Kim Hodge and David J. Sauchyn

Kim Hodge, GIS Specialist, Agriculture and Agri-Food Canada, Prairie Farm Rehabilitation Administration, Calgary, Alberta, Canada. (403) 292-4562 ph, (403) 292-5659 fax, pf10390@em.agr.ca email

David J. Sauchyn, Professor, Department of Geography, University of Regina, Regina, Saskatchewan, Canada. (306) 585-4030 ph, (306) 585-4815 fax, sauchyn@leroy.cc.uregina.ca email

Prairie potholes (sloughs/wetlands/kettle-holes/marshes etc.) are important economic and environmental components of the prairie landscape. They provide habitat for wildlife, reduce potential flooding and maintain local and regional water table levels. Despite this, they have not been subject to quantitative regional analysis. Traditionally, potholes have been examined at a local rather than regional scale. As environmental and agricultural issues come to the forefront of public concerns, decisions about potholes will need to be made at both the local and regional scale. This paper examines whether information, such as pothole-ground water relationships, topography and surficial geology derived at a local scale can be applied to a region of potholes, to facilitate decision making, with the use of a GIS. Surficial geology and topography showed a strong relationship with the distribution of potholes. Till deposits, and specifically hummocky moraine and till plains, have the largest number of potholes. It was expected that irregular slopes and Gleysolic soils would display a strong relationship with potholes. However, slope and soil tests, with the exception of soil water storage capacity, did not show expected results. However these results were impacted by data quality, scale and availability.

The Great Plains of North America and the prairie pothole zone (Figure 1.1) have diverse topography, geology, and hydrology. They are bounded by parkland to the east and north and by the Rocky Mountain Foothills to the west, and are typically characterized as dry, flat, homogeneous environments. However, at local (a few square kilometres) and regional (10's to 100's of square kilometres) scales and over time the vegetation, soil, topography and hydrology of these environments are diverse and dynamic.

Potholes are important to the economic, hydrologic and ecologic systems of the prairie environment. They act as chemical, nutrient and hydrologic retention areas, and provide habitat for flora and fauna. In addition, potholes maintain local and regional water table levels and reduce the frequency and magnitude of floods. Prairie potholes have been subjected to intensive examination on an individual scale, such as detailed measurements of groundwater flow around an individual pothole, however, there has not been a similar level of examination at a regional scale (Novitzki, 1979; 1989; Hubbard and Linder, 1986; Woo et al., 1993).

Pothole hydrology appears to be straightforward. They are viewed as temporary runoff storage areas and a hindrance to farming. However, the dynamics of pothole hydrology includes their influence on soil properties, local weather, stream flow, flooding and groundwater (Carter, 1996). Due to the economic, hydrologic and ecologic significance of potholes, an understanding of their relationship with other geomorphic, hydrologic and topographic systems at a regional scale is important for an overall understanding of the prairie environment.

Many authors (Meyboom, 1966; 1967; Shjeflo, 1968; Woo et al., 1993) comment on the distribution and characteristics of potholes, however, these statements are not backed with quantitative data, but qualitatively and intuitively, it is recognized that potholes occur in characteristic environments. This lack of data leaves a large hole in the basic knowledge of prairie landscapes in general and potholes in particular (Sloan, 1972; Hubbard, 1988; Zebarth, et al., 1988).

The purpose of this paper is to determine how information about potholes, derived at a local scale, such as hydrologic processes, soil characteristics, surficial geology and topography, can provide a better understanding of pothole distribution at a regional scale. The Geographic Information System (GIS), ArcInfo will be used to determine whether current assumptions about pothole distribution and characteristics hold true at a regional scale.

For the purpose of this paper a pothole is defined as a body of water, either permanent or temporary, of all depths and areal extent, and with or without vegetation (Meyboom, 1966; Eisenlohr, 1969; Woo et al., 1993; Tiner, 1996). Potholes can occur on all types of surficial material ranging from till to flat lacustrine material to sand dunes (Carter, 1996; Tiner, 1996).

