. Oil spill detection by satellite remote sensing. Remote Sens. Environ.
**2005**, 95, 1–13. [Google Scholar] [CrossRef] - Yu, Y.; Yang, X.; Zhang, W.; Duan, B.; Cao, X.; Leng, H. Assimilation of sentinel-1 derived sea surface winds for typhoon forecasting. Remote Sens.
**2017**, 9, 845. [Google Scholar] [CrossRef] [Green Version] - Chen, R.; Zhang, W.; Wang, X. Machine learning in tropical cyclone forecast modeling: A review. Atmosphere
**2020**, 11, 676. [Google Scholar] [CrossRef] - Xi, X.F.; Zhou, G.D. A survey on deep learning for natural language processing. Acta Autom. Sin.
**2016**, 42, 1445–1465. [Google Scholar] - Lee, H.; Pham, P.T.; Largman, Y.; Ng, A.Y. Unsupervised feature learning for audio classification using convolutional deep belief networks. Adv. Neural Inf. Process. Syst.
**2009**, 22, 1096–1104. [Google Scholar] - Sattar, N.S.; Arifuzzaman, S. Community Detection using Semi-supervised Learning with Graph Convolutional Network on GPUs. In Proceedings of the 2020 IEEE International Conference on Big Data (Big Data), Atlanta, GA, USA, 10–13 December 2020; pp. 5237–5246. [Google Scholar]
- Jain, V.; Murray, J.F.; Roth, F.; Turaga, S.; Zhigulin, V.; Briggman, K.L.; Helmstaedter, M.N.; Denk, W.; Seung, H.S. Supervised Learning of Image Restoration with Convolutional Networks. In Proceedings of the 2007 IEEE 11th International Conference on Computer Vision, Rio de Janeiro, Brazil, 14–21 October 2007; pp. 1–8. [Google Scholar]
- Shi, X.; Chen, Z.; Wang, H.; Yeung, D.Y.; Wong, W.K.; Woo, W.C. Convolutional Lstm Network: A Machine Learning Approach for Precipitation Nowcasting; MIT Press: Cambridge, MA, USA, 2015. [Google Scholar]
- Liu, D.; Jiang, W.; Mu, L.; Wang, S. Streamflow Prediction Using Deep Learning Neural Network: Case Study of Yangtze River. IEEE Access
**2020**, 8, 90069–90086. [Google Scholar] [CrossRef] - Petrou, Z.I.; Tian, Y. Prediction of sea ice motion with convolutional long short-term memory networks. IEEE Trans. Geosci. Remote Sens.
**2019**, 99, 1–12. [Google Scholar] [CrossRef] - Chen, R.; Wang, X.; Zhang, W.; Zhu, X.; Li, A.; Yang, C. A hybrid cnn-lstm model for typhoon formation forecasting. GeoInformatica
**2019**, 23, 375–396. [Google Scholar] [CrossRef] - Winona, A.Y.; Adytia, D. Short Term Forecasting of Sea Level by Using LSTM with Limited Historical Data. In Proceedings of the 2020 International Conference on Data Science and Its Applications (ICoDSA), Bandung, Indonesia, 5–6 August 2020; pp. 1–5. [Google Scholar]
- Kun, X.; Shan, T.; Yi, T.; Chao, C. Attention-based long short-term memory network temperature prediction model. In Proceedings of the 2021 7th International Conference on Condition Monitoring of Machinery in Non-Stationary Operations (CMMNO), Guangzhou, China, 11–13 June 2021. [Google Scholar]
- Krishnamurti, T.N.; Kishtawal, C.M.; LaRow, T. Improved weather and seasonal climate forecasts from multimodel superensemble. Science
**1999**, 285, 1548–1550. [Google Scholar] [CrossRef] [Green Version] - Xu, Z.; Wang, Y.; Fan, G. A two-stage quality control method for 2-m temperature observations using biweight means and a progressive eof analysis. Mon. Weather Rev.
**2013**, 141, 798–808. [Google Scholar] [CrossRef] - Zhang, X.; Gao, S.; Wang, T.; Li, Y.; Ren, P. Correcting Predictions from Simulating Wave Nearshore Model via Gaussian Process Regression. In Proceedings of the Global Oceans 2020: Singapore—U.S. Gulf Coast, Biloxi, MS, USA, 5–30 October 2020; pp. 1–4. [Google Scholar]
- Doroshenko, A.; Shpyg, V.; Kushnirenko, R. Machine Learning to Improve Numerical Weather Forecasting. In Proceedings of the 2020 IEEE 2nd International Conference on Advanced Trends in Information Theory (ATIT), Kyiv, Ukraine, 25–27 November 2020. [Google Scholar]
