Transfer learning for neural network model in chlorophyll-a dynamics prediction

  • Wenchong Tian
  • Zhenliang LiaoEmail author
  • Xuan Wang
Research Article


Neural network models have been used to predict chlorophyll-a concentration dynamics. However, as model generalization ability decreases, (i) the performance of the models gradually decreases over time; (ii) the accuracy and performance of the models need to be improved. In this study, Transfer learning (TL) is employed to optimize neural network models (including feedforward neural networks (FNN), recurrent neural networks (RNN) and long short-term memory (LTSM)) and overcome these problems. Models using TL are able to reduce the influence of mutable data distribution and enhance generalization ability. Thus, it can improve the accuracy of prediction and maintain high performance in long-term applications. Also, TL is compared with parameter norm penalties (PNP) and dropout—two other methods used to improve model generalization ability. In general, TL has a better prediction effect than PNP and dropout. All the models, including FNN with different architectures, RNN and LSTM, as well as models optimized by PNP, dropout, and TL, are applied to an estuary reservoir in eastern China to predict chlorophyll-a dynamics at 5-min intervals. According to the results of this study, (i) models with TL produce the best prediction results; (ii) the original models and the models with PNP and dropout lose their ability to predict within 3 months, while TL models retain a high prediction accuracy.


Transfer learning Chlorophyll-a dynamics Feedforward neural networks Recurrent neural network Long short-term memory 


Funding information

This study was financially supported by The National Key Research and Development Program of China (Grant No.2016YFE0123300 and 2017YFC0405406) and the National Natural Science Foundation of China (Grant No. 51578396).