The area chosen for study is the 1:250,000 NTS mapsheet 62F, located in the core of the prairie pothole region (Figure 1.1 and Figure 2.1). The study area covers 16,000 km² and extends from 100^o W to 102^o W, is bounded by the 49^th parallel to the south and by the 50^th parallel to the north. This study area is representative of the prairie pothole region with a diverse assemblage of landscapes and landforms including broad topographic highs, sand dunes, vast glaciolacustrine deposits, and large areas of potholes.

Land use within and surrounding the study area includes numerous regional, provincial and international parks (Figure 2.1) and recreation areas, intensive livestock operations, ranching, specialty farming and annual cropping. Virden, in the north central part of the study area, is the largest urban centre with a population of 2,956 (Statistics Canada, 1996). The city of Brandon is located approximately 3 km east of the study area and has a population of 39,175 (Statistics Canada, 1996).

Glacial activity formed much of the topographic and hydrologic features in the study area (Figure 2.1). Landforms range from hummocky dead ice moraine superimposed on a Tertiary upland, to flat glacial lacustrine plains. During deglaciation numerous glacial and postglacial lakes were formed including glacial Lake Agassiz which covered nearly all of Manitoba. The drainage of glacial lakes left broad lacustrine plains and formed new, as well as reactivated old glacial spillways. Examples of such spillways include the Souris Valley, Assiniboine Valley and Qu'Appelle Valley (Figure 2.1). Lacustrine deposits and low topography occur over much of the study area extending north and west from Turtle Mountain. A number of moraines exist throughout the study area and include an end moraine that extends south-east from Brandon.

Turtle Mountain, in the south-east corner of the study area, is a significant geographical and hydrological feature. It rises approximately 275 m above the surrounding plains, has a dense network of ponded surface water, is a local recharge area and has a poorly integrated drainage network (Vance and Last, 1994). Most of the streams that do originate from Turtle Mountain flow northward and disappear into the ground. A few streams flow into the large lake basin, Whitewater Lake, 20 km away (Bamburak, 1978).

Two major rivers, the Souris and Assiniboine, as well as numerous smaller, often intermittent streams, meander throughout the study area (Figure 2.1). The Souris River originates in Saskatchewan, flows through North Dakota and then into Manitoba. It connects with the Assiniboine River near Brandon, which originates near Duck Mountain along the Saskatchewan-Manitoba border, and flows east.

Two major lake basins occur within the study area. Whitewater Lake is a large semi-permanent, saline lake, north of Turtle Mountain and west of Boissevain (Figure 2.1). It has an area of approximately 106 km². Whitewater Lake is located in a shallow depression that is believed to have formed as Devonian evaporite deposits dissolved, and the surface collapsed (Christiansen, 1967). Whitewater Lake is fed primarily by groundwater. A few streams flow into the lake, however, no streams flow from the lake. Oak Lake, located south-east of Virden, has an area of 79 km² and is a series of interconnected lake basins surrounded by intermittent waterbodies (Figure 2.1).

Primary data sources for this study are the Canadian National Topographic Survey (NTS) mapsheets (1:50,000), which include hydrologic and topographic data, the Soil Landscapes of Canada (SLC) (1:1,000,000) version 2.1, and surficial geology (1:50,000 digital format, 1:100,000 paper format) from the Geological Survey of Canada (Sun and Fulton, 1995). Each dataset chosen was the smallest scale available for the study area. The majority of the spatial analysis was accomplished with the GIS ArcInfo.

A pothole map was created from 16 1:50,000 NTS mapsheets which were appended to form a seamless data layer. From the water bodies indicated on the NTS sheets only 'wetlands' (NTS code 01492), 'other water bodies' (NTS code 01450) and 'intermittent waterbodies' (NTS code 01452) were used. These best fit the definition of potholes used for this study. The contour map used for topographical analysis and delineation of slopes was derived from the same 1:50,000 NTS mapsheet. Numerous edits were necessary before the topographic and hydrologic data would fit together seamlessly. At the time of completion, detailed digital soil data at a scale that approximated the NTS and surficial geology data was not available for the study area. In addition, only four NTS sheets of surficial geology (62F/01,02,07 and 08) were available. Detailed digital groundwater data does not exist for Manitoba or Saskatchewan. As a substitute, SLC data was used for available water capacity in upper 120 cm, depth to water table, soil type (Great Group Classification) and texture. These data are at a 1:1,000,000 scale and are very coarse for such an exercise, however, it is the only digital data available at this time.

