Spatiotemporal variability and landscape controls of specific discharge in a Boreal landscape

Thursday, 25 September 2014
Reinert Huseby Karlsen1, Kevin H Bishop1,2, Thomas Grabs1 and Jan Seibert3, (1)Uppsala University, Uppsala, Sweden, (2)Swedish University of Agricultural Science, Uppsala, Sweden, (3)University of Zurich, Zurich, Switzerland
Abstract:
Spatial and temporal variability of specific discharge in the boreal landscape is rarely documented, and even less well understood. Pioneering work on 14 sub-catchments in a nested boreal meso-scale (67 km2) catchment in Northern Sweden reveal spatial and temporal patterns in both the magnitude and variability of specific discharge. Here we explore the structure of this variability and its connection to the landscape using a 5-year time series of continuous flow records. The variations in spatiotemporal variation in specific discharge have important implications for biogeochemical exports, since they are often calculated based on the assumption of a spatially constant specific discharge.

The aim of the present study is to quantify both the spatial and temporal variability in specific discharge, as well as how different landscape characteristics such as soil, vegetation and topography can be related to the observed variability. Continuous time series records allowed studying the spatial variability over a range of antecedent weather conditions. An important task was the filling of gaps during winter periods using a modelling approach. This was done to allow comparisons of, for instance, annual mean discharges. The catchments vary in size from 12 ha headwater catchments to the 4th order main outlet of 6 700 ha. Lower elevations of the catchment are covered with fluvial sediment deposits, while glacial till deposits are found at the higher elevations. Topography is gentle and long term gross precipitation is assumed to be equally distributed over the catchment.

Results show a long term (5 years) variation between 73 % and 132 % from the main outlet, where the higher flows are seen in wetland dominated catchments (Spearman rank correlation p=0.81 between flow and wetland coverage). However, rank correlations with other spatial characteristics over the long term show weaker relationships. When looking at seasonal and short term flows, both more pronounced variability and stronger links to different landscape properties are seen. Spring flood magnitudes show strong correlations (p=0.61 - 0.80) with open and wet areas, while summer flow is negatively correlated to catchment tree volumes (p=-0.61 - -0.75) and potential evaporation (p=-0.48 - -0.78), see Figure (a).

On shorter timescales, from daily to monthly, other catchment properties also present themselves in the spatial discharge variability. During dry periods of low summer baseflows, catchments with deep fluvial deposits maintain a higher discharge than the catchments with shallow soils. This pattern is reversed during summer stormflow (Figure (b)). In general, catchments with more fluvial deposits are also larger in size, and it is thought that this temporally dependent spatial pattern is a combined result of both scale and storage characteristics.

The results show that there is a spatial structure in the specific discharge that is temporally variable. Various landscape traits influence the flows at different time scales, and the spatiotemporal discharge variability depends on seasonal climatic variability. The observed structure does not only influence mass balance calculations, but also gives pointers to how runoff is generated in boreal landscapes at different spatial and temporal scales.

Figure:

a) Scatter plots showing total spring flood magnitude against fraction of wet area (wetlands and lakes) covering the catchments (top), and total summer flow against catchment treevolume (bottom).

b) Daily specific discharge from reference site 7 (top) and spearman rank correlation between flow and deep fluvial cover of all 14 sites (bottom). Data is aggregated over periods from 1 day to 1 year. Specific discharge and fluvial sediment soils are positively correlated during drier periods and negatively correlated during runoff events. Winter periods December - March, where most data is infilled using model simulation, are shaded.