  1. Coad P, Cathers B, Ball JE, Kadluczka R (2014) Proactive management of estuarine algal blooms using an automated monitoring buoy coupled with an artificial neural network. Environ Model Softw 61(C):393–409CrossRefGoogle Scholar
  2. García-Camacho F, López-Rosales L, Sánchez-Mirón A, Belarbi EH, Chisti Y, Molina-Grima E (2016) Artificial neural network modeling for predicting the growth of the microalga karlodinium veneficum. Algal Res 14:58–64CrossRefGoogle Scholar
  3. Graves A. 2008 Offline handwriting recognition with multidimensional recurrent neural networks [C]// International Conference on Neural Information Processing Systems. Curran Associates Inc 2008:545-552Google Scholar
  4. Graves A, Jaitly N 2014 Towards end-to-end speech recognition with recurrent neural networks [C]// International Conference on Machine Learning. 2014:1764-1772Google Scholar
  5. Grover JP (1991) Resource competition in a variable environment: phytoplankton growing according to the variable-internal-stores model. Am. Nat. 1991:811–835CrossRefGoogle Scholar
  6. Janssen F, Neumann T, Schmidt M (2004) Inter-annual variability in cyanobacteria blooms in the Baltic Sea controlled by wintertime hydrographic conditions [J]. Mar Ecol Prog 275(1):59–68CrossRefGoogle Scholar
  7. Jeong KS 2000 Statistical evaluation and application of artificial neural networks on water quality of the lower Nakdong River. MSc thesis, Pusan National University, Busan, p. 74Google Scholar
  8. Jeong KS, Kim DK, Whigham P et al (2003) Modelling microcystis aeruginosa, bloom dynamics in the Nakdong River by means of evolutionary computation and statistical approach [J]. Ecological Modelling 161(1–2):67–78CrossRefGoogle Scholar
  9. Jeong KS, Recknagel F, Joo GJ 2006 Prediction and elucidation of population dynamics of the blue-green algae microcystis aeruginosa, and the diatom Stephanodiscus hantzschii, in the Nakdong river-reservoir system (South Korea) by a recurrent artificial neural network [M]// Ecological Informatics. Springer Berlin Heidelberg, 2006:196-213Google Scholar
  10. Kingma D, Ba J 2014 Adam: a method for stochastic optimization [J]. Computer ScienceGoogle Scholar
  11. Lecun Y, Bengio Y, Hinton G (2015) Deep learning [J]. Nature 521(7553):436–444CrossRefGoogle Scholar
  12. Lilover MJ, Stips A (2008) The variability of parameters controlling the cyanobacteria bloom biomass in the Baltic Sea [J]. J Mar Syst 74(20):S108–S115CrossRefGoogle Scholar
  13. Nausch M, Nausch G, Mohrholz V et al (2012) Is growth of filamentous cyanobacteria supported by phosphate uptake below the thermocline [J]. Estuar Coast Shelf Sci 99(99):50–60CrossRefGoogle Scholar
  14. Oh HM, Ahn CY, Lee JW, Ahn CY, Lee JW, Chon TS, Choi KH, Park YS (2007) Community patterning and identification of predominant factors in algal bloom in Daechung Reservoir (Korea) using artificial neural networks [J]. Ecol Model 203(1-2):109–118CrossRefGoogle Scholar
  15. Pan SJ, Yang Q (2010) A Survey on Transfer Learning [J]. IEEE Trans Knowl Data Eng 22(10):1345–1359CrossRefGoogle Scholar
  16. Pappu JSM, Vijayakumar GK, Ramamurthy V (2013) Artificial neural network model for predicting production of Spirulina platensis, in outdoor culture [J]. Bioresour Technol 130(2):224–230CrossRefGoogle Scholar
  17. Recknagel F (1997) ANNA – artificial neural network model for predicting species abundance and succession of blue-green algae. Hydrobiologia. 349(1):47–57CrossRefGoogle Scholar
  18. Recknagel F (2001) Applications of machine learning to ecological modelling [J]. Ecol Model 146(1):303–310CrossRefGoogle Scholar
  19. Recknagel F, French M, Harkonen P, Yabunaka KI (1997) Artificial neural network approach for modelling and prediction of algal blooms. Ecol Model 96(s 1–3):11–28CrossRefGoogle Scholar
  20. Seitzinger SP (1991) 1991. The effect of pH on the release of phosphorus from Potomac estuary sediments: implications for blue-green algal blooms. J. Estuarine Coastal Shelf Sci. 33(4):409–418CrossRefGoogle Scholar
  21. Srivastava N, Hinton G, Krizhevsky A et al (2014) Dropout: a simple way to prevent neural networks from overfitting [J]. J Mach Learn Res 15(1):1929–1958Google Scholar
  22. Tian W, Liao Z, Zhang J (2017) An optimization of artificial neural network model for predicting chlorophyll-a dynamics [J]. Ecol Model 364:42–52CrossRefGoogle Scholar
  23. Vahtera E, Laamanen M, Rintala J-M (2007) Use of different phosphorus sources by the bloom-forming cyanobacteria Aphanizomenon flos-aquae and Nodularia spumigena. Aquat Microb Ecol 46:225e237CrossRefGoogle Scholar
  24. Vitányi PMB (2013) Conditional Kolmogorov complexity and universal probability [J]. Theor Comput Sci 501:93–100CrossRefGoogle Scholar
  25. Wei B, Sugiura N, Maekawa T (2001) Use of artificial neural network in the prediction of algal blooms [J]. Water Res 35(8):2022CrossRefGoogle Scholar
  26. Wei L, Qin B, Zhu G (2014) Forecasting short-term cyanobacterial blooms in Lake Taihu, China, using a coupled hydrodynamic-algal biomass model. Ecohydrology 7(2):794–802 (9)CrossRefGoogle Scholar
  27. Zhang X, Recknagel F, Chen Q, Cao H, Li R (2014) Spatially-explicit modelling and forecasting of cyanobacteria growth in Lake Taihu by evolutionary computation. Ecol Model 306:216–225CrossRefGoogle Scholar
  28. Zhang Y, Huang JJ, Chen L, Lan Q (2015) Eutrophication forecasting and management by artificial neural network: a case study at Yuqiao Reservoir in North China. J Hydroinf 17:4CrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2019

Authors and Affiliations

  1. 1.UNEP-Tongji Institute of Environment for Sustainable Development, College of Environmental Science and EngineeringTongji UniversityShanghaiPeople’s Republic of China
  2. 2.Shanghai Institute of Pollution Control and Ecological SecurityShanghaiPeople’s Republic of China
  3. 3.Key Laboratory of Yangtze River Water Environment (Ministry of Education),Tongji UniversityShanghaiPeople’s Republic of China
  4. 4.College of Civil Engineering and ArchitectureXinjiang UniversityUrumqiPeople’s Republic of China

Personalised recommendations