Abstract
The recession curve provides important information about the low flow behaviour of rivers, and tells in a general way about the natural water storages feeding the stream. This work has, however, found a high time variability in recessions. In order to derive recession characteristics which are representative for the storage properties of a catchment, it is important to have quantitative knowledge about the factors influencing the recession variability. Using a new, automatic method for calculating recession characteristics developed in this study, the following issues were addressed:
1)The regional variability in the recession rate. The adoption of an average recession constant for the catchment is shown to be useful in regional analysis. The most important charac-teristics found to affect the recession rate, are related to geological characteristics, relief and climate.
2)The time variability in the recession rate. Variations due to limitations in the recession model and climatic influences during the recession period are studied. The results demonstrate the need for incorporating these effects in recession analysis.
3)The low flow model performance of the conceptual HBV rainfall-runoff model. The climatic factors precipitation and temperature are to a much lesser extent than in the observed series, responsible for the time variability in the simulated series.
4)The influence of evapotranspiration on recession variability. This relationship is indicated by the correlation found between recession rate and temperature. In order to quantify the water lost by evapotranspiration, an appropriate model for application on a local scale is required. This study introduces AMOR, a modified MORECS (Meteorological Office (UK) Rainfall and Evaporation Calculation System) soil water budget model. It is concluded that this approach provides a promising method for estimating evapotranspiration based on routinely observed climatic data. An objective of future investigations is to prove the general applicability of the AMOR model.