The following additional data sets were chosen from the SLC data based on quality and importance: local surface expression, depth to water table, soil texture, soil type, ability of soil to drain water and available water capacity.

NTS contour data (8 m interval) was used to determine the topographic (slope) characteristics where potholes are most prevalent. Slopes were delineated by creating a Triangulated Irregular Network (TIN). From this dataset, the GIS generated polygons of equal slope value. Seven classes were created, using natural breaks in the data:

1. less than (lt) 1%
2. greater than or equal to (ge) 1% and (lt) 2%
3. ge 2% and lt 3%
4. ge 3% and lt 5%
5. ge 5% and lt 10%
6. ge 10% and lt 20%
7. ge 20%

The natural breaks were checked manually to ensure appropriate breaks were chosen. For each slope class, the number of potholes that fell in each slope range was determined by overlaying the pothole layer on the slope class layer in the GIS.

Surface expression was derived from SLC data. The pothole layer was overlain and the number of potholes in each class was determined. Surface expression classes include water (not applicable), dissected, hummocky, knoll and kettle, level and undulating.

This test sought to determine whether there is a correlation between the location of potholes and known or interpreted water table elevation, available water capacity and soil drainage. SLC data utilizes an 8 class system for available water capacity in the upper 1.2 m of soil; 50 mm, 100 mm, 150 mm, 200 mm, 250 mm, water, saline and high water table. Soil drainage data in this study area includes 4 classes, water or not applicable, imperfect, poor and well drained. The depth to water table values ranged from 0 to 2 m, 2 to 3 m and greater than 3 m. The pothole data was overlain on each of the variables to determine how many potholes occurred within each type of soil water polygon.

This exercise was designed to determine the relationship between potholes and soil, specifically, texture and type. Soil types include water (not applicable), Black Chernozemic, Dark Gray Chernozemic/Luvisolic, Back Solonetzic, Regosolic and Gleysolic. Soil textures include water (not applicable), clay, clay loam, fine sandy loam, fine sand, loam, loamy sand and sandy loam. For this study, the potholes were overlain the soil textures and soil types from SLC data.

Unpublished surficial geology was available in digital format for only 4 of the 16 NTS mapsheets of the study area from the Geological Survey of Canada (GSC). Changes may be incorporated prior to their publication and after this study. The surficial geology and pothole layer were overlain and the number of potholes per surficial geology unit was determined. A description of surficial units can be found in Appendix A

The results of this analysis determined that potholes dominantly occurred on low slopes (less than 1%) and had the highest density with 3.68 potholes/km², although they were not absent in steeper slope classes. Much of the topographic detail, such as individual depressions, which may contain water, is lost with the 8 m contour interval. Much more detailed data, such as a 1 metre contour interval, would be needed for identification of every depression where potholes could occur. Complete results can be seen in Table 4.1 and Table 4.2.

The steepest slopes have the fewest potholes resulting from a number of factors. Most of the highest slope values are in the valleys of the Souris River and the Assiniboine River. The steep glacial valleys of the Souris and Assiniboine Rivers preclude any significant pothole development. High slope values are also found in Turtle Mountain. It has numerous potholes but they tend to be larger and more permanent, indicative of stagnant ice melt and subsequent letdown of glacial debris. The distribution of potholes relative to slopes can be seen in Figure 4.1.

The low to moderate slope classes, less than 3%, had a total of 99% of all potholes (Table 4.1). This seems to contradict what is known about pothole distribution, however, hummocky and rolling topography, where potholes are reported to be dominant, do not necessarily have steep slopes. Ridged till plains are the most common pothole substrate with less than 5 m of relief (Sun and Fulton, 1995). Similarly, flat till plain, the second most common pothole substrate, has a relief of less than 2 m (Sun and Fulton, 1995).

The SLC database classified much of the landscape in the study area as hummocky, while Sun and Fulton (1995), classed the area as undulating and level. The difference can likely be accounted for with the 1:1,000,000 scale of the SLC data and the generalizations necessary to make such a map, as well as to differences in terminology and classification.

Local surface expression, as derived from the SLC data, can be seen in Figure 4.2. The knoll and kettle landscape, which has no external drainage and slopes generally greater than 3% has potholes occupying about 15-20% of the area. Similarly, this study found it has the highest percentage of total potholes, 62% (33,656 potholes) as well as the highest density with 5.71 potholes/km². Level topography (4,196 km²) accounts for the next most numerous class with 7,236 potholes, 13% of all potholes with a density of 1.72 potholes/km². Undulating topography (2,500 km²) has 6,744 potholes resulting in a density of 2.70 potholes/km². The remaining results can be found in Table 4.2.

Topography, climate, hydrology and surficial geology contribute to the frequency and above average size of potholes on hummocky moraine. Hummocky moraine (1,553 km²) has 3,981 potholes and a density of 2.60 potholes/km². The local topography is developed in till, a morainal deposit that formed as debris-rich ice melted leaving debris and random large closed depressions. The depressions, combined with complex and frequent interactions with various scale groundwater systems, account for their unique characteristics and high pothole density relative to other surficial material (Sun and Fulton, 1995).

Moisture in the upper parts of the soil profile is important for pothole development and maintenance. From the SLC data 3 moisture variables were measured; available water capacity, depth to the water table and ability of these soils to drain. These 3 variables were analyzed against pothole distribution.

The tests which examined the relationship of pothole location to depth of water table, showed that the greater the depth to the water table the larger the number of potholes (Table 4.2). Density reached a maximum of 4.27 potholes/km² in the greater than 3 m class (Appendix B).

The high frequency of potholes in areas with deep water tables corresponds to areas with well drained sub-surficial material and high available water capacity. The scale of the water table examined is again too coarse to thoroughly examine and determine the relationship of potholes to groundwater at a regional scale.

Most potholes occur where the groundwater is farthest from the surface because this groundwater is part of a regional groundwater system. Potholes interact with this groundwater both directly through seasonal mounds as described by Meyboom (1966) and indirectly as the regional groundwater system influences local scale groundwater systems.

The water class (lakes) had few potholes, and corresponds with flat lacustrine deposits and Gleysolic soils found around Whitewater and Oak Lakes. Potholes occurred on the water class because of different delineation of boundaries for Whitewater and Oak Lakes. Potholes are densely concentrated around Oak Lake on the imperfect and poorly drained classes. Numerous marshy potholes occur around Oak Lake. Typically however, these flat lacustrine plains with extremely low slopes and poor drainage make unfavourable conditions for pothole development and maintenance. Without steeper slopes natural depressions do not form which decreases the chance of pothole development. Additionally, the areas with poor drainage can only be identified at a very coarse scale. Finer scale data are needed to adequately examine potholes and groundwater in more detail over a larger region.

Soil drainage data consisted of 4 classes, water, imperfect, poor, and well drained (Table 4.2; Appendix B). Well drained soils had the most potholes, with a density of 4.03 potholes/km², because of several factors including irregular topography and favourable surficial geology. Well drained soil coincides with topography ranging from level to undulating and a 100 to 150 mm available water content, providing an optimum environment for pothole development and maintenance.

The poorly drained soils were located adjacent to the major permanent/semi-permanent waterbodies, Whitewater and Oak Lakes. The combination of level topography with highly impermeable clay-rich soil, which inhibits groundwater flow, lowers the possibility of numerous potholes developing.

The saline soil water class has 153 potholes (1.21 potholes/km²) and the high water table zone has 1,433 (1.15 potholes/km²), which includes areas in and around Whitewater and Oak Lakes (Table 4.2). These occurrences account for less than 3% of the total number of potholes. The high water table areas, where pothole interaction with the groundwater should yield more potholes, have an insignificant number of potholes. Two primary reasons account for this. The first is that the scale of the groundwater data is too coarse to examine the characteristics of the water table around a pothole. The second is that the location of this high water table, relative to topographic conditions and drainage characteristics, is significantly different from what is required for pothole development. These high water table areas occur adjacent to Whitewater and Oak Lakes. Given the same high water table levels with an appropriate landscape of poorly integrated drainage and irregular, hummocky style topography, potholes could likely develop.

Optimum pothole development increases with the available water capacity to densities as high as 4.49 potholes/km² in the 250 mm water class. There is a strong relationship between the available water capacity of the soil and the number of potholes. The increase in the available water capacity results in a near exponential increase in both the number and density of potholes.

The 250 mm area corresponds with the immediate area around Whitewater and Oak Lakes which have Gleysolic soils (Figure 4.3), low relief and high water table values. These characteristics, like available water capacity, suggested to be strongly associated with pothole development by Duchaufour (1982) and Scott (1995), were somewhat unfavourable to pothole development in this study. However, the 250 mm class around Oak Lake had numerous potholes. The hydrology of these potholes appear to be connected with Oak Lake.

The effort to determine how many potholes occur on each soil Great Group was negated by lack of available digital data. The SLC data was used, however, it only provides general soil type (Great Group) and texture at a 1:1,000,000 scale. Precise observations about detailed soil characteristics and pothole occurrences proved difficult given these scale limitations. Figure 4.3 displays the soil groupings available for the study area and Table 4.2 provides the results.

According to other researchers, Gleysolic soils are formed from regular interaction with water in potholes and other variables such as groundwater and surface water (Duchaufour, 1982; Scott, 1995). Gleysolic soils did not display the expected strong correlation with pothole location in the study. The Gleysolic soils have a limited number of potholes, as they coincide with the Oak and Whitewater Lakes. The Gleysolic soil appears to have been influenced not by potholes but rather by the regular interaction of the large lakes. The lakes, coupled with the poor soil drainage, level relief and high ground water table, provide the necessary conditions for Gleysolic soil development.

In addition, many of the potholes in this particular study area are ephemeral which reduces the time available for water in a pothole to alter the soil. Finally, the scale of source data was far too coarse to adequately display small areas of Gleysolic development that likely exist in and around potholes. Examination of 1:100,000 scale soil maps from outside of the study area, but within the Prairie Pothole Region, indicates that a much stronger relationship exists between pothole location and Gleysolic soils than indicated here.

Potholes attained the highest densities (3.62 potholes/km²) on Black Chernozemic and Dark Gray Chernozemic/Luvisolic (3.79 potholes/km²). The Dark Gray Chernozemic/Luvisolic class covers only 535 km² and occurs on Turtle Mountain (Figure 4.3). These soils types occur no where else in the study area and include gleyed characteristics indicating the possible influence of potholes on soil type.

Potholes occurred most often and with the highest density on loam texture soil. This material covered 10,050 km² and had 44,566 potholes and a density of 4.43 potholes/km². The second most numerous soil texture was clay loam with 2,887 potholes (2.89 potholes/km²). Clay loam occurs on Turtle Mountain accounting for this high density.

Many authors (Meyboom, 1962; 1966; Sloan, 1972; Mills and Zwarich, 1986; Hubbard, 1988; Zebarth, et al., 1988) have stated that potholes generally occur on certain types of glacial material such as impermeable fine grained till. However, the authors do not quantify the relationship between surficial geology and pothole location. By overlaying the pothole layer and the surficial geology layer the number of potholes that occur on specific surficial materials was determined (Figure 4.4; Table 4.3).

The following discussion centres on the 1:50,000 surficial geology data obtained from the GSC which covers 4,000 km², 4 of the 16 NTS sheets of the study area. Till is the most common surficial material on which potholes are found, with 86% of the total potholes (see morainal deposits in Appendix A for a description of till). It is composed of a variety of materials, ranging from clay to cobbles and boulders, and is a glacier deposit (Dreimanis, 1977). Till is a highly variable material, defined not by its composition, but rather, by its origin. This is reflected in most classification and mapping systems that use both a morphogenetic and textural classification. Given the great range of till characteristics and recognized importance of till in pothole location and development a discussion of these units is warranted. The following discussion is based on the surficial geology map from the study area (Sun and Fulton, 1995) and the results of this study.

