Dynamika transportu fluwialnego górnej Parsêty jako odbicie funkcjonowania systemu zlewni

Dynamics of fluvial transport of the upper Parsêta River as a response of the catchment system

Andrzej Kostrzewski, Malgorzata Mazurek, Zbigniew Zwoliñski

9. Multi-parametric model of fluvial transport

The data presented in this book also help identify factors other than water discharge causing variations in fluvial transport in lakeland areas of the temperate zone. With this aim in mind, the available data were used to model the fluvial transport pattern of the upper ParsÍta. Only those parameters were included which vary over time, namely precipitation, air and ground temperature, duration of the vegetation season, moisture of the soil cover, water discharge and specific conductivity, concentrations of ionic macrocomponents and ionised silica, concentrations of dissolved and suspended material as well as bedload, and total flow of river material. On the basis of selection following indices treated as independent variables: precipitation indices Px, Pix, PPx, where x = 1, 2 ... 14, 15, 20, 25 and 30 days and APIx where x = 9, T and ET, seasonal indices SI and COSSIN; and water discharge Q. Fluvial transport parameters were adopted as dependent variables, i.e. concentrations of ionic macrocomponents Ca2+, Mg2+, Na+, K+, HCO3-, Cl- and SO42- (N = 52), specific conductivity SEC, concentrations of ionised silica SiO2, concentrations of dissolved material Cd, suspended material Cs and bedload Cb, and total flow of river material AL (N = 731).

Table 9.3.1. gives a matrix of correlations between ionic component concentrations, fluvial parameters of the upper ParsÍta, and the independent variables. The analysis of part A of the matrix leads to the conclusion that the discharge remains the most important factor responsible for changes in the concentration of the listed components (except Mg2+). The small number of significant coefficients of correlation for precipitation indices may be taken to suggest a substantial transformation of precipitation water in the catchment system. Part B of the correlation matrix shows that the dependences are inversely proportional for dissolved substances, and directly proportional for solid particles. After the preliminary analysis of the independent variables and selection of those that turned out to be significant, possible predictive models of the dependent variables under study were determined (Table 9.3.2) using the procedure of multiple regression with a stepwise elimination of variables. Most of the models thus derived are weakly predictive. In the case of sodium and potassium ions, the models are more sensitive than those based solely on variations in water discharge. There is a notable lack of hydrometeorological indices in predictive models for calcium and sulphate ions (Table 9.3.2a) and a total lack of significant dependences for magnesium ions. Better fits of predictive models were obtained for AL and SiO2, and poorer for Cd, Cs and Cb (Figs. 9.3.1a, 9.3.2a, 9.3.3a, 9.3.4a, 9.3.5a, Table 9.3.2b).

Precipitation is the main factor responsible for the dynamics of supply, transformation and carrying away of material in a river catchment system. The reason why a group of precipitation indices forms a pattern of time intervals of up to one week, and usually up to four days, seems to lie in the catchment's retaining capacity which defines the duration of rainwater circulation in the form of overland flow and throughflow. The lengthening of the 4-day impact of retention in the upper ParsÍta catchment is the result of overlap of such factors as the intensity of rain, dryness or moisture of the soil covers, hydrogeological conditions of the substratum, and the distance of the particular subcatchments from the Storkowo measurement profile. The influence of 20-day precipitation indices on the transport of chemical substances is especially well marked in the predictive model for potassium ions. The appearance in the predictive models of precipitation indices of varying time intervals makes it possible to draw conclusions about the productivity of flood discharges.

A verification procedure (e.g. residual variable analysis - Figs. 9.3.1b, 9.3.ba, 9.3.3b, 9.3.4b, 9.3.5b, 9.3.6, 9.3.7, Table 9.3.3) has shown that for five dependent variables (AL, SiO2, HCO3-, K+ and SO42-) the predictive models constructed can be used in forecasting (Table 9.5.1).

Previous Contents Next