Issue Paper/ Are Water Tables a Subdued Replica of the Topography? by Henk M. Haitjema1 and Sherry Mitchell-Bruker2 Abstract The water table in unconfined aquifers is often believed to be a subdued replica of the topography or land surface. However, this assumption has not been widely tested and in some cases has been found to be in error. An analysis of ground water rise in regional unconfined aquifers, using both a two-dimensional boundary element model and a one-dimensional Dupuit-Forchheimer model, reveals the conditions under which the water table does or does not closely follow the topography. A simple decision criterion is presented to estimate in advance under which conditions the water table is expected to be largely unrelated to the topography and under which conditions the topography controls the position of the water table. Introduction When formulating mathematical solutions to ground water flow problems in unconfined aquifers, the a priori unknown water table is treated as a ‘‘free surface,’’ its location depending on the solution itself. Early computational work often focused on local flow through or underneath dams and near stream or canal boundaries (Muskat 1937; Polubarinova-Kochina 1952, 1962; Harr 1962; Davis and De Wiest 1966). The closed-form analytic solutions to flow for these cases were developed by use of the hodograph method, an advanced conformal mapping technique to treat the a priori unknown upper aquifer boundary. Dupuit (1863) was probably the first to address problems of truly regional ground water flow. He simplified the problem by ignoring vertical flow, thus reducing the problem to that of one- or two-dimensional horizontal flow. Later, Forchheimer (1886) presented a differential equation for the water table elevation in unconfined aquifers, referring to Dupuit’s approximation. Hubbert (1940), using a flow net construction, was one of the first who provided a conceptual ground water flow model for flow between 1Corresponding author: SPEA 439, Indiana University, Bloomington, IN 47405; (812) 855-0563; fax (812) 855-7802; [email protected] 2Everglades National Park, 950 North Krome Avenue, Homestead, FL 33030; (305) 224-4286; fax (305) 224-4147; [email protected] Received August 2004, accepted January 2005. Copyright ª 2005 National Ground Water Association. doi: 10.1111/j.1745-6584.2005.00090.x (parallel) streams that included resistance to vertical flow. Toth (1963) also offered a regional ground water flow solution that included resistance to vertical flow, but he assumed the position of the water table in advance: a subdued replica of the topography. While the water table in his example was observed to, at least approximately, follow the topography, in many other settings the water table and the surface topography appear poorly correlated if not unrelated (Desbarats et al. 2002; Moore and Thompson 1996; Blaskova et al. 2002; Shaman et al. 2002). Desbarats et al. (2002) suggested that the concept of a correlation between the water table and the topography may be valid locally but not universally. The question then is under what circumstances does the topography control the water table and under what circumstances does it not. In this paper, we explore the different conditions that affect the position of the water table and its possible interaction with the topography. Whether using a twodimensional boundary element model or a one-dimensional Dupuit-Forchheimer model, we arrive at the same conclusions regarding the conditions under which the water table will or will not interact with the topography. Finally, we offer criteria for when the water table is controlled by the surface topography and when it is mostly independent of the topography and controlled by hydrological features. Regional Ground Water Flow The theories of Dupuit-Forchheimer and Toth are seemingly conflicting conceptual models of regional Vol. 43, No. 6—GROUND WATER—November–December 2005 (pages 781–786) 781 ground water flow. Toth (1963) introduced the concept of nested regional, intermediate, and local flow cells, with both horizontal and vertical components of flow being important. Dupuit and Forchheimer proposed to ignore vertical flow on a regional scale (Dupuit 1863; Forchheimer 1886). While in a modern interpretation of the Dupuit-Forchheimer approximation vertical flow is allowed but resistance to vertical flow is ignored (Kirkham 1967; Strack 1984), there is no opportunity for the formation of nested flow cells such as presented by Toth. We argue here that both models of regional flow can be expected to occur in nature but they apply to very different hydrogeological conditions. Moreover, we demonstrate that