Ridged till plain (Tr), as identified by Sun and Fulton (1995), is the most common surficial unit on which potholes are found with 2,510, 24% of total potholes covering an area of 351 km² with a density of 7.16 potholes/km² (Figure 4.4; Table 4.3). Ridged till plain is composed of stratified sediment in the form of broad, closely spaced ridges and hummocks with local relief as high as 30 m. It was formed by ice thrusting of till and ablation processes. Stratified sediment, which allows easy groundwater movement, moderate relief and ridges, which form numerous closed depressions, combine to provide the best conditions for pothole development (Sun and Fulton, 1995). Closed depressions allow water to pond and interact with the groundwater table. These characteristics have been identified as key components of pothole development.

The second most common surficial material for potholes is flat till plain with rimmed ridges (Tl+c) (Figure 4.4; Table 4.3) with 18% (1,893) of total potholes. It covers 422 km² and has 4.49 potholes/km². A flat till plain is till overlain by massive clayey silt less than 1.5 m thick. Relief is low (< 2 m) varying from flat to gently undulating topography and forms low rises and shallow depressions.

The third most common substrate is hummocky moraine (Th) with 17% of total potholes (Figure 4.4; Table 4.3). It covers a smaller area than flat till plain with rimmed ridges (355 km²) and has nearly as many potholes with 1,838 and consequently a higher density of potholes at 5.18 potholes/km². It is composed of silt, sand and gravel interstratified with till. The surface is marked by sharp ridges, hummocky mounds and hills with numerous closed depressions ranging in size from shallow and small, to deep and large. Relief is greater than 5 m. These are excellent characteristics for pothole development.

The fourth most common substrate for potholes is a flat till plain (Tl) with 9% of total potholes (Figure 4.4; Table 4.3). It covers 293 km² and has 968 potholes (3.31 potholes/km²). This material has a gently undulating topography with relief of less than 2 m with low rises and shallow depressions. It is composed of glacial till overlain by massive clayey silt (Sun and Fulton, 1995).

The material with the highest density of potholes is ridged till plain with low mounds (Tr+m) (Figure 4.4; Table 4.3; Appendix A). It covers only 9 km² (0.23%) and has 85 potholes (0.79% of total), or 9.20 potholes/km². There is no information in the SCL data nor the topographic data to explain this high concentration of potholes. More detailed sight-specific work is required to explain this relationship.

Three of the four most common substrates are variations of till plains (Figure 4.4; Table 4.3) that, when combined, total 5,371 or 51% of total potholes compared with only 1,838 or 17% of the total potholes for hummocky moraine. However, the density of these till plains is 5.04 potholes/km², slightly lower than hummocky moraine. Additionally, if all till plains are averaged they have a density of only 3.63 potholes/km², much lower than hummocky moraine. Hummocky moraine, found only on Turtle Mountain, has the largest average size of potholes. This is skewed by the inclusion of all waterbodies in the study area except the largest lakes (Whitewater Lake and a portion of Oak Lake). Large waterbodies have developed on Turtle Mountain as a result of large closed depressions and high relief. In addition, groundwater supply is likely high.

When examining the 3 till plain types with the highest number of potholes, there is little difference in terms of pothole frequency and density in comparison to the other surficial geology classes (Figure 4.4; Table 4.3). It is notable that the second most common pothole surface material, flat till plain with rimmed ridges, has relief of less than 2 m compared with hummocky moraine with relief from 5 to 30 m. Flat till plains, while having a low relief, have gently undulating topography and form numerous depressions separated by shallow rises. Three possibilities can account for the numerous potholes which occur on material which, according to the literature, is not ideal for pothole development. The first is that the texture of the material, which is massive clayey silt, is excellent for pothole development. Second, there is a significant interaction with the water table allowing potholes to develop. Thirdly, there may be closed depressions not identifiable with the 8 m contour data. Although groundwater data at an appropriate scale is not available, the large number of potholes on flat till plains with fine grained massive sediment suggests significant groundwater interaction with surface depressions.

