LDMS: A Low-Dimensional Modeling System for Hillslope, Catchment and River-Basin Runoff
Abstract
In this research we have developed a strategy for dynamical modeling of multi-scale hydrologic systems. The approach assumes that soil moisture and saturated groundwater storage serve as essential state variables in the rainfall-runoff process and that natural variations in topography, drainage area, and depth of moisture penetration serve to define the particular flow geometry. The state variables are formed by "weighted averaging", where the weighting is a conditional probability for terrain features (altitude, aspect and depth of moisture penetration) determined from digital terrain data and a GIS. Similarly, "volume weighted" storage-flux relations are determined. The strategy of "terrain averaging" is proposed here to simplify the model dynamics, retain essential nonlinearity while preserving the local space-time scale (Duffy, 1996). In large regions, however, runoff represents an unknown number of space-time scales, and the question becomes: What is the dimension and complexity of this extended dynamical system? To this end we apply the signal processing technique of Multichannel Singular Spectrum Analysis (MSSA), a generalization of Principal Components Analysis for space-time processes. From multiple precipitation and streamflow time series across the drainage basin, we estimate the time scales of the response and the complexity of our dynamic system.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 28, 2000
- Accession Number
- ADA378858
Entities
People
- Christopher J. Duffy
- David Brandes
- Karsten Sedmera
- Tong-ying Shun
Organizations
- Pennsylvania State University