the Dupuit-Forchheimer approximation is often valid in areas with properties typical of ‘‘usable aquifers,’’ that is, aquifers from which adequate quantities of water can be pumped for domestic, industrial, or public use. Toth’s Model of Three-Dimensional Flow Cells Toth developed his theory of regional flow while studying the hummocky terrain of Alberta, Canada. This region is known to contain areas of deep aquifers of lowpermeability shales and sandstone overlain by a relatively thin layer of higher permeability sands (Ophori and Toth 1989). This stratigraphy accommodates large infiltration rates in the surface sands that provide continuous recharge to the underlying low-permeable aquifer. Toth was interested in deep geologic formations, so the aquifers that he modeled were deep relative to the distance between drainage basins. In one of his examples, the aquifer depth is 3048 m (10,000 feet) while the distance between the drainage divide and the stream is 6096 m (20,000 feet). In order for the water table to rise to the highest point in this theoretical watershed, the ratio of recharge over hydraulic conductivity must be greater than 0.2 (Mitchell-Bruker 1993). For example, assuming a recharge rate somewhere between 25.4 mm/year (1 inch/year) and 508 mm/year (20 inches/year) implies that the hydraulic conductivity of the basin must be somewhere between 0.0003 m/d (0.001 ft/ d) and 0.007 m/d (0.023 ft/d). A recharge over hydraulic conductivity ratio of 0.2 is quite high but could possibly occur when the low-permeable basin is overlain by a layer of much more permeable sand, as is the case in Toth’s study area (Ophori and Toth 1989). The more permeable sands could store much of the precipitation, rather than letting it run off overland, and cause a sustained infiltration into the underlying basin. In models where the water table is specified (e.g., Toth 1963), the distribution of recharge and discharge follows from the solution. An alternative approach, which is more commonly used in ground water modeling, is to apply recharge to the aquifer top and only specify the head where streams interact with the aquifer. In Figures 1 and 2, a steady-state, two-dimensional boundary element model is used to compare these two approaches (Mitchell-Bruker 1993). In this model, linear strength line sink functions (Strack 1989) are used to create either recharge- or headspecified boundaries to the flow regime. The basin is delineated by no-flow boundaries on the sides and bottoms, which are obtained by image line sinks (Haitjema 1995). In the recharge-specified simulation, the recharge rate is specified for the line sinks and the water table is obtained from the solution (Figure 1a). In the headspecified simulations (Figure 2), the water table is specified and the recharge distribution is obtained from the solution. Figure 1a represents the case where the ratio of recharge over hydraulic conductivity is relatively low (R/k ¼ 1/12). In this situation, the recharge is equal to 25.4 mm/year (1 inch/year) throughout the watershed and the water table does not reach the land surface, except all the way to the left where it equals the water level in a drainage stream. In the higher regions of the watershed, the water table is far below the land surface. Figure 1b represents the case where the ratio of recharge to hydraulic conductivity is much higher (R/k ¼ 5/12). In this situation, the water table rises to intersect the land surface well Figure 1. Flow simulation of theoretical aquifer using recharge-specified boundary conditions: (a) recharge ¼ 25.4 mm/year, hydraulic conductivity ¼ 0.3 m/d; (b) recharge ¼ 126 mm/year, hydraulic conductivity ¼ 0.3 m/d (after Mitchell-Bruker 1993). 782 H.M. Haitjema, S. Mitchell-Bruker GROUND WATER 43, no. 6: 781–786 Figure 2. Flow simulation of theoretical aquifer using head-specified boundary conditions: (a) 1000 feet aquifer depth, (b) 10,000 feet aquifer depth (after Mitchell-Bruker 1993). before it links up with the left-hand drainage feature, forming a local flow cell near the bottom of the watershed. The land surface near the bottom of the watershed limits the water table rise, which is modeled by replacing the discharge-specified line sinks in that area by headspecified ones with the head equal to the land surface. These head-specified line sinks become discharge features as is seen from the flow lines in the left-hand part of the watershed in Figure 1b. Further to the right, the water table rises along with the ground surface; however, the distance between the water table and the ground surface increases as the land surface rises. When measuring the water table in Figure 1b at several locations in