Pattern clustering of forest fires based on meteorological variables and its classification using hybrid data mining methods

This paper outlines two hybrid approaches to investigate the nonlinear relationship between size of a forest fire and meteorological variables (temperature, relative humidity, wind speed and rainfall). Self organizing map was used to cluster the historical meteorological variables. The clustered data were then used as inputs for two different approaches, the back-propagation neural network and the rule generation approaches. A back-propagation neural network was trained based on these inputs to classify the output (burnt area) in categorical form, namely; small, medium, large and extremely large. Several sets of rules were also generated from the data clustered by the self organizing map. Experimental results showed that both approaches gave considerable accuracy. Back-propagation neural network achieved a higher rate of accuracy than rule generation approach because the rule generation approach could not predict any criterion that goes beyond the set of rules .

[1]  G. Clifford Carter,et al.  Empirical results of using back-propagation neural networks to separate single echoes from multiple echoes , 1993, IEEE Trans. Neural Networks.

[2]  M. W Gardner,et al.  Artificial neural networks (the multilayer perceptron)—a review of applications in the atmospheric sciences , 1998 .

[3]  Andrew Rowell,et al.  Global review of forest fires , 2000 .

[4]  Jean-François Mas,et al.  Modelling deforestation using GIS and artificial neural networks , 2004, Environ. Model. Softw..

[5]  M. Flannigan,et al.  Fire weather index system components for large fires in the Canadian boreal forest , 2004 .

[6]  F. Sunar,et al.  Forest fire analysis with remote sensing data , 2001 .

[7]  Osmar R. Zaïane,et al.  Application of Data Mining Techniques for Medical Image Classification , 2001, MDM/KDD.

[8]  Orbita Roswintiarti,et al.  Development of the Indonesian and Malaysian Fire Danger Rating Systems , 2006 .

[9]  S. Džeroski,et al.  LEARNING TO PREDICT FOREST FIRES WITH DIFFERENT DATA MINING TECHNIQUES , 2006 .

[10]  Teuvo Kohonen,et al.  Self-organized formation of topologically correct feature maps , 2004, Biological Cybernetics.

[11]  Nigel J. Tapper,et al.  The Sensitivity of Australian Fire Danger to Climate Change , 2001 .

[12]  John William McManus,et al.  Design and analysis techniques for concurrent blackboard systems , 1996, IEEE Trans. Syst. Man Cybern. Part A.

[13]  R. Soni.  The International Union for the Conservation of Nature and Natural Resources , 1960, Oryx.

[14]  Mohammad Bagher Menhaj,et al.  Training feedforward networks with the Marquardt algorithm , 1994, IEEE Trans. Neural Networks.

[15]  Ku Ruhana Ku-Mahamud,et al.  Pattern Extraction and Rule Generation of Forest Fire using Sliding Window Technique , 2009, Comput. Inf. Sci..

[16]  Geoff West,et al.  BACKPROPAGATION NEURAL NETWORK FOR SOIL MOISTURE RETRIEVAL USING NAFE ’ 05 DATA : A COMPARISON OF DIFFERENT TRAINING ALGORITHMS , 2008 .

[17]  Chih-Cheng Hung,et al.  An artificial neural network for classifying and predicting soil moisture and temperature using Levenberg-Marquardt algorithm , 1999, Proceedings IEEE Southeastcon'99. Technology on the Brink of 2000 (Cat. No.99CH36300).

[18]  Jeremy S. Fried,et al.  Predicting the impacts of global warming on wildland fire , 1992 .

[19]  P. Cortez,et al.  A data mining approach to predict forest fires using meteorological data , 2007 .

[20]  A. M. Kalteh,et al.  Review of the self-organizing map (SOM) approach in water resources: Analysis, modelling and application , 2008, Environ. Model. Softw..