- Wang, X.; Li, X.; Zhu, J.; Xu, Z.; Yu, K. A local similarity-preserving framework for nonlinear dimensionality reduction with neural networks. In Proceedings of the The 26th International Conference on Database Systems for Advanced Applications (Dasfaa 2021), Tai Pei, China, 11–14 April 2021. [Google Scholar]
- Wang, A.; Xu, L.; Li, Y.; Xing, J.; Zhou, Z. Random-forest based adjusting method for wind forecast of WRF model. Comput. Geosci.
**2021**, 55, 104842. [Google Scholar] [CrossRef] - Zheng, G.; Li, X.; Zhang, R.H.; Liu, B. Purely satellite data–driven deep learning forecast of complicated tropical instability waves. Sci. Adv.
**2020**, 6, eaba1482. [Google Scholar] [CrossRef] [PubMed] - Makarynskyy, O. Improving wave predictions with artificial neural networks. Ocean Eng.
**2004**, 31, 709–724. [Google Scholar] [CrossRef] - Xu, X.; Liu, Y.; Chao, H.; Luo, Y.; Chu, H.; Chen, L. Towards a precipitation bias corrector against noise and maldistribution. arXiv
**2019**, arXiv:1910.07633. [Google Scholar] - Wang, T.; Gao, S.; Xu, J.; Li, Y.; Li, P.; Ren, P. Correcting Predictions from Oceanic Maritime Numerical Models via Residual Learning. In Proceedings of the 2018 OCEANS—MTS/IEEE Kobe Techno-Ocean. (OTO), Kobe, Japan, 28–31 May 2018; pp. 1–4. [Google Scholar]
- Rasp, S.; Lerch, S. Neural networks for post-processing ensemble weather forecasts. Mon. Weather Rev.
**2018**, 146, 3885–3900. [Google Scholar] [CrossRef] [Green Version] - Deshmukh, A.N.; Deo, M.C.; Bhaskaran, P.K.; Nair, T.; Sandhya, K.G. Neural-network-based data assimilation to improve numerical ocean wave forecast. IEEE J. Ocean. Eng.
**2016**, 4, 944–953. [Google Scholar] [CrossRef] - Yang, X.Q.; Anderson, J.L. Correction of systematic errors in coupled gcm forecasts. J. Clim.
**2000**, 13, 2072–2085. [Google Scholar] [CrossRef] [Green Version] - Han, Y.K.; Dan, Y.U.; Shen, X.Y.; Zhou, Y.Y. Study on the correction of SST prediction of HYCOM. Mar. Forecast.
**2018**, 35, 5. (In Chinese) [Google Scholar] - Zhang, P.J.; Zhou, S.H.; Liang, C.X. Study on the correction of SST prediction in South China Sea using remotely sensed SST. J. Trop. Oceanogr.
**2020**, 39, 59–67. (In Chinese) [Google Scholar] - Ji, S.; Xu, W.; Yang, M.; Yu, K. 3d convolutional neural networks for human action recognition. IEEE Trans. Pattern Anal. Mach. Intell.
**2013**, 35, 221–231. [Google Scholar] [CrossRef] [Green Version] - Woo, S.; Park, J.; Lee, J.Y.; Kweon, I.S. Cbam: Convolutional block attention module. In Proceedings of the European Conference on Computer Vision (ECCV), Munich, Germany, 8–14 September 2018. [Google Scholar]
- Mnih, V.; Heess, N.; Graves, A.; Kavukcuoglu, K. Recurrent models of visual attention. Adv. Neural Inf. Processing Syst.
**2014**, 2, 2204–2212. [Google Scholar] - Bleck, R. An oceanic general circulation model framed in hybrid isopycnic-cartesian coordinates. Ocean Modeling
**2002**, 4, 88. [Google Scholar] [CrossRef] - Metzger, E.J.; Smedstad, O.M.; Thoppil, P.G.; Hurlburt, H.E.; Cummings, J.A. US Navy Operational Global Ocean and Arctic Ice Prediction Systems. Oceanography
**2014**, 27, 32–43. [Google Scholar] [CrossRef] - Reynolds, R.W.; Smith, T.M.; Liu, C.; Chelton, D.B.; Casey, K.S.; Schlax, M.G. Daily High-Resolution-Blended Analyses for Sea Surface Temperature. J. Clim.
**2007**, 20, 5473–5496. [Google Scholar] [CrossRef] - Liu, Y.; Weisberg, R.H.; Law, J.; Huang, B. Evaluation of Satellite-Derived SST Products in Identifying the Rapid Temperature Drop on the West Florida Shelf Associated With Hurricane Irma. Mar. Technol. Soc. J.
**2018**, 52, 43. [Google Scholar] [CrossRef]

