XVIII
Contents
Part IV Prediction and Elucidation of Lake and Marine
Ecosystems ....................................................................................... 247
13. A Comparison between Neural Network Based and Multiple
Regression Models in Chlorophyll-a Estimation .................. 249
13.1 Introduction ......................................................................................... 249
13.1.1Eutrophication in Water Bodies and Relevant Models ............. 249
13.1.2Artificial Neural Networks ........................................................ 250
13.1.3The Use of Artificial Neural Networks in Environmental
Modelling .................................................................................. 251
13.2Data and Lakes .................................................................................... 251
13.3Methodology ....................................................................................... 253
13.3.1.1 Artificial Neural Network Approach ........................................... 254
13.3.1.2 Training Method .......................................................................... 254
13.3.1.3 Data Pre-Processing ..................................................................... 256
13.3.1.4 Improving Generalisation ............................................................ 256
13.3.2 Multiple Regression Modelling Approach ................................... 257
13.4Results ................................................................................................. 257
13.5Conclusions and Recommendations ................................................... 260
13.5.1 Conclusions ................................................................................. 260
13.5.2 Recommendations ........................................................................ 261
Acknowledgments ................................................................................. 261
References ............................................................................................. 262
14. A Generic Artificial Neural Network Model for Dynamic
Predictions of Aigal Abundance in Freshwater Lakes ......... 265
14.1 Introduction ......................................................................................... 265
14.2Issues to be Adddressed by Algal Bloom Models ............................... 266
14.2.1 Input Layer Design ...................................................................... 266
14.2.2 Control of Overfitting .................................................................. 267
14.2.3 Linear versus Non-Linear Decision Boundaries .......................... 269
14.3Implementation of the Generic ANN Algal Bloom Model... .............. 269
14.3.1 Input Layer Design ......................................................................... 269
14.3.2Model Approximation (Training ..................................................... 272
14.3 .3Control of Overfitting ...................................................................... 273
14.3.4Model Assessment ........................................................................... 274
14.4Results ................................................................................................ 275
14.5Discussion ........................................................................................... 285
14.6 Conclusion ......................................................................................... 286
Acknowledgements ........................................................................... 287
References ......................................................................................... 287
15. Predictive Rules for Phytoplankton Dynamics in Freshwater
Lakes Discovered by Evolutionary Algorithms .................... 291
15.1 Introduction ......................................................................................... 291
15.1.1 Know ledge Generalisation and Representation ........................... 292
Contents
Part IV Prediction and Elucidation of Lake and Marine
Ecosystems ....................................................................................... 247
13. A Comparison between Neural Network Based and Multiple
Regression Models in Chlorophyll-a Estimation .................. 249
13.1 Introduction ......................................................................................... 249
13.1.1Eutrophication in Water Bodies and Relevant Models ............. 249
13.1.2Artificial Neural Networks ........................................................ 250
13.1.3The Use of Artificial Neural Networks in Environmental
Modelling .................................................................................. 251
13.2Data and Lakes .................................................................................... 251
13.3Methodology ....................................................................................... 253
13.3.1.1 Artificial Neural Network Approach ........................................... 254
13.3.1.2 Training Method .......................................................................... 254
13.3.1.3 Data Pre-Processing ..................................................................... 256
13.3.1.4 Improving Generalisation ............................................................ 256
13.3.2 Multiple Regression Modelling Approach ................................... 257
13.4Results ................................................................................................. 257
13.5Conclusions and Recommendations ................................................... 260
13.5.1 Conclusions ................................................................................. 260
13.5.2 Recommendations ........................................................................ 261
Acknowledgments ................................................................................. 261
References ............................................................................................. 262
14. A Generic Artificial Neural Network Model for Dynamic
Predictions of Aigal Abundance in Freshwater Lakes ......... 265
14.1 Introduction ......................................................................................... 265
14.2Issues to be Adddressed by Algal Bloom Models ............................... 266
14.2.1 Input Layer Design ...................................................................... 266
14.2.2 Control of Overfitting .................................................................. 267
14.2.3 Linear versus Non-Linear Decision Boundaries .......................... 269
14.3Implementation of the Generic ANN Algal Bloom Model... .............. 269
14.3.1 Input Layer Design ......................................................................... 269
14.3.2Model Approximation (Training ..................................................... 272
14.3 .3Control of Overfitting ...................................................................... 273
14.3.4Model Assessment ........................................................................... 274
14.4Results ................................................................................................ 275
14.5Discussion ........................................................................................... 285
14.6 Conclusion ......................................................................................... 286
Acknowledgements ........................................................................... 287
References ......................................................................................... 287
15. Predictive Rules for Phytoplankton Dynamics in Freshwater
Lakes Discovered by Evolutionary Algorithms .................... 291
15.1 Introduction ......................................................................................... 291
15.1.1 Know ledge Generalisation and Representation ........................... 292