Presenter/Author Information

Edwin Roehl
John Risley
Jana Stewart
Matthew Mitro

Keywords

clustering, classification, neural network, model

Start Date

1-7-2006 12:00 AM

Abstract

Natural systems exhibit random, chaotic, and multiply periodic behaviors that are driven by gravity, weather, and man-made disturbances. Modeling them on a large scale is challenging because behaviors vary discontinuously both spatially and in time. Modeling requires calibration and validation data that represent a diversity of causes and effects. Measured variables are either categorical (static) or dynamic (time series). Integrating multiple data types and reducing large numbers of variables to a select set often leads to subjective decision-making that has significant ramifications when applying state-of-the-art multi-step modeling approaches, e.g., land-use models driving finite element flow models. This paper is Part 1 of a two-part treatment that describes an alternative approach that employs a sequence of numerically optimized data mining algorithms. They include 1) signal decomposition to separate static, chaotic and periodic time series components that are attributable to different forcing functions; 2) time series clustering to segment monitored sites by their dynamic behaviors; 3) non-linear, multivariate sensitivity analysis using multi-layer perceptron artificial neural networks (ANN) to determine the relative importance of categorical variables at predicting site-to-site behavioral variability; 4) spatially interpolating dynamic behaviors with ANNs; and 5) assembling an end-user application that integrates data, site attribute classifiers, and prediction models to model an expansive, behaviorally heterogeneous natural system. This paper also describes applications of this approach that predict water levels and stream temperatures.

COinS
 
Jul 1st, 12:00 AM

Numerically Optimized Empirical Modeling of Highly Dynamic, Spatially Expansive, and Behaviorally Heterogeneous Hydrologic Systems – Part 1

Natural systems exhibit random, chaotic, and multiply periodic behaviors that are driven by gravity, weather, and man-made disturbances. Modeling them on a large scale is challenging because behaviors vary discontinuously both spatially and in time. Modeling requires calibration and validation data that represent a diversity of causes and effects. Measured variables are either categorical (static) or dynamic (time series). Integrating multiple data types and reducing large numbers of variables to a select set often leads to subjective decision-making that has significant ramifications when applying state-of-the-art multi-step modeling approaches, e.g., land-use models driving finite element flow models. This paper is Part 1 of a two-part treatment that describes an alternative approach that employs a sequence of numerically optimized data mining algorithms. They include 1) signal decomposition to separate static, chaotic and periodic time series components that are attributable to different forcing functions; 2) time series clustering to segment monitored sites by their dynamic behaviors; 3) non-linear, multivariate sensitivity analysis using multi-layer perceptron artificial neural networks (ANN) to determine the relative importance of categorical variables at predicting site-to-site behavioral variability; 4) spatially interpolating dynamic behaviors with ANNs; and 5) assembling an end-user application that integrates data, site attribute classifiers, and prediction models to model an expansive, behaviorally heterogeneous natural system. This paper also describes applications of this approach that predict water levels and stream temperatures.