A Non-stationary NDVI Time Series with Big Data: A Deep Learning Approach

Abstract

AbstractIn this paper, a suitable method to forecast the normalized difference vegetation index (NDVI) time series (TS) is deep learning in the context of remote sensing big data. In fact, we proposed a non-stationary NDVI time series forecasting model by combining big data system, wavelet transform (WT) and long short-term memory (LSTM) neural network. In the first step, the MapReduce was investigated to extract NDVI TS. Then, WT and LSTM were applied for the analysis and forecasting of the NDVI. Our results show that our methodology gives a good result for forecasting NDVI time series in terms of root mean square error (RMSE) of 0.05.KeywordsNon-stationary time seriesBig dataDeep learningWavelet transform