Advanced

Forecasting discharge in Amazonia using artificial neural networks

Uvo, Cíntia Bertacchi LU ; Tölle, Ute and Berndtsson, Ronny LU (2000) In International Journal of Climatology 20(12). p.1495-1507
Abstract

The Amazon, located in northern South America, is the world's largest river basin, and covers an area of about 6.5 million km2. The observed interannual variability in precipitation and water availability during its main discharge season has been shown to be influenced by Pacific and Atlantic Ocean sea surface temperatures (SSTs). However, the links between large-scale atmospheric motion and local and regional runoff patterns are essentially complex and still not fully understood. The processes involved are believed to be highly non-linear, spatially and temporally variable, and not easily described by physical or conceptual models. Artificial neural networks (NN) were trained to forecast discharge, one and two seasons in... (More)

The Amazon, located in northern South America, is the world's largest river basin, and covers an area of about 6.5 million km2. The observed interannual variability in precipitation and water availability during its main discharge season has been shown to be influenced by Pacific and Atlantic Ocean sea surface temperatures (SSTs). However, the links between large-scale atmospheric motion and local and regional runoff patterns are essentially complex and still not fully understood. The processes involved are believed to be highly non-linear, spatially and temporally variable, and not easily described by physical or conceptual models. Artificial neural networks (NN) were trained to forecast discharge, one and two seasons in advance, at ten river sites in Amazonia from Pacific and Atlantic Ocean SST anomalies. The NN with an input layer of eight neurons, one hidden layer with 20 neurons and a one-neuron output layer was trained using back-propagation with momentum and gradient descendent. Results confirmed that different oceanic regions have distinct influences on different parts of the Amazonian basin. Better forecasts for basins in the northern part of Amazonia were obtained from Pacific Ocean SST and from Atlantic Ocean SST for basins in the southern part. Correlation coefficients between observed and estimated discharge (validation) were as high as 0.76 at some of the sites studied. The inclusion of precipitation as input improved the forecast for sites where NN did not perform well with training by SST only as input. The results obtained during this study corroborate and improve results obtained previously by means of linear statistical methods. Copyright (C) 2000 Royal Meteorological Society.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Amazon River, Amazonia, Curua-Una River, Discharge, Discharge forecast, Neural networks, Sea surface temperature
in
International Journal of Climatology
volume
20
issue
12
pages
13 pages
publisher
John Wiley & Sons
external identifiers
  • scopus:0033757099
ISSN
0899-8418
DOI
10.1002/1097-0088(200010)20:12<1495::AID-JOC549>3.0.CO;2-F
language
English
LU publication?
yes
id
da64d98b-ea82-477b-972b-9d86794cd27c
date added to LUP
2018-11-01 12:24:42
date last changed
2019-11-25 09:24:48
@article{da64d98b-ea82-477b-972b-9d86794cd27c,
  abstract     = {<p>The Amazon, located in northern South America, is the world's largest river basin, and covers an area of about 6.5 million km<sup>2</sup>. The observed interannual variability in precipitation and water availability during its main discharge season has been shown to be influenced by Pacific and Atlantic Ocean sea surface temperatures (SSTs). However, the links between large-scale atmospheric motion and local and regional runoff patterns are essentially complex and still not fully understood. The processes involved are believed to be highly non-linear, spatially and temporally variable, and not easily described by physical or conceptual models. Artificial neural networks (NN) were trained to forecast discharge, one and two seasons in advance, at ten river sites in Amazonia from Pacific and Atlantic Ocean SST anomalies. The NN with an input layer of eight neurons, one hidden layer with 20 neurons and a one-neuron output layer was trained using back-propagation with momentum and gradient descendent. Results confirmed that different oceanic regions have distinct influences on different parts of the Amazonian basin. Better forecasts for basins in the northern part of Amazonia were obtained from Pacific Ocean SST and from Atlantic Ocean SST for basins in the southern part. Correlation coefficients between observed and estimated discharge (validation) were as high as 0.76 at some of the sites studied. The inclusion of precipitation as input improved the forecast for sites where NN did not perform well with training by SST only as input. The results obtained during this study corroborate and improve results obtained previously by means of linear statistical methods. Copyright (C) 2000 Royal Meteorological Society.</p>},
  author       = {Uvo, Cíntia Bertacchi and Tölle, Ute and Berndtsson, Ronny},
  issn         = {0899-8418},
  language     = {eng},
  month        = {11},
  number       = {12},
  pages        = {1495--1507},
  publisher    = {John Wiley & Sons},
  series       = {International Journal of Climatology},
  title        = {Forecasting discharge in Amazonia using artificial neural networks},
  url          = {http://dx.doi.org/10.1002/1097-0088(200010)20:12<1495::AID-JOC549>3.0.CO;2-F},
  doi          = {10.1002/1097-0088(200010)20:12<1495::AID-JOC549>3.0.CO;2-F},
  volume       = {20},
  year         = {2000},
}