Review Articles

Application of neural network to model rainfall pattern of Ethiopia

Gemechu Abdisa Atomsa ,

School of Statistics, East China Normal University, Shanghai, People's Republic of China

Yingchun Zhou

KLATASDS-MOE, School of Statistics, East China Normal University, Shanghai, People's Republic of China

Pages | Received 16 Feb. 2022, Accepted 02 Oct. 2022, Published online: 31 Oct. 2022,
  • Abstract
  • Full Article
  • References
  • Citations

In this paper, we have constructed Artificial Neural Network models which could capture rainfall pattern of Ethiopia. The data was collected from 147 stations across Ethiopia. Seven homogenized rainfall stations have been created based on both local and global patterns of datasets. Back-of-Word algorithm was used for extracting patterns of the datasets. K-means algorithm was used for clustering purpose. Each of the data of homogenized regions was interpolated using a spatial average. Two time series models, ARMA and Facebook's Prophet, have been fitted for each of spatial averages as baseline models. Both have been shown to perform weak for generalization purpose as spatially averaged datasets lose their strong seasonal pattern. On the other hand, the proposed Long Short Term Memory (LSTM) was found to be the best fitted model in comparison to the baseline models. The hyperparameters of the LSTM have been tuned to get optimal parameters. Besides, the RMSE of the baseline model was used as a benchmark for tuning the LSTM used.


  • Abbot, J., & Marohasy, J. (2018). Forecasting of medium-term rainfall using Artificial Neural Networks: Case studies from Eastern Australia. In T. V. Hromadka & P. Rao (Eds.), Engineering and Mathematical Topics in Rainfall (pp. 33–56). Books on Demand. 
  • Berhanu, B., Seleshi, Y., & Melesse, A. M. (2014). Surface water and groundwater resources of ethiopia: Potentials and challenges of water resources development. In A. Melesse, W. Abtew, & S. Setegn (Eds.), Nile River Basin (pp. 97–117). Springer. 
  • Box, G. E. P., & Jenkins, G. M (1976). Time series analysis. Forecasting and control. In G. C. Reinsel (Ed.), Holden-day series in time series analysis (Revised ed.) Holden-Day. 
  • Brownlee, J. (2016). Time series prediction with lstm recurrent neural networks in Python with Keras. Available at: 
  • Carpenter, B., Gelman, A., Hoffman, M. D., Lee, D., Goodrich, B., Betancourt, M., Brubaker, M., Guo, J., Li, P., & Riddell, A. (2017). Stan: A probabilistic programming language. Journal of Statistical Software76(1), 1–32. 
  • Chen, C., & Liu, L.-M. (1993). Joint estimation of model parameters and outlier effects in time series. Journal of the American Statistical Association88(421), 284–297. 
  • Dabakoglu, C. (Jun 2019). Time series forecasting—Arima, LSTM, Prophet with python. Online. Retrieved June 30, 2019. 
  • Golden, R. M. (1996). Mathematical methods for neural network analysis and design. MIT Press. 
  • Greff, K., Srivastava, R. K., Koutník, J., Steunebrink, B. R., & Schmidhuber, J. (2017). Lstm: A search space Odyssey. IEEE Transactions on Neural Networks and Learning Systems28(10), 2222–2232. 
  • Haining, R. P., & Haining, R. (2003). Spatial data analysis: Theory and practice. Cambridge University Press. 
  • Hijmans, R. J. (2017). Geosphere: Spherical trigonometry. R package version 1.5-7. 
  • Hochreiter, S. (1998). Recurrent neural net learning and vanishing gradient. International Journal of Uncertainity, Fuzziness and Knowledge-Based Systems6(2), 107–116. 
  • Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation9(8), 1735–1780. 
  • Hyndman, R. J. (2013). FPP data for “forecasting: Principles and practice”. R package version 0.5. 
  • Hyndman, R., Athanasopoulos, G., Bergmeir, C., Caceres, G., Chhay, L., O'Hara-Wild, M., Petropoulos, F., Razbash, S., Wang, E., & Yasmeen, F. (2018). forecast: Forecasting functions for time series and linear models. R package version 8.4. 
  • Hyndman, R. J., & Khandakar, Y. (2008). Automatic time series forecasting: The forecast package for R. Journal of Statistical Software27(3), 1–22. 
  • Jiang, Y., & Zhang, Y. (2018). Exploration of predicting power of Arima, Facebook Prophet and lstm on time series. Stanford University. Online. Retrieved June 30, 2019. 
  • Lin, J., Keogh, E., Lonardi, S., & Patel, P. (2002). Finding motifs in time series. In Proc. of the 2nd workshop on temporal data mining (pp. 53–68). 
  • Lin, J., Keogh, E., Wei, L., & Lonardi, S. (2007). Experiencing sax: A novel symbolic representation of time series. Data Mining and Knowledge Discovery15(2), 107–144. 
  • Lin, J., & Li, Y. (2009). Finding structural similarity in time series data using bag-of-patterns representation. In M. Winslett (Ed.), International conference on scientific and statistical database management (pp. 461–477). Springer. 
  • Longley, P. A., Goodchild, M. F., Maguire, D. J., & Rhind, D. W. (2005). Geographic information systems and science (2nd ed.). John Wiley & Sons.
  • Shen, S. S. P. (2017). R programming for climate data analysis and visualization: Computing and plotting for NOAA data applications. John Wiley & Sons. 
  • Strauß, M. (2018). Time series forecasting with LSTMS and Prophet, nov. Online. Retrieved June 30, 2019.
  • Taylor, S. J., & Letham, B. (2018a). Forecasting at scale. The American Statistician72(1), 37–45.
  • Taylor, S. J., & Letham, B. (2018b). Prophet: Automatic forecasting procedure. R package version 0.4. 
  • Tsidu, G. M. (2012). High-resolution monthly rainfall database for ethiopia: Homogenization, reconstruction, and gridding. Journal of Climate25(24), 8422–8443. 
  • Veenstra, J. Q. (2012). Persistence and anti-persistence: Theory and software [PhD thesis]. Western University. 

To cite this article: Gemechu Abdisa Atomsa & Yingchun Zhou (2023) Application of neural network to model rainfall pattern of Ethiopia, Statistical Theory and Related Fields, 7:1, 69-84, DOI: 10.1080/24754269.2022.2136266 To link to this article: