Showing papers by "Linda See published in 2006"
TL;DR: The results showed that networks trained with pre-processed data performed better than networks trained on undecomposed, noisy raw signals, and the best results were obtained using the data partitioning technique.
Abstract: The evaluation of surface water resources is a necessary input to solving water management problems. Neural network models have been trained to predict monthly runoff for the Tirso basin, located in Sardinia (Italy) at the S. Chiara section. Monthly time series data were available for 69 years and are characterized by non-stationarity and seasonal irregularity, which is typical of a Mediterranean weather regime. This paper investigates the effects of data preprocessing on model performance using continuous and discrete wavelet transforms and data partitioning. The results showed that networks trained with pre-processed data performed better than networks trained on undecomposed, noisy raw signals. In particular, the best results were obtained using the data partitioning technique.
229 citations
TL;DR: This paper presents a methodology for the comparison of different land cover datasets and illustrates how this can be extended to create a hybrid land cover product.
Abstract: This paper presents a methodology for the comparison of different land cover datasets and illustrates how this can be extended to create a hybrid land cover product. The datasets used in this paper are the GLC-2000 and MODIS land cover products. The methodology addresses: 1) the harmonization of legend classes from different global land cover datasets and 2) the uncertainty associated with the classification of the images. The first part of the methodology involves mapping the spatial disagreement between the two land cover products using a combination of fuzzy logic and expert knowledge. Hotspots of disagreement between the land cover datasets are then identified to determine areas where other sources of data such as TM/ETM images or detailed regional and national maps can be used in the creation of a hybrid land cover dataset
80 citations
TL;DR: A symbiotic adaptive neuro-evolutionary algorithm is used to breed neural network models for the River Ouse catchment and it is shown that at longer lead times the evolved neural networks outperform the conventional ones in terms of overall performance.
44 citations
01 Sep 2006
TL;DR: In this article, the application of multiple linear regression and three different data-driven modeling techniques to river level forecasting for the river Ouse catchment in northern England was presented. But the results show that the data driven approaches generally outperformed the statistical approach and that M5 model trees have great potential for the development of transparent river-level forecasting models.
Abstract: This paper outlines the application of multiple linear regression and three different data-driven modeling techniques to river level forecasting for the river Ouse catchment in northern England. Lead times of 6 and 24 hours ahead were modelled. The results show that the data driven approaches generally outperformed the statistical approach and that M5 model trees have great potential for the development of transparent river level forecasting models.
14 citations
17 Mar 2006
1 citations