Meyboom (1966), Williams and Farvolden (1967), Mills and Zwarich (1986) and Winter (1989) all suggest that till, and particularly hummocky moraine till, is most suitable for pothole development. These qualitative assumptions hold, at least partially, in this study. While 17% of potholes in this study occur on hummocky moraine, at a density of 5.18 potholes/km², it is not as dominant as implied by these researchers. One possibility is the change over time of knoll and kettle and hummocky terminology. Efforts in using and creating surficial deposit maps are hampered by a continuous evolution in legend structure and terminology. As surficial geology maps are created or re-evaluated the legends must change to incorporate new information and changes in interpretation or terminology (Fulton, 1992). This study does not refute the assumptions made in the literature about hummocky moraine. However, it does introduce the importance of till, with topography other than hummocky, as important in determining the number of potholes on a particular substrate. Till plains of all varieties accounted for 70% of all pothole occurrences in the study area reaching densities as high as 9.20 potholes/km².

Potholes on lacustrine and glaciolacustrine material total 922 or 9% of potholes (Figure 4.4; Table 4.3). Glaciolacustrine material has neither topography nor surficial material which is advantageous for pothole development. Glaciolacustrine deposits develop as material is deposited into lakes formed from glacial runoff (Shelby, 1995). The relief on lacustrine and glaciolacustrine material is typically very low, less than 2 m. The material deposited ranges from clay to pebbles, but is dominated by finer material such as clay and silt (Sun and Fulton, 1995). Clay will hold large amounts of water but its discontinuous pore structure provides low permeability which is not conducive to groundwater movement or pothole movement.

The low relief means that the surface will not often intersect the groundwater table to produce potholes. Potholes may form in small depressions in the surface, either where the depression intersects the groundwater table for part or all of the year, or where water ponds during storm events or with spring runoff. Reduced permeability, coupled with the low relief, accounts for the relatively few potholes on this material.

The surficial geology, slope and pothole layer were overlain to determine the number of potholes that fell on each slope class and surficial geology unit. Table 4.4 displays the results.

Of particular interest is the relatively small area of hummocky moraine on Turtle Mountain. While having the third highest number and density of potholes it comprises a very small area. Similarly, it has the largest mean pothole size reinforcing the impact surficial geology and topography have on pothole location and characteristics. The topography, surficial geology and hydrologic characteristics of Turtle Mountain are very different from the rest of the study area. Specifically, Turtle Mountain has more irregular topography and a less integrated drainage network than the rest of the study area which are ideal parameters for numerous and large potholes to develop.

Ridged till plain (Tr) and slope class 1 (slope of less than or equal to 1%) had the most potholes with 2,411, covered 323 km² and had a density of 7.64 potholes/km². Second was flat till plain with rimmed ridges (Tl+c) on slope class 1 covering 420 km² with 1,891 potholes (4.50 potholes/km²). Flat till plain (Tl) was third with 928 potholes on slope class 1 which covered 264 km² with 3.52 potholes/km². Hummocky moraine (Th) on slope class 1 was fourth with 848 potholes covering 172 km² and a density of 4.93 potholes/km². The surficial geology and slope class combinations that had the most potholes were ridged till plain (Tr) on slope class 1. However, the highest density of potholes occurs on ridged till plains with mounds (Tr+m) on slope class 1, which covers only 6.74 km² but has 68 potholes (10.09 potholes/km²).

These results of finding potholes on low slope class on hummocky moraine are somewhat unexpected. Hummocky moraine has a topography which is irregular and relief is as high as 30 m. One explanation is that slope class 1 covered most of the study area, including Turtle Mountain which reduces the chance of potholes occurring on other material. Hummocky moraine does occur on the higher slope classes.

The till plain categories also had high pothole occurrences in combination with slope class 1. These categories cover a large percentage of the study area. The till plains have much lower relief than hummocky moraine and as a result, have fewer large depressions where the slope calculation could measure steeper slopes. This is expected since the 8 m contour interval is too coarse to identify small closed depressions on a till plain.

There is a considerable body of knowledge about prairie potholes examining all components from vegetation to wildlife to hydrology. However, there lacks an integrated and regional approach to examining the variables determining pothole distribution. The purpose of this thesis was to determine how information about potholes derived at a local scale, could provide a better understanding of pothole distribution at a regional scale. A GIS (ArcInfo) was used to determine whether current assumptions about pothole distribution and characteristics hold true at a regional scale.

Slope information, calculated from NTS data, yielded a strong relationship with pothole distribution. Low slopes contained the largest percentage and highest densities of potholes. As slopes increased, the number of potholes decreased. Further work should use more detailed elevation data to determine slope more precisely however, a strong relationship between topography and pothole location was evident.