**Figure 7.**The location of test area: (

**a**) Satellite image of the test area location, the box is the test area; (

**b**) SST map of the test area location, the box is the test area.

**Figure 8.**The experimental results of different methods for SST correction. (

**a**) Truth; (

**b**) forecast; (

**c**) linear regression; (

**d**) SVR; (

**e**) LSTM; (

**f**) CONVLSTM; (

**g**) CONVLSTM-AT; (

**h**) 3DCNN-CONVLSTM-AT; (

**i**) 3DCNN-CBAM-CONVLSTM-AT.

**Figure 9.**The comparisons of difference between the truth and the correction output. (

**a**) Difference between the truth and the forecast; (

**b**) difference between the truth and the linear regression result; (

**c**) difference between the truth and the SVR result; (

**d**) difference between the truth and the LSTM result; (

**e**) difference between the truth and the CONVLSTM result; (

**f**) difference between the truth and the CONVLSTM-AT result; (

**g**) difference between the truth and the 3DCNN-CONVLSTM-AT result; (

**h**) difference between the truth and the 3DCNN-CBAM-CONVLSTM-AT result.

**Figure 10.**The experimental results of the new hybrid SST correction model in different timesteps. The units of RMSE, MAE, and MSE are °C, the unit of MAPE is %, and the unit of timestep is day.

**Figure 11.**The experimental results of the new hybrid SST correction model in different learning rate. The units of RMSE, MAE, and MSE are °C and the unit of MAPE is %.

**Figure 12.**The experimental results of the new hybrid SST correction model in different epochs. The units of RMSE, MAE, and MSE are °C and the unit of MAPE is %.

MAPE | MAE | MSE | RMSE | Improve | |
---|---|---|---|---|---|

Forecast | 1.6118 | 0.4587 | 0.3600 | 0.6000 | |

Linear Regression (LR) | 1.4592 | 0.4075 | 0.3005 | 0.5482 | 8.67% |

Support Vector Regression (SVR) | 1.3767 | 0.3832 | 0.2536 | 0.5036 | 16.17% |

LSTM | 1.2781 | 0.3553 | 0.2115 | 0.4599 | 23.35% |

CONVLSTM | 1.1679 | 0.3312 | 0.1842 | 0.4292 | 28.47% |

CONVLSTM-AT | 1.1071 | 0.3139 | 0.1623 | 0.4028 | 32.92% |

3DCNN-CONVLSTM-AT | 1.0033 | 0.2839 | 0.3600 | 0.3690 | 38.5% |

3DCNN-CBAM-CONVLSTM-AT | 0.9546 | 0.2641 | 0.1239 | 0.3520 | 41.33% |

Parameters | Train(s) | Test(s) | |
---|---|---|---|

LSTM | 13,601 | 271.52 | 0.55 |

CONVLSTM | 44,993 | 236.98 | 0.46 |

CONVLSTM-AT | 46,079 | 437.67 | 0.94 |

3DCNN-CONVLSTM-AT | 13,197 | 272.57 | 0.55 |

3DCNN-CBAM-CONVLSTM-AT | 13,560 | 223.15 | 0.43 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2022 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 (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Fei, T.; Huang, B.; Wang, X.; Zhu, J.; Chen, Y.; Wang, H.; Zhang, W.
A Hybrid Deep Learning Model for the Bias Correction of SST Numerical Forecast Products Using Satellite Data. *Remote Sens.* **2022**, *14*, 1339.
https://doi.org/10.3390/rs14061339

**AMA Style**

Fei T, Huang B, Wang X, Zhu J, Chen Y, Wang H, Zhang W.
A Hybrid Deep Learning Model for the Bias Correction of SST Numerical Forecast Products Using Satellite Data. *Remote Sensing*. 2022; 14(6):1339.
https://doi.org/10.3390/rs14061339

**Chicago/Turabian Style**

Fei, Tonghan, Binghu Huang, Xiang Wang, Junxing Zhu, Yan Chen, Huizan Wang, and Weimin Zhang.
2022. "A Hybrid Deep Learning Model for the Bias Correction of SST Numerical Forecast Products Using Satellite Data" *Remote Sensing* 14, no. 6: 1339.
https://doi.org/10.3390/rs14061339