Framework

Mapping connectivity

Development of simple methods that allow a rapid assessment of the potential hydraulic connection between groundwater and surface water systems are of great value to natural resource managers, State and catchment water agencies and in the development of water policy. Such mapping of connectivity potential can be useful in water accounting as well as for conjunctive water management purposes. A simple GIS-based method has been developed to map the potential hydraulic connection between groundwater and surface water systems in a catchment, taking into account hydrological and hydrogeological factors. This method can handle spatially distributed catchment-scale data and can be modified depending on data availability. The methodology provides sufficient information for a first-cut prioritisation of stream reaches, enabling targeting of further investigations and management. It is important to remember that this GIS methodology focuses on the conductance of the geological material to derive an indicator for the potential for water movement. This is a precursor to methods such as hydrographic analysis or numerical modelling to derive an understanding of the direction and magnitude of seepage flux.

A connectivity index model

Stream-aquifer connectivity potential in a catchment can be determined by means of a rating index approach. To fully describe the catchment processes, many input data and their spatial and temporal variability has to be taken into account. Hydrological and hydrogeological factors mapped at the catchment scale can be combined to obtain a final rating. The general data inputs needed for this method is:

1. depth to water table;
2. stream bed characteristics;
3. geology; and
4. geomorphology

One of the prime considerations is the conductance of the geological material. As well as the aquifer itself, this includes the material of the stream bed and banks, the interface between stream and aquifer. For example, it is assumed that connectivity will be high where stream bed sediments consist of gravel and the aquifer consists of alluvial sands. Alternatively, the connectivity will be low where there is an intervening layer of clay of significant thickness.

A numerical rating and ranking system was devised for the four parameters mentioned above. The system includes ranges, ratings and weights. The dataset representing each parameter is subdivided into meaningful ranges, with a rating assigned to each data range. The rating represents the relative influence on connectivity. The weights determine the relative importance of each parameter. Each parameter is assigned a relative weight ranging from 1 to 5; a weight of 1 is the least significant and a weight of 5 is the most significant. Sensitivity analysis can be undertaken to explore the effects of different weightings placed on the parameters. The following is an example of the overall additive model where a numerical connectivity index is obtained.

Connectivity Index	= Potential for groundwater-surface water connectivity
	= (3 x depth to water table) + (5 x stream bed sediments) + (5 x aquifer material) + (2 x geomorphology)

Figure 1 outlines a flow chart for this process. The calculated index identifies the stream reaches that will have potential for groundwater-surface water connectivity. The higher the connectivity index, the greater the potential for stream-aquifer connectivity. The rigour in the index model has been achieved without making it a data-hungry model and sacrificing its practicality. The potential connectivity ratings for different river reaches provided by this model can be compared with actual field measurements.

Figure 1: Flow chart showing structural component of the connectivity index model

In an initial implementation, the methodology used a simple spreadsheet software format (Figure 2). The tool allows the user to select appropriate ranges and ratings for each parameter using a drop down menu. The spreadsheet then converts the various input parameters into numerical values using a series of mathematical and logical steps using standard spreadsheet functions. Variables can be changed by the user, and the connectivity indices automatically calculated. The spreadsheet presents indices of groundwater-surface water connectivity as high, moderate and low potential. The spreadsheet can be used as an exploratory tool to determine the relative sensitivity of each parameter, so that data collection and preparation can be focussed on the most critical data sets.

Figure 2 Spreadsheet implementation of the potential connectivity index model

Integration of connectivity index model in a GIS environment

The index method was implemented in a GIS environment using ESRI ArcGIS 9.0 to enable the creation of maps of connectivity potential at the catchment scale. The GIS-based approach was trialled using datasets for the Border Rivers catchment. Potential connectivity was mapped spatially by combining four raster datasets into one representing water table depth, stream sediments, aquifer material and geomorphology. Map algebra was used to derive and assign a single numerical index along the river reach of the catchment. The map algebra equation is formulated from the connectivity index model equation based on weighting individual data parameters and combining the results into a single index value. The higher the single index output, the greater the potential for groundwater-surface water connectivity. The final single index value output from the map algebra equation for each grid cell in the raster is further categorised into low, medium and high connectivity potential classes based on the output classification classes in the connectivity model. These categories can then be mapped using a standardised legend to spatially represent the estimated potential connectivity along the river reaches (Figure 3).

The following four catchment datasets for the Border Rivers catchment were rasterised and combined to derive a connectivity index (Figure 3):

1. Depth to water table: The depth to groundwater measurements from existing State water agency borehole monitoring were interpolated into a gridded surface (250m cell size) of the depth to the water table. The resulting watertable depth raster was then reclassified into three broad categories according to the weighted value in the connectivity index model. Shallow watertables (<10 m) were assumed to reflect higher connectivity with streams when compared with deeper watertables (>20m).
2. Stream/river bed characteristics: The NLWRA soil saturated hydraulic conductivity (permeability) of Layer 2 gridded (1.1km cell size) national dataset (1999) was used as a surrogate for stream bed characteristics. The dataset was categorised into five permeability classes of very low; low; moderate; high and very high. The permeability classes were then reclassified into a weighted value based on values from the connectivity index model. Finally the raster dataset was resampled to 250m cell size for the catchment using the bilinear resampling technique. River beds composed of sand and gravel deposits were assumed to have high connection, any silt and clays a low connection.
3. Geology: The aquifer material data was sourced from the 1:250,000 scale Geology of the Murray-Darling Basin digital dataset. Lithology units from the dataset were used to identify the type of aquifer material. Where lithology units do not differentiate between sediment types lithological logs sourced from borehole databases maintained by State agencies can be used as an alternate dataset. Borehole data can provide greater detail in the spatial variability of aquifer lithology. However, using drillers logs to classify the aquifer material distribution in the profile is more time-consuming. It is assumed that aquifers with gravel and sand materials tend to have higher potential for connectivity with streams and are thus rated more highly. Geological units dominated by silt, clay or fractured rocks are assumed to have lower potential for connection and are thus assigned low index values.
4. Geomorphology: The Multi-resolution Valley Bottom Flatness Index (MrVBF; Gallant and Dowling, 2003) was used to interpret landscape geomorphology. The MrVBF index allows for the delineation of erosional and depositional environments based on an algorithm applied to a Digital Elevation Model at multiple scales. Narrow alluvial valleys with high rainfall and shallow groundwater are assumed to have high connection with streams and are thus assigned higher values compared with lower values assigned to wide, arid alluvial plains with deep groundwater levels.

Figure 3: GIS-based approach for mapping stream-aquifer connectivity applied in the Border Rivers catchment

References

Gallant JC, Dowling TI, 2003. A multi-resolution index of valley bottom flatness for mapping depositional areas. Water Resources Research 39(1):1347.