Sprungmarken

 

The analysis of hydrological and hydrochemical time series with multi-variate non-linear procedures

BITÖK-A15

From 01/1998 to 12/2000

Principal Investigator: Michael Hauhs
Staff: Gunnar Lischeid, Tobias Rötting
Grant: 0339476 C Grundlagen zur nachhaltigen Entwicklung von Ökosystemen bei veränderter Umwelt

A variety of models exist to describe the chemical hydrograph of the runoff of small catchments. However, inverse modelling as a tool to assess parameters and for process identification usually is far from being unique, mainly due to overparametrization. Thus this project aims at reducing model complexity as far as possible by applying artificial neural networks. They allow for mapping non-linear multivariate relationships without any a priori assumption about the basic processes. Models of this type successfully simulated SO4, NO3 and CI dynamics. Visualization of the regression planes helps to identify the predominating processes. Furthermore, neural networks allow for a more detailed trend analysis, revealing shifts that are limited on certain parts of the regression planes only.

List of publications of this Project

Lischeid, G; Moritz, K; Bittersohl, J; Alewell, C; Matzner, E: Sinks of anthropogenic nitrogen and sulphate in the Lehstenbach catchment (Fichtelgebirge): lessons learned concerning reversibility., Silva Gabreta, 4, 41-50 (2000) -- Details
Hauhs, M; Lischeid, G; Lange, H: Auswertung von forsthydrologischen Monitoringdaten, Stoffhaushalt von Waldökosystemen, 7, 11-18 (1999) -- Details
Lischeid, G: Woher kommt das Wasser? Abflussgenerierung in kleinen bewaldeten Einzugsgebieten in Bayreuther Institut für Terrestrische Ökosystemforschung (BITÖK): Bayreuther Forum Ökologie, Selbstverlag, 9-16 (1999)
Lange, H; Lischeid, G; Hauhs, M: Complexity analysis of time series from two headwater catchments in South Germany in European Academy, Bozen: Hydrology, Water Resources and Ecology of Mountain Areas, Tappeiner, U; Ruffini, FV; Fumai, M, 103-106 (1998)

last modified 2002-04-09