Correo Electrónico
DNINFOA - SIA
Bibliotecas
Convocatorias
Identidad U.N.
Publicado
Análisis Espectral Singular y redes neuronales aplicadas en Hidrología
Palabras clave:
Redes neuronales-Hidrología, AES-Análisis Espectral Singular, Predicción lluvias y caudales (es)Descargas
El presente trabajo considera dos metodologías que utilizan la dinámica no lineal (caos) y su aplicación en la predicción de lluvias y caudales. La motivación de estos trabajos está en la oscilación interanual del clima tropical conocida como El Niño-Oscilación del Sur (ENOS) es un fenómeno no lineal de clara influencia en la hidrología de Colombia.
La primera es el Análisis Espectral Singular (AES), metodología útil en el filtrado de señales, con una amplia aplicación en el campo de la electrónica y recientemente en la hidrometeorología. Se busca capturar la componente no lineal presente en la dinámica de las series temporales naturales mediante el estudio del atractor en el espacio fásico y la utilización de métodos tradicionales de predicción en modelos autorregresivos AR(P) para extrapolar el comportamiento dinámico futuro. Este procedimiento fue aplicado a la modelación de lluvias y caudales en diversas cuencas del departamento de Antioquia, así como caudales en el territorio nacional.
La otra metodología es la de las redes neuronales (modelo de retro-propagación), que han tomado gran auge en diferentes campos de la actividad investigativa, presentando nuevos caminos para el estudio de fenómenos no lineales y que es aplicada a la predicción de lluvias y caudales, considerando información de precipitación, caudales, manchas solares, temperatura superficial del mar (TSM), el Indice de Oscilación del Sur (IOS) y datos de la oscilación quasi-bienal del viento (OQB). Los resultados plantean la posibilidad de considerar las redes neuronales como métodos de predicción con un cierto éxito.
The El Nino-Southem Oscillation (ENOS) influence on Colombian hydrology motivated the study of two non-linear methodologies for rainfall and streamflow prediction.
The first methodology is Singular spectral analysis which is a very powerful method in nonlinear (chaos) analysis of time series. The main idea is to filter the deterministic nonlinear component of a time series and to model it by means of traditional time series methods (autoregressive methods). The original series is embedded in a higher dimensional space for the reconstruction of the attractor of the dynamic system producing the series. The method is equivalent to the expansion in Fourier series but instead of using prescribed sine and cosine functions empirical eigen functions are used. The eigen functions come from the eigen vectors of the covariance matrix of the series in the phase space. Predictions are made extrapolating tendencies in the attractor. The method is applied to rainfall data in Colombia.
The second method is Neural Network Analysis, currently a very popular tool for modeling a great variety of nonlinear phenomena. This work presents the application of neural networks to the prediction of streamflows and rainfall in Colombia, using the southern oscillation index, QBO index (Quasibiennal stratospheric wind tropical velocity) and Pacific sea surface temperatures as predictors.
Referencias
Akaike, H. (1969). Fitting autoregressive models for prediction, Ann? Inst. Stat. Math., 21, 243-247.
Allen, M. R., Read, P. L. y Smith, L. A. Bidecadal oscillations and the global temperature record, Nature 355 (1992) 686.
Amari, S. (1972). "Learning patterns and patterns sequences by self organizing nets of treshold elements," IEEE Trans. Computers, C-21,1197-1206.
Anderson, J.A., Silverstein, J.W., Ritz, S.A., y Jones, R.S. (1977). "Distinctive features, categorial perception, and probability learning: some applications of a neural model, "Psych. Review, 84, 413-451, y reimpreso en J.A. Anderson y E. Rosenfeld (Eds.) (1998). Neurocomputing, MIT Press, Cambridge, MA.
Box, G.E.P. y Jenkins, G.M. (1970). Time Series Analysis, Forecasting and Control, Holden-Day, San Francisco.
Broomhead, D. S., y King, G. P. (1986). Extracting qualitative dynamics from experimental data, Physica D, 20,217- 236.
Burg, J.P. (1967). Maximun entropy spectral analysis, Presentation at 37th an- naul meeting of the society of exploratory Geophysicists, Oklahoma City, reimpreso en Modem Spectrum Analysis (1968), D. G. Childers (ed.), 43-48.
Carpenter, G.A., y Grossberg, S. (March, 1988). "The ART of adaptive pattern recognition," Computer, 77-88.
Carvajal, L. F. Acerca de la Predicción No lineal en Hidrología. Tesis de Magíster en Aprov. de Rec. Hid., Facultad de Minas, Medellin, 1994. 136 p.