Surface expression, derived from the SLC data, also displayed a strong relationship with pothole occurrence. Most of the study area was classified as knoll and kettle topography which had both the highest occurrence and the greatest density of potholes.

Analysis of depth to water table with pothole distribution showed that the deeper the water table the higher the occurrence and density of potholes. Potholes were expected to occur where the water table was close to the surface and regular interaction with groundwater could occur. The data, while showing high watertable areas, only displayed large areas of high water table where lakes exist.

The analysis of the ability of soils to drain showed that more and a greater density of potholes occurred in well drained soils. Poorly drained soils covered a very small portion of the study area and were located around Oak and Whitewater Lakes. The inclusion of larger potholes adjacent to Oak Lake accounts for the relatively high density of potholes in this substrate.

There is a strong relationship between available water capacity and pothole distribution. The density of potholes increased significantly from the 50 mm class to the 250 mm class. Pothole were highly concentrated in the small highly saturated areas represented by the 250 mm class.

Gleysolic soils, reported to be commonly formed where groundwater regularly interacts with the surface, were not found to occur in areas of high pothole density. This is not surprising given the lack of detailed digital soil data. Pothole distribution should provide an indication of where Gleysolic soils occur. It is expected that with more detailed soil data, Gleysolic soils will appear around individual potholes as it does around Whitewater and Oak Lakes, and discontinuously over broad areas where potholes are dense. Black Chernozemic soils had the most potholes (51,443) but also covered the largest area. Dark Gray Chernozemic/Luvisolic soils covered a very small area but had the greatest density of potholes. Like the poor soil drainage class this area is adjacent to Oak and Whitewater Lakes. The large number of potholes around Oak Lake included in this study account for the high density.

Surficial geology analyses yielded results that directly linked the number of potholes with particular surficial geology units. Till plains, ridged till plains in particular, had the greatest number and highest density of potholes. Flat till plains with rimmed ridges had the second highest number of potholes in part due to its large area. An irregular or hummocky moraine was expected to have the highest density of potholes. Hummocky moraine did not have the greatest number of potholes primarily because of the relative small area it occupies, however it does have the second highest density of potholes.

Pothole characteristics and location are dominantly controlled by topography and surficial geology. Similar results were found in this study. Potholes were most common on ridged till plain in combination with low slopes. Surficial geology and topography are strongly interrelated and showed a strong correlation between results of local scale studies and the more regional perspective taken here.

In general, potholes were most common in areas of well drained soil, greater than 3 m depth to water table, loamy-Black Chernozemic soil, high available water capacity, knoll and kettle topography, slopes less than 3% and till plains.

The results of this study can be deemed marginally successful. The application of site specific scale data and concepts was not complete due to data constraints. This study could not follow the guidelines established at the outset as regional scale data, specifically soil and groundwater data does not exist for any portion of the study area in digital format. Digital mapping of soil data is currently been undertaken by the provincial governments of Manitoba and Saskatchewan.

The primary suggestion for further research is to use appropriate data for a similar study. Soil data may have yielded more representative results had more detailed information been available. More importantly, a clear picture of pothole-groundwater relationships, such as how pothole permanence is affected by groundwater, could be determined had detailed groundwater data for the area, or even a portion of the area, been available.

An important aspect of this research worth further examination is groundwater supply. As landuse intensifies, with activities such as intensive livestock operations and continuous cropping, greater pressure will be placed on both ground and surface water resources. Further work should attempt to model groundwater flow using pothole distribution as a surrogate.

A field of research receiving more attention is ecosystem health. Potholes may play an important role as indicators of ecosystem health. Both ground and surface water supply and quality should be important aspects of ecosystem health worth further examination.

The authors would like to thank Dr. Robert J. Fulton of the Geological Survey of Canada for his guidance and patience and the Geological Survey of Canada and the NATMAP project for data and support. We would like to thank Agriculture and Agri-Food Canada-Prairie Farm Rehabilitation Administration for use of computing and library facilities. Thanks to Bob Parkinson (PFRA) for assistance with figures and finally, we would like to thank Glenda Samuelson for providing comments, assistance and support throughout this study.