the watershed, it appears as if the water table is a subdued replica of the land surface, at least in a general sense. However, if a model is developed where the water table is everywhere equal to the land surface, a very different flow system is obtained (Figure 2a). In this simulation, numerous local flow cells are formed with zones of significant vertical flow. If the same head-specified upper boundary conditions (water table equal to the land surface) are applied to a deeper aquifer, as in Figure 2b, a system of local, intermediate, and regional flow is obtained. The last Figure 2b is similar to those presented by Toth (1963). Water Table in a Dupuit-Forchheimer Model The analysis presented previously is based on a twodimensional cross-sectional flow model, thus including resistance to vertical flow. Shallow regional aquifers, however, have been modeled successfully by use of the Dupuit-Forchheimer approximation, which ignores resistance to vertical flow. Such a model, however, leads to the same conclusions regarding the interaction of the water table and the topography as arrived at previously. The water table rise in a Dupuit-Forchheimer model of the theoretical watershed in Figure 1 can be described with a simple formula as shown subsequently. In Figure 3, areal recharge due to precipitation generates flow toward head-specified boundaries. If Figure 3 is interpreted as representing one-dimensional flow between two parallel streams, the mounding h [m], midway between the streams, follows from (Haitjema 1995) h ¼ RL2 8T ð1Þ where R [m/d] is the areal recharge rate, L [m] the distance between the streams, and T [m2/d] the average Figure 3. Ground water mounding due to recharge may interact with the topography. H.M. Haitjema, S. Mitchell-Bruker GROUND WATER 43, no. 6: 781–786 783 aquifer transmissivity. Figure 1 may be seen as the left half of Figure 3. Figure 3 may also be interpreted as flow in a circular island with diameter L [m], in which case the mounding in the center of the island becomes: h ¼ RL2 16T ð2Þ The average transmissivity T may be written as the aquifer hydraulic conductivity k [m/d] times the average aquifer thickness H [m], hence: T ¼ kH ð3Þ With Equation 3, both Equations 1 and 2 may be rewritten as: h ¼ RL2 mkH ð4Þ where m is either 8 or 16, depending on the flow problem being one-dimensional or radial symmetric, respectively. Equation 4 illustrates that the water table rise is very sensitive to the distance L between hydrological boundaries (surface waters). Given a particular aquifer geometry and assuming a constant saturated thickness H, however, the water table rise depends only on the ratio of recharge over hydraulic conductivity, R/k, which is dimensionless. Under unconfined flow conditions, the average saturated thickness H will increase with a rising water table and the response to R/k becomes nonlinear. The result for the mounding at the center of the aquifer in Equation 4 is obtained by ignoring resistance to vertical flow in the aquifer, which constitutes the DupuitForchheimer approximation. The validity of this approximation depends on the aquifer geometry and, to a lesser extent, on the degree of vertical anisotropy. The DupuitForchheimer approximation is discussed in most standard texts on ground water flow (e.g., Freeze and Cherry 1979; Bear 1972; Harr 1962). A detailed discussion on the Dupuit-Forchheimer approximation is provided by Kirkham (1967). As a rule of thumb, the DupuitForchheimer approximation is acceptable when the distance between hydrological boundaries is large compared to the ‘‘effective’’ aquifer thickness: rffiffiffiffiffi kh L H kv In Figure 4 a cross section over a regional aquifer is sketched with a water table in response to aquifer recharge for two different cases. In Figure 4a, a high recharge rate in combination with a low aquifer hydraulic conductivity would lead to the dashed water table if the aquifer would extend that high, while in Figure 4b a relatively low recharge rate and high aquifer hydraulic conductivity results in a water table that barely touches the terrain surface. In fact, what matters is not the absolute values of the recharge and the conductivity but the dimensionless ratio of R/k (Equation 4). In Figure 4a the water table can be characterized as ‘‘topography controlled,’’ while in Figure 4b there is essentially no correlation between the topography and the water table. We demonstrate that the nature of the water table (topography controlled or not) depends on the recharge, the aquifer transmissivity, the aquifer geometry, and to some degree the topography itself. We introduce three dimensionless ratios to characterize the ground water