Estimation of recharge using a revised CRD method

Y Xu1* and GJ van Tonder2

Department of Water Affairs and Forestry, Pretoria Institute for Groundwater Studies, University of the Free State, Bloemfontein

Abstract
The cumulative rainfall departure (CRD) method, based on the water-balance principle, is often used for mimicking of water level fluctuations. Because of its simplicity and minimal requirement of spatial data, the CRD method has been applied widely for estimating either effective recharge or aquifer storativity, and consequently gained a focus in South Africa. This paper critically reviews this method and proposes expanded algorithm. Validation of the method under typical South African conditions is discussed based on model-generated and known cases. The study is aided with a user-friendly Excel program called Recharge Estimation Model in Excel (REME).

Introduction
Background Hydrogeologists often compare rainfall and groundwater levels for estimation of groundwater recharge. The reader may refer to Wenzel (1936), Sophocleous (1991) and Wu et al.(1996). In South Africa Bredenkamp et al. (1995) applied the CRD method in dolomitic aquifers and promoted the method through their publication entitled Manual on Quantitative Estimation of Groundwater Recharge and Aquifer Storativity. Their approach is based on the premise that equilibrium conditions develop in an aquifer over time, i.e. average rate of losses equating to average rate of recharge of the system. They clearly showed that natural groundwater level fluctuation is related to that of the departure of rainfall from the mean rainfall of the preceding time. If the departure is positive, the water level will rise and vise versa. However, it can be demonstrated that as long as there is a surplus of recharge over discharge of an aquifer, even though the departure is negative, the natural water level may continue to rise. Purpose

Theory
Groundwater balance Assuming an aquifer of area (A) receiving recharge from rainfall (QR) with production boreholes (Qp) tapping the aquifer and with natural outflow (Qout), a simple water balance equation for a given time interval i can be written as follows:

QRi = Q pi + Qouti + hi AS

(i = 1, 2, 3...N )

(1)

where hi is water level change and S aquifer storativity (specific yield). If QRi is averaged over such a time interval where hi is zero, the system may be treated as in equilibrium. This is, however, seldom the case in reality. If Qpi is a constant rate, aquifer storage (hiAS) adjusts to accommodate for net balance between QRi and Qouti. This adjustment of the storage would be reflected in piezometric surface or water level change in boreholes. The cause-effect relationship between rainfall oscillation and water-level fluctuation is effectively represented by the correlation between the CRD and water level fluctuation. Recharge formulae

The purpose of this paper is to revisit the existing method and to improve the algorithm to accommodate a wide variety of circumstances. Following improvement of the algorithm a userfriendly tool could be developed for groundwater practitioners. Such a need was identified by the Department of Water Affairs and Forestry, which sponsored a project aimed at promoting the effective use of simple yet powerful methods for recharge estimation. This paper summarises some results of this project.

Bredenkamp formula Bredenkamp et al. (1995) defined CRD as follows:

CRDi = Rn Rav (i = 0, 1, 2, 3, ... N )

(2)

where R is rainfall amount with subscript i indicating the i-th month, av the average and = 1+(Qp+Qout)/(ARav). = 1 indicates that pumping does not occur and > 1 if pumping and/or natural outflow takes place. It is assumed that a CRD has a linear relationship with a monthly water level change. Bredenkamp et al. (1995) derived hi = (r / S) (1vCRDi) (i = 0, 1, 2, 3, ...N) a (3)

* To whom all correspondence should be addressed. New address: Department of Earth Sciences, University of The Western Cape, Private Bag x17, Bellville 7535, South Africa ((021) 959-2223; fax (021) 959-2438; e-mail: [email protected] Received 6 July 2000; accepted in revised form 3 April 2001.

where r is a percentage of the CRD which results in recharge from rainfall.

Available on website http://www.wrc.org.za

ISSN 0378-4738 = Water SA Vol. 27 No. 3 July 2001

341

Eq. (3) may be used to estimate the ratio of recharge to aquifer storativity through simple regression between CRDi and hi (Bredenkamp et al., 1995).

Recharge Estimated through Simulation of Water level Fluctuation Rain (mm) 12.00 10.00 8.00 water levels (m) 6.00 4.00 2.00 0.00 -2.00 -4.00 -6.00 1 3 5 7 9 11 13 15 17 19 21 Time (months) 23 Water level dh(new) dh(crd) 180.00 160.00 140.00 rainfall(mm) 120.00 100.00 80.00 60.00 40.00 20.00 0.00

New formula It is often the case that an appropriate value of the parameter in Eq. (2) must be chosen to mimic adequately the water level fluctuation in boreholes. However, its physical meaning is still unclear. Rainfall time series in general are composed of random and deterministic components, the latter is in the form of trends and periodicities. A short series of data often displays a trend to a certain degree, which cannot be reflected in Eq. (2). A new CRD has therefore been formulated to account for such a trend: i i i 1 1 R n Rt t CRDi = Rn 2 Rav i n =1 n =1 n =1 (4) (i = 1, 2, 3, ... N )
