DOI QR코드

DOI QR Code

Prediction of pollution loads in the Geum River upstream using the recurrent neural network algorithm

  • Lim, Heesung (Agricultural and Rural Engineering, Chungnam National University) ;
  • An, Hyunuk (Agricultural and Rural Engineering, Chungnam National University) ;
  • Kim, Haedo (Rural research institute, Korea Rural Community Corporation) ;
  • Lee, Jeaju (Rural research institute, Korea Rural Community Corporation)
  • Received : 2018.10.25
  • Accepted : 2018.11.13
  • Published : 2019.03.01

Abstract

The purpose of this study was to predict the water quality using the RNN (recurrent neutral network) and LSTM (long short-term memory). These are advanced forms of machine learning algorithms that are better suited for time series learning compared to artificial neural networks; however, they have not been investigated before for water quality prediction. Three water quality indexes, the BOD (biochemical oxygen demand), COD (chemical oxygen demand), and SS (suspended solids) are predicted by the RNN and LSTM. TensorFlow, an open source library developed by Google, was used to implement the machine learning algorithm. The Okcheon observation point in the Geum River basin in the Republic of Korea was selected as the target point for the prediction of the water quality. Ten years of daily observed meteorological (daily temperature and daily wind speed) and hydrological (water level and flow discharge) data were used as the inputs, and irregularly observed water quality (BOD, COD, and SS) data were used as the learning materials. The irregularly observed water quality data were converted into daily data with the linear interpolation method. The water quality after one day was predicted by the machine learning algorithm, and it was found that a water quality prediction is possible with high accuracy compared to existing physical modeling results in the prediction of the BOD, COD, and SS, which are very non-linear. The sequence length and iteration were changed to compare the performances of the algorithms.

Keywords

CNNSA3_2019_v46n1_67_f0001.png 이미지

Fig. 1. Structure of recurrent neural networks. A, weight; ht, hidden layer; h0, hidden layer 0; h1, hidden layer 1; h2, hidden layer 2; Xt, input layer; X0, input layer 0; X1, input layer 1; X2, input layer 2.

CNNSA3_2019_v46n1_67_f0002.png 이미지

Fig. 2. Structure of LSTM (long short-term memory) network. A, weight; ht-1, hidden layer (t-1); ht, hidden layer; ht+1, hidden layer (t+1); Xt-1, input layer (t-1); Xt, input layer; Xt+1, input layer (t+1); σ, sigmoid layer; tanh, hyperbolic tangent; X, multiplication; +, plus.

CNNSA3_2019_v46n1_67_f0003.png 이미지

Fig. 3. Study area (Okcheon-gun, Chungcheongbuk-do, Korea).

CNNSA3_2019_v46n1_67_f0004.png 이미지

Fig. 4. RNN (recurrent neural network) scatter plot of water quality. BOD, biochemical oxygen demand; COD, chemical oxygen demand; SS, suspended solids.

CNNSA3_2019_v46n1_67_f0005.png 이미지

Fig. 5. LSTM (long short-term memory) scatter plot of water quality. BOD, biochemical oxygen demand; COD, chemical oxygen demand; SS, suspended solids.

Table 1. Meaning of R2 and RMSE (root mean square error) index.

CNNSA3_2019_v46n1_67_t0001.png 이미지

Table 2. Model performance results of the BOD (biochemical oxygen demand) RNN (recurrent neural network) model.

CNNSA3_2019_v46n1_67_t0002.png 이미지

Table 3. Model performance results of the COD (chemical oxygen demand) RNN (recurrent neural network) model.

CNNSA3_2019_v46n1_67_t0003.png 이미지

Table 4. Model performance results of the SS (suspended solids) RNN (recurrent neural network) model.

CNNSA3_2019_v46n1_67_t0004.png 이미지

Table 5. Model performance results of the BOD (biochemical oxygen demand) LSTM (long short-term memory) model.

CNNSA3_2019_v46n1_67_t0005.png 이미지

Table 6. Model performance results of the COD (chemical oxygen demand) LSTM (long short-term memory) model.

