Looking at catchments in a new way: The role of connectivity, scale, and location to understand the export of diverse biogeochemical elements

Abstract:

Catchment morphology is a fundamental component regulating stream hydrology and biogeochemistry in heterogeneous landscapes. Using advanced tools from high-resolution digital data to add to the current state of understanding of catchment systems, we show that three different landscape attributes provide spatially explicit approaches to predict stream chemistry. Elements with low affinity for organic matter were best predicted using the areal coverage of landscape types while elements with high affinity for organic matter were best predicted using an approach that included measurements of connectivity of landscape types. The location of the landscapes relative to the outlet also showed strong influence on some elements, particularly those with high atmospheric deposition. This presentation, which is the second of two, is based on an analysis of Krycklan Catchment Study in northern Sweden.

Background

Stream water quality reflects the hydrological and biogeochemical characteristics of the catchment. Defining linkages between stream chemistry and landscape features is therefore an effective tool to understand catchment functioning. Catchment topography determines the hydrological flow pathways while the landscape type determines the mobility, retention, and export of different biogeochemical constituents to streams. In order to understand how the catchment functions as a source and control of stream chemistry, it is necessary to understand why some elements are retained, while other are released to adjacent streams through different pathways.

Stream water quality consists of a multitude of dissolved and suspended materials. While the focus of previous researches have been on understanding the sources and controls of single parameters or elements with similar characteristics, less focus have been placed on understanding elements in a wider context. Elements have important functions as nutrient sources for biota. However, there are also toxic elements which can become a source of pollution to water quality depending on their concentrations and forms. A critical point in sustaining and managing water quality therefore lies in understanding how many of these classes of elements are regulated in the catchment.

The link between landscape types and stream chemistry provides an explicit way of deciphering various functions of the catchment [Lidman et al., 2014]. Downstream concentrations can be predicted by mixing headwater end members in proportion to their areal coverage. The areal coverage is a relatively simple technique which lumps all processes into a simple, easy to implement model by assuming that the unique properties and processes of each landscape types are captured by their sizes. Other digital terrain analysis, can be used to show how specific landscape features influences stream biogeochemistry [Creed and Sass, 2011]. Since the biochemical reactivity and mobility of different elements is determined by the connectivity of the landscape features to the stream, wetter landscape features or those located closer to streams may have greater influences than others may. Therefore, defining how water is routed in the catchment and the location of different landscape features that can act as hot spots relative to the stream may explain some of the variability in stream chemistry.

In this work, we ask the question if a multi-method approach combining hydrological and morphological features at multiple scales can improve the explanation of spatial variability of diverse chemical elements and if this can help decipher the function of the catchment from a landscape perspective. Based on our previous work, we hypothesize that for some solutes (Aluminum (Al), Rubidium (Rb), Cesium (Cs), Cadmium (Cd) etc.), the parts of the catchment that are closest and more connected to the streams will have a disproportionately high impact on the stream water chemistry. Other solutes (base cations (Ca, Mg, Na, K)) are more associated with mineral soil weathering at more distant catchment locations.

Method

This research was carried out in a boreal catchment (Krycklan catchment) which is located in Northern Sweden. In this presentation, which is a follow up to Laudon et al. (also in this session), we applied three approaches (table 1) to model stream concentrations variability in 14-nested subcatchment during the period of 2006-2013. These nested subcatchments have varying sizes, landscape types, and hydrological conditions. Three landscape types (peat, till and sediments) were created by classifying the quaternary deposits map (Swedish Geological Survey, Uppsala Sweden, 1:100000) in dominant classes. Rocks, arable land and lakes were excluded from the analysis because their coverage was not significant enough to be included in the analysis (<3%). Digital terrain data were created from high-resolution LiDAR data (Light Detection and Ranging) (Lantmateriet, the Swedish Mapping, Cadastral and Land Registration Authority). A Digital Elevation Model (DEM) with a resolution of 2*2m was first corrected to create a flow compatible model (i.e. burning ditches across roads and filling depression) before any of the analysis was carried out.

Table 1 Appraoches used to model catchment characteristics

Landscape features	Assumptions	Calculated in GIS as
Areal coverage	Different quaternary deposits have different characteristics and processes which can be represented by their areal coverage.	Areal coverage of each quaternary deposit within each subcatchment
Connectivity	Water accumulates in locations in the catchment based on the topography. Different soil types are associated with different topographies. Soils which have higher flow accumulations are more connected to streams and should be should be more important for stream biogeochemistry.	Sum of flow accumulation in each soil polygon
Location	Soil types located closer to the catchment outlet will have stronger influence on the stream chemistry than those located further away.	Sum of downslope flowpath distance of each pixel to the outlet

Stream exports of total organic carbon (TOC), base cations (BC), and a wide range of other elements (La, Se, Si, Cl, Al etc) were tested using multiple regression analysis as a function of the dominant landscape types and pH. A landscape-mixing model was created using the three dominant landscape types (peat, till and sediments) as end members to model stream chemistry in the 14 nested subcatchments. The first approach used the landscape-mixing model with simple sub-catchment areal coverage to model stream chemistry. Areal coverage of each landscape types was calculated from the quaternary deposit map in Arc GIS. The second approach used flow accumulation within each landscape type to represent connectivity to streams. Flow accumulation was calculated from the corrected DEM using a multiple flow direction algorithm (MDInf) [Seibert and McGlynn, 2007]. The third approach which represented the effects of location of landscape types on stream chemistry used the downslope flow path length of each landscape type to the catchment outlet. This was calculated along the flow direction which was defined by the Deterministic 8 algorithm (D8) [Ocallaghan and Mark, 1984].

Results and discussion

Large spatial variability in stream water quality can be seen in the Krycklan catchment based on the analysis of 37 elements (figure 1). Some elements are associated with certain catchments characteristics. This association is linked to the dominant landscape types which shows three distinct groups based on areal coverage. Catchments with high portions of peats were more strongly associated with lead (Pb), TOC, iron (Fe) and antimony (Sb) while catchments with high portions of tills have corresponding high concentrations of elements such as (Si, Na, SO4, Cu). Other elements such as K, La, Ca, U, Rb etc. were more associated with catchments with high portions of sediments.

Table 2 The element are categorized according to the approach with the best model according to the coefficient of regression (r²). Each element was modeled using all three approaches (areal, connectivity and location) but only the highest r² is shown.

Areal	r2	Connectivity	r2	Location	r2
Ca	0.98	Al	0.94	Cr	0.90
Fe	0.74	Ba	0.91	Ni	0.94
K	0.93	Mn	0.58	Cu	0.92
Mg	0.94	S	0.97	Se	0.96
Na	0.98	Si	0.97	Rb	0.79
Zn	0.65	Sr	0.97	Cd	0.92
Pb	0.93	Ti	0.60	Cs	0.85
SO4	0.95	Sc	0.96
		Co	0.67
		Ge	0.77
		As	0.53
		Y	0.97
		Sb	0.87
		La	0.95
		Th	0.97
		U	0.98
		TOC	0.85
		Cl	0.72
		F	0.56

The areal coverage method was best for predicting weathering derived elements with a relatively conservative geochemical behavior. Modeling stream chemistry with this method showed particularly high r² values for Ca, Mg, Na, Si etc. (table 2). The areal coverage of peats, tills soil, and sediments with pH can explain most of the variability in these elements whose sources are mainly from mineral soils and have low affinity to organic material. As discussed by Lidman et al. [2014], elements with low affinity for organic matter do not appear to accumulate in wetlands in any higher degree so the stream water concentrations are largely determined by the availability of the mineral soils, weathering of which are their primary source.

TOC, some redox sensitive elements (Mn, Sb, Co etc) and organophillic elements (U, Th) were better predicted using the flow accumulation approach (table 2). The improvement for TOC seems logical given that the riparian soils along with wetlands are known to be the major source for TOC in boreal streams [Laudon et al., 2011]. The wetter and more organic environment in the riparian zone as compared to ordinary forest soils is also likely to cause more reducing redox conditions, which could be expected to affect the mobility of primarily redox sensitive elements and organophillic elements. For instance, the mobility of Mn is favored by more reducing conditions. Some strongly pH dependent elements, e.g. Al, were also better predicted using this approach. In the case of Al previous research has demonstrated that the mobility of Al is limited by precipitation of Al(OH)₃ in streams with higher pH [Kohler et al., 2014]. Large differences in Al concentrations between uphill and riparian soils have previously been observed, suggesting that the riparian zone has a fundamental impact on the mobility on elements such as Al [Cory et al., 2007].

Elements that could be largely influenced by atmospheric deposition (Cr, Ni, Cu etc) were best predicted with the downslope flowpath length to the outlet (Approach 3). The elements including Rb, Cs, Cd and Se were also better predicted with this approach. This reflects that the location of landscape types have a strong effect on the concentrations of these elements in the catchment.

Conclusion

In this research, we used different methods to determine how landscape features control the functions and characteristics of a meso-scale catchment. In general, the study shows that different attributes of the catchment (areal coverage, connectivity, or location) can be used to improve the predictions of the different biogeochemical constituents. The areal coverage was a better predictor for elements, which are derived by weathering of mineral soils and which have a comparatively conservative behavior, e.g. most base cations. This suggests that mobility rarely is an issue for these elements, which is concordant with our general understanding of their biogeochemical properties. Instead, the kinetically controlled weathering is the limiting factor, and, consequently, the proportion of weatherable soils is decisive for the stream water chemistry. Elements with high affinity for organic matter, redox sensitive elements, and TOC were much better predicted with the flow accumulation approach. These elements with complex chemistry are usually associated with the conditions that tend to prevail in riparian soils. The location of the landscape was also found to be important for influencing some elements.

Reference

Cory, N., H. Laudon, S. Kohler, J. Seibert, and K. Bishop (2007), Evolution of soil solution aluminum during transport along a forested boreal hillslope, Journal of Geophysical Research-Biogeosciences, 112(G3), doi: 10.1029/2006jg000387.

Creed, F. I., and Z. G. Sass (2011), Digital terrain analysis: Approaches for tracking hydrological and biogeochemical pathways and processes in forested landscapes, in Forest hydrology and biogeochemistry:Synthesis of past research and future directions, edited by F. D. Levia, Darryl M-C and TanakaT., Springer Netherlands.

Kohler, S. J., F. Lidman, and H. Laudon (2014), Landscape types and pH control organic matter mediated mobilization of Al, Fe, U and La in boreal catchments, Geochimica Et Cosmochimica Acta, 135(190-202), doi: 10.1016/j.gca.2014.03.033.

Laudon, H., M. Berggren, A., Agren, I. Buffam, K. Bishop, T. Grabs, M. Jansson, and S. Kohler (2011), Patterns and Dynamics of Dissolved Organic Carbon (DOC) in Boreal Streams: The Role of Processes, Connectivity, and Scaling, Ecosystems, 14(6), 880-893, doi: 10.1007/s10021-011-9452-8.

Lidman, F., S. J. Kohler, C. M. Morth, and H. Laudon (2014), Metal Transport in the Boreal Landscape-The Role of Wetlands and the Affinity for Organic Matter, Environmental Science & Technology, 48(7), 3783-3790, doi: 10.1021/Es4045506.

Ocallaghan, J. F., and D. M. Mark (1984), The Extraction of Drainage Networks from Digital Elevation Data, Computer Vision Graphics and Image Processing, 28(3), 323-344, doi: 10.1016/S0734-189x(84)80011-0.

Seibert, J., and B. L. McGlynn (2007), A new triangular multiple flow direction algorithm for computing upslope areas from gridded digital elevation models, Water Resources Research, 43(4), doi: 10.1029/2006wr005128.