Loading...

Using Echo State Networks for Modeling and Prediction of Drought Based on Remote Sensing Data

Mohammadinejad, Amir | 2012

1234 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 44037 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Jalili, Mahdi
  7. Abstract:
  8. Iran is regarded as a dry land and has suffered from extreme to severe drought conditions in recent years. Drought – which is mainly caused by shortage in rainfall – affects the normal life in Iran. Development of tools for effectively monitoring and predicting drought intensity might help the policy makers to reduce the vulnerability of the areas affected by drought. In this thesis, we showed that the intensity of drought can be predicted using satellite imagery data and recurrent neural networks. To this end, the standardized precipitation index (SPI) was chosen as an index for drought and normalized deviation of vegetation index (NDVI) as a remote sensing measure extracted from NOAA-AVHRR images. We then build a model which received the NDVI as input and outputted the SPI class. In other words, the model aimed at predicting the drought conditions based only on satellite imagery data.
    We used a number of classification methods in order to perform the prediction task. These included multilayer perceptrons (MLPs), support vector machines (SVMs) and recurrent networks based on reservoir computing (RC). RC methods – inspired by the mechanisms of real neuronal networks – are increasingly used in many applications. The drawback of these methods is the way their internal weights are optimized. In this works, we used Kronecker product in reducing the number of parameters to optimize. We then used a number of optimization algorithms including gradient descent, simulated annealing, genetic algorithms and differential evolution. RCs optimized by differential evolution showed superior performance than MLPs and SVMs. We expect these efforts to have positive influence on drought modeling in Iran
  9. Keywords:
  10. Drought ; Recurrent Neural Networks ; Forecasting ; Remote Sensing Data ; Echo State Networks

 Digital Object List