flow regime: (1) The recharge over hydraulic conductivity, R/k; (2) the distance between hydrological boundaries over the saturated aquifer thickness, L /H; and (3) the distance between hydrological boundaries over the maximum terrain rise, L /d. The maximum terrain rise d is defined here as the largest distance between the (average) surface water elevation and the terrain surface, as measured at the highest terrain elevation (Figure 3). The maximum ground water mounding h in an aquifer may be estimated by the following dimensionless relationship in terms of the previously defined ratios: rffiffiffiffiffi h 1 R L L kh ¼ H ð6Þ L d mk Hd kv ð5Þ where the square root term scales the aquifer thickness H to account for the difference in horizontal conductivity kh and vertical conductivity kv. In practice, it is often sufficient for L to be five times larger than the effective aquifer thickness (square root term times H ). In the event of an anisotropic aquifer, the hydraulic conductivity k in Equation 4 must be interpreted as the horizontal hydraulic conductivity kh. However, anisotropy does affect ground water mounding indirectly since it may invalidate the use of the Dupuit-Forchheimer approximation by violating Equation 5. Under these circumstances, the ground water mounding predicted by Equation 4 should be considered an underestimation. 784 Topography Vs. Recharge-Controlled Water Tables Figure 4. (a) Low-permeable aquifer with topography-controlled water table; (b) highly permeable aquifer with a recharge-controlled water table (after Haitjema 1995). H.M. Haitjema, S. Mitchell-Bruker GROUND WATER 43, no. 6: 781–786 The term h/d is the ratio of the maximum mounding over the maximum terrain rise, both measured with respect to the surface water elevations of the surrounding hydrological boundaries. Equation 6 is obtained from Equation 4, whereby k must be interpreted as the horizontal hydraulic conductivity. The integer constant m in Equation 6 depends on the type of flow. For one-dimensional flow between two parallel surface water boundaries m ¼ 8, while for radial flow in an aquifer of diameter L, m ¼ 16. In practice, therefore, m is expected to be somewhere in between these values: 8 m 16 ð7Þ Equation 6 is obtained by use of the DupuitForchheimer approximation, thus the condition that the distance between boundary conditions must be large compared to the aquifer thickness (Equation 5). If the ratio h/d is larger than 1, the water table will likely intersect the terrain surface at one or more locations. For values of h=d 1; the situation depicted in Figure 4a occurs where the topography essentially controls the water table location. For this case, the recharge rate R must be interpreted as the potential recharge rate since the actual recharge rate will vary over the terrain, being positive along the hillslopes and negative at local topographic lows. Conversely, when h=d 1 the situation in Figure 4b occurs where the water table is controlled by the recharge rate (and other aquifer parameters) and essentially unrelated to the topography. For the sake of brevity, we will refer to the situation in Figure 4b as that of a ‘‘recharge-controlled’’ water table. Equation 6, of course, does not offer a clean cut between the two cases in Figure 4. If the maximum ground water mounding is about the same as the maximum terrain rise, h=d ffi 1; the ground water flow regime is neither entirely recharge controlled nor entirely topography controlled. Yet, as a general rule of thumb the following may be stated: RL2 > 1 water table is topography controlled mkHd RL2 <1 mkHd water table is recharge controlled ð8Þ When reviewing the case presented by Toth (1963), we find that the distance between the regional water divide (on the right-hand side of his basin) and the primary discharge area on the left-hand side of the basin is 6096 m (20,000 feet). Assuming that the regional water divide is midway between two (regional) drainages, the distance L for his case is 12,192 m (40,000 feet). With a basin depth H of 3048 m (10,000 feet), we find that L /H ¼ 4. Estimating the maximum terrain rise d (difference between the terrain elevation on the right and the left of the basin) to be 30.48 m (100 feet), we find that L/d ¼ 400. Finally, from the maximum recharge over conductivity ratio that is consistent with the mounding on the right-hand side of the basin, we observed earlier that R/k ¼ 0.2. This maximum recharge over conductivity ratio (R/k ¼ 0.2) is interpreted as belonging to the ‘‘potential recharge rate.’’ With these numbers, and m ¼ 8, the dimensionless factor in Equation 8 is equal to 40, which explains the topography-controlled water table in Toth’s study. In fact, the use of the Dupuit-Forchheimer approximation in this assessment is suspect for this case with L /H ¼ 4, so that in view of the actual resistance to vertical flow, the water table mounding is likely greater than predicted by Equation 6, further ensuring a topography-controlled water table. For the case discussed previously, the factor R/k is rather large, R/k ¼ 0.2. Most productive aquifers (from which usable quantities of water can be pumped) are moderately to highly permeable with recharge over conductivity ratios in the range of 1026 to 1023 (at least 2 orders of magnitude lower than for Toth’s problem). With these lower R/k ratios, the water table tends to be more often recharge controlled. To illustrate this, consider the following example: The distance L ¼ 5000 m, the thickness H ¼ 50 m, the hydraulic conductivity k ¼ 50 m/d, the recharge rate R ¼ 25.6 mm/year (10 inch/year), which is 0.0000701 m/d. With these data and assuming a maximum terrain rise d ¼ 10 m, the dimensionless factor in Equation 8 becomes 0.0088. This is 2 orders of magnitude smaller than 1; hence, the water table will almost certainly not reach the terrain surface and is thus unrelated to the topography. Topography-controlled water tables are most likely to occur in low-permeable sediments, which often are not thought of as ‘‘aquifers,’’ at least not aquifers from which usable quantities of water can be pumped. They can be expected in systems where, due to circumstances, relatively high recharge rates are realized. These high recharge rates may be due to wetland conditions, where ponded water can augment aquifer recharge beyond that derived from normal rainfall. In situations where a lowpermeable formation is overlain by a (thin) layer of more permeable soils, a kind of ‘‘subsurface ponding’’ may occur resulting in relatively high recharge rates in the underlying low-permeable formation. In theory, the recharge over conductivity ratio can be R/k ¼ 1 for incipient ponding or slightly higher for true ponding. Additionally, if the terrain surface is relatively flat or the water table is generally shallow, the factor L /d may be large as well (small d ), further increasing the potential for topographic control of the water table. It should be pointed out that even if the water table is topography controlled, it does not automatically follow that the water table must be locally parallel to the terrain surface. In other words, even though the water table is found to be a ‘‘subdued replica of the topography,’’ the local hydraulic gradient may be quite different from the slope of the local terrain surface. If the water table is not known in advance of modeling, it is not advisable to specify a water table based on interpolation of a sparse set of measured water table elevations, much less on an assumed water table. When the head is specified instead of recharge rates, the recharge is not constrained to realistic rates and, if the water table is not represented accurately, the solution produces erratic and erroneous recharges and discharges (Stoerz and Bradbury 1989). H.M. Haitjema, S. Mitchell-Bruker GROUND WATER 43, no. 6: 781–786 785 Conclusions Acknowledgments Consistent with Toth’s original analysis, topography can significantly constrain ground water mounding. This type of topography-controlled water table most likely occurs in relatively low-permeable and/or anisotropic aquifers subjected to unusually high (in view of the low permeability) areal recharge rates or in nearly flat terrain. However, most aquifers that are developed for industrial or public water supply are relatively permeable and show moderate to small degrees of ground water mounding in response to average recharge. Such productive aquifers are generally recharge controlled. Recharge and conductivity, however, are not the only parameters that determine the difference between topography- or recharge-controlled water tables. As is seen from Equation 6, the aquifer thickness and distance between surface water features are also important. Consequently, shallow aquifers in flat or gently rolling terrain may exhibit a relatively low R/k ratio and still exhibit a water table that seems a subdued replica of the terrain surface. In fact, in a very general sense the water table and the topography are always related. After all, water table lows occur at surface waters, which in turn occur in topographically low areas. We offer a simple dimensionless decision criterion to assess the likelihood for either topography-controlled or recharge-controlled water tables. The authors thank Randy Hunt of the USGS in Madison, Wisconsin, for his valuable suggestions regarding this