Predicting summer rainfall in the Yangtze River basin with neural networks

Abstract

Summer rainfall in the Yangtze River basin is predicted using neural network techniques. Input variables (predictors) for the neural network are the Southern Oscillation Index (SOI), the East Atlantic/Western Russia (EA/WR) pattern, the Scandinavia (SCA) pattern, the Polar/Eurasia (POL) pattern and several indices calculated from sea surface temperatures (SST), sea level pressures (SLP) and snow data from December to April for the period from 1993 to 2002. The output variable of the neural network is rainfall from May to September for the period from 1994 to 2002, which was previously classified into six different regions by means of a principal component analysis (PCA). Rainfall is predicted from May to September 2002.The winter SST and SLP indices are identified to be the most important predictors of summer rainfall in the Yangtze River basin. The Tibetan Plateau snow depth, the SOI and the other teleconnection indices seem to be of minor importance for an accurate prediction. This may be the result of the length of the available time series, which does not allow a deeper analysis of the impact of multi‐annual oscillations.The neural network algorithms proved to be capable of explaining most of the rainfall variability in the Yangtze River basin. For five out of six regions, our predictions explain at least 77% of the total variance of the measured rainfall. Copyright © 2007 Royal Meteorological Society

Forecasting Victorian spring rainfall using ENSO and IOD: A comparison of linear multiple regression and nonlinear ANN

“…Many studies have tried to find the relationships between large-scale climate modes and rainfall in different parts around the world using different linear and nonlinear methods [1][2][3][4][5][6]. It is believed that Australian rainfall is affected by several major climate patterns.…”

Seasonal precipitation forecasts over China using monthly large-scale oceanic-atmospheric indices

“…The predictors used were six large-scale atmospheric indices including the Antarctic Oscillation, the Ural circulation, the East Asian circulation, the Southern Pacific sea level pressure, the meridional wind shear and the Yangtze River vorticity. Hartmann et al (2008) employed a neural network technique to forecast summer rainfall over the Yangtze River basin, using a large number of predictors including the Southern Oscillation Index, the East Atlantic and Western Russia pattern, the Scandinavia pattern, the Polar and Eurasia pattern and 11 indices derived from sea surface temperatures (SSTs), sea level pressures and snow data. Wu et al (2009) proposed a regression model to forecast an East Asian summer monsoon (EASM) index using a North Atlantic Oscillation index, an ENSO developing index and an ENSO decaying index as predictors.…”

Neural Network Based Seasonal Predictions of Lake-Effect Snowfall

“…Here, "backpropagation" training is performed: weights are adjusted in order to reduce the error between actual and desired network outputs backward from the output layer to the input layer (Backhaus et al, 2003). As in Hartmann et al (2008aHartmann et al ( , 2008b) the architecture of the NN was determined by a trial and error approach. The starting point was a small network consisting of one hidden layer and four processing elements (PEs).…”