CNNSA3_2019_v46n1_67_t0006.png 이미지

Table 7. Model performance results of the SS (suspended solids) LSTM (long short-term memory) model.

CNNSA3_2019_v46n1_67_t0007.png 이미지

References

  1. Brown LC, Barnwell TO. 1987. The enhanced stream water quality models QUAL2E and QUAL2E-UNCAS. US Environmental Protection Agency, Georgia, USA.
  2. Hochreiter S, Schmidhuber J. 1997. Long short-term memory. Neural computation 9:1735-1780. https://doi.org/10.1162/neco.1997.9.8.1735
  3. Jeong HJ, Lee SJ, Lee HK. 2002. Water quality forecasting of Chungju Lake using artificial neural network algorithm. Journal of the Environmental Sciences 11:201-207. [in Korean] https://doi.org/10.5322/JES.2002.11.3.201
  4. Jeong HW, Ki SJ, Jeon DJ, Kim JH. 2016. Development of system based on weather radar images for predicting rainfall events using machine learning models in watershed of Yeong-san River. pp. 283-284. Proceedings of 2016 Joint Conference of Korean Society on Water Environment and Korean Society of Water & Wastewater, Korea. [in Korean]
  5. Kim MS, Han JS. 2002. Artificial neural networks for forecasting of short-term river water quality. Journal of the Korean Geo-Environmental Society 3:11-17. [in Korean]
  6. Lee GH, Kim IH, Moon BS. 2001. Water quality prediction of river using intelligent model. pp. 179-182. Proceedings of 2001 Joint Conference of Korean Society of Water & Wastewater and Korean Society on Water Environment, Korea. [in Korean]
  7. Oh CR, Park SC, Lee HM, Pyo YP. 2002. A forecasting of water quality in the Youngsan River using neural network. Jouranl of The Korean Society of Civil Engineers 22:371-382. [in Korean]
  8. Park SC, Lee HM, Oh CR. 2000. The application of artificial neural network for forecasting of DO, BOD concentration. Journal of Environmental Research 5:31-48. [in Korean]
  9. Roh TH, Lee TH, Han IG. 2005. Forecasting the volatility of KOSPI 200 using neural network-financial time series model. Korean Management 34:683-713. [in Korean]
  10. Seo IW, Yun SH. 2016. Forecasting water quality by ANN model at the downstream of Cheongpyeong Dam. pp. 41-42. Korean Society of Civil Engineers (KSCE) 2017 Convention, Seoul, Korea. [in Korean]
  11. Shin DS, Kwun SK. 1997. Water quality modeling for Bokha Stream by WASP5 model. Korean Journal of Environmental Agriculture 16:233-238. [in Korean]
  12. Seo DI, Lee JH, Lee EH, Ko IH. 2004. Analysis on errors of water quality modeling for the Geum River down stream areas using QUAL2E. Korean Society of Environmental Engineers 26:933-940. [in Korean]
  13. Shin CM, Kim KH. 2016. Operational water quality forecast for the Nakdong River basin using HSPF watershed model. Journal of Korean Society on Water Environment 32:570-581. [in Korean] https://doi.org/10.15681/KSWE.2016.32.6.570
  14. Tran Q, Song S. 2017. Water level forecasting based on deep learning: A use case of Trinity River-Taxas-The United States. Journal of Korean Institute of Information Scientists Engineers 44:607-612. [in Korean]

Cited by

  1. A Review of the Artificial Neural Network Models for Water Quality Prediction vol.10, pp.17, 2020, https://doi.org/10.3390/app10175776
  2. Flood prediction in the Namgang Dam basin using a long short-term memory (LSTM) algorithm vol.47, pp.3, 2019, https://doi.org/10.7744/kjoas.20200036
  3. Prediction of the DO concentration using the machine learning algorithm: case study in Oncheoncheon, Republic of Korea vol.47, pp.4, 2019, https://doi.org/10.7744/kjoas.20200086
  4. The Use of Artificial Neural Networks to Predict the Physicochemical Characteristics of Water Quality in Three District Municipalities, Eastern Cape Province, South Africa vol.18, pp.10, 2021, https://doi.org/10.3390/ijerph18105248