article. We also thank the reviewers, George Hornberger and Ken Bradbury for their constructive remarks. RL2 >1 mkHd RL2 <1 mkHd topography controlled recharge controlled where R [m/d] is the average annual recharge rate, L [m] the average distance between surface waters, m [2] a factor between 8 and 16 for aquifers that are strip like or circular in shape, respectively. The term k [m/d] is the (horizontal) aquifer hydraulic conductivity, H [m] the aquifer thickness, and d [m] the maximum distance between the average surface water levels and the terrain elevation. These parameters are indicated in Figure 3. The criterion has been developed for Dupuit-Forchheimer flow conditions, which generally occur when the distance between surface waters is large compared to the aquifer thickness. In case the Dupuit-Forchheimer approximation is not valid, some additional water table rise must be anticipated, somewhat increasing the chance for topography-controlled water table conditions when the factor in Equation 8 is near unity. Recharge-controlled water tables tend to be dominated by horizontal flow with mostly regional flow cells, while topography-controlled water tables tend to form local flow cells. Deep flow systems with topographycontrolled water tables tend to form local and regional flow systems as pointed out by Toth (1963). In most cases where the water table is topography controlled, it still is likely not a truly subdued replica of the topography in that the local hydraulic gradient is not correlated with the local slope of the terrain surface. 786 References Bear, J. 1972. Dynamics of Fluids in Porous Media. New York: American Elsevier. Blaskova, S., K. Beven, P. Tachecci, and A. Kulasova. 2002. Testing the distributed water table predictions of TOPMODEL (allowing for uncertainty in model calibration): The death of TOPMODEL? Water Resources Research 38, no. 11: 1257–1268. Davis, S.N., and R.J.M. De Wiest. 1966. Hydrogeology. New York: Wiley. Desbarats, A.J., C.E. Logan, M.J. Hinton, and D.R. Sharpe. 2002. On the Kriging of water table elevations using collateral information from a digital elevation model. Journal of Hydrology 255, no. 1: 25–39. Dupuit, J. 1863. E´tudes The´oriques et Practiques sur le Mouvement des Eaux dans les Canaux Decouverts et à Travers les Terrains Perme´ables, 2nd ed. Paris, France: Dunod. Forchheimer, P. 1886. Über die Ergiebigkeit von Brunnen-Anlagen und Sickerschlitzen. Zeitschrift des Architectenund Ingenieurs Vereins zu Hannover 32, 539–564. Freeze, R.A., and J.A. Cherry. 1979. Groundwater. Englewood Cliffs, New Jersey: Prentice Hall. Haitjema, H.M. 1995. Analytic Element Modeling of Groundwater Flow. San Diego, California: Academic Press. Harr, M.E. 1962. Groundwater and Seepage. New York: McGrawHill. Hubbert, M.K. 1940. The Theory of ground-water motion. The Journal of Geology 48, no. 8: 785–944. Kirkham, D. 1967. Explanation of paradoxes in DupuitForchheimer seepage theory. Water Resources Research 3, no. 2: 609–622. Mitchell-Bruker, S. 1993. Modeling steady state groundwater flow and surface water interactions. Ph.D. diss., School of Public and Environmental Affairs, Indiana University. Moore, R.D., and J.C. Thompson. 1996. Are water table variations in a shallow forest soil consistent with the TOPMODEL concept? Water Resources Research 32, no. 3: 663–669. Muskat, M. 1937. The Flow of Homogeneous Fluids through Porous Media. Ann Arbor, Michigan: McGraw-Hill. Ophori, D.U., and J. Toth. 1989. Characteristics of ground water flow by field mapping and numerical simulation, Ross Creek Basin, Alberta, Canada. Ground Water 27, no. 2: 193–201. Polubarinova-Kochina, P.Y. 1962. Theory of Groundwater Movement. Translated by J.M.R. De Wiest. Princeton, New Jersey: Princeton University Press. Polubarinova-Kochina, P.Y. 1952. Theory of Groundwater Movement [in Russian]. Moscow, Russia: Gostekhizdat. Shaman, J., M. Stieglitz, V. Engel, R. Koster, and C. Stark. 2002. Representation of subsurface storm flow and a more responsive water table in a TOPMODEL-based hydrology model. Water Resources Research 38, no. 8: 1156–1172. Stoerz, M.W., and K.R. Bradbury. 1989. Mapping recharge areas using a ground-water flow model: A case study. Ground Water 27, no. 2: 220–228. Strack, O.D.L. 1984. Three-dimensional streamlines in DupuitForschheimer models. Water Resources Research 20, no. 7: 812–822. Strack, O.D.L. 1989. Ground Water Mechanics. Englewood Cliffs, New Jersey: Prentice Hall. Toth, J. 1963. A theoretical analysis of groundwater flow in small drainage basins. Journal of Geophysical Research 68, no. 16: 4795–4812. H.M. Haitjema, S. Mitchell-Bruker GROUND WATER 43, no. 6: 781–786
© Copyright 2026 Paperzz