Colebrook, J. M. Continuous plankton records: zooplankton and environment, North-East Atlantic and North Sea, 1948-1975, Oceanol. Acta, 1, 9-23, 1978.
Eberhart, Russell C. y Dobbins, Roy W (1990). Neural Network PC Tools, A Practical Guide. Academic Press, San Diego, California, 35-49.
Farley, B. y Clarck, W. A. (1954). "Simulation of self-organizing systems by digital computer," IRE Transaction of information Theory, 4,76-84.
Fraedrich, k. (1986). Estimating the dimensions of weather and climate attractors, J. Atmos. Sci., 43,419-432.
Fukishima, K., Miyake, S. y Ito, T. (1975). "Neocognitron: a neural network model for a mechanism of visual pattern recognition," IEEE Transactions on Systems, Man, and Cybernetics, SMC- 13, 826-834.
Hopfield, J.J. (1982). "Neural network and physical systems with emergent collective computational abilities", Proc. of the National Academy of Sciences, USA, 79, 2554-2558.
Hopfield, J.J., y Tank, D. (1986). "Computing with neural circuits: A model," Science, 233, 625-633.
James, W. (1980). Phychology, Holt, New York, chapter XVI, 253-279.
Keppenne, C.L. y Ghil, M. (1992). Adaptive filtering and prediction of the southern oscillation index. J. Geophys. Res., sub judice.
Kohonen, T. (1972). "Correlation matrix memories," IEEE Trans. Computer C- 21,353-359.
Kosko, B. (1987c). "Fuzzy Associative Memories," Fuzzy Expert Systems, Addison* Wesley, Reading, MA.
Kosko, B. (1987a). "Adaptive bidirectional associative memories," Applied Optics, 26,4947-4960.
Kosko, B. (1987b). "Constructing and Associative Memory," BYTE, Sept., 137- 144.
Kosko, B. (1986). "Fuzzy cognitive maps," International Journal of Man-Machine Studies, 24,65-75..
LeCun, Y. (1986). "Learning processes in an asymmetric treshold network," Disordered Systems and Biological Organization, E. Beinenstock, Fogelman Souli, and G. weisbuch (Eds.), Springer, Berlin.
Loöve, M. (1962). Probability Theory, 3rd ed, Van Nostrand, Princeton.
Maren, Alianna; Harston, Craig y Pap, Robert (1990). Handbook of Neural Computing Applications. Academic Press, San Diego, California, 13-28,45-57.
McCulloch, W.S. y Pitts, W. (1943). A logical calculos of the ideas immanent in nervous activity, "Bulletin of Mathematical Biophysics, 5,115-133.
Parker, D. (1985). "Learning logic," Technical report TR-87, Center for Computational Research in Economics and Ma- nagement Science, MIT Press, Cambridge, MA.
Penland, C., Ghil, M. y Weickmann, K. M. (1991). Adaptive Filtering and maxi- mun entropy spectra with application to changes in atmospheric angular momentum. J. Geoph. Res., 96, No. D12, 22,659-22,671.
Press, William H. Numerical Recipes The Art of Scientific Computing. Cambridge University Press, 1989.
Rochester, N., Holland, J. H., Haibt, L. H. y Duda, W. L. (1956). "Test on a cell asembly theory of the action of the brain, using a large digital computer, "IRE Transactions of Information Theory, IT-2,80-93.
Rosenblatt, F. (1958). "The Perceptron: A probabilistic model for information so- targe and organization in the brain," Psych. Review, 65,386-408, y reimpreso en Neurocomputing, J. Anderson y E. Rosenfeld (Eds.) (1988). MIT Press, Cambridge, MA., 92-114.
Rumelhart, D. E. y Me Clellandlland, "Parallel Distributed Processing, Explorations in the Microstructure of the Cognition. Vol. 1: Foundations." MIT Press, Cambridge, MA, 1986.
Tank, D. W., y Hopfield, J.J. (1987). Collective computation in neuronlike circuits," Scientific American, 257, 104- 114,158.
Vautard, R., Yiou, P. y Ghil, M. (1992). Singular-spectrum analysis: A toolkit for short, noisy chaotic signals. Physica D 58,95-126.
Vautard, R. y Ghil, M. (1889). Singular spectrum analysis in nonlinear dynamics, with applitions to paleoclimatic time series. Physica D 35,395-424.
Werbos, P. (1974). Beyond Regression: New Tools for Prediction and Analysis in the Behavioral Sciences. Ph.D. dissertation, Harvard University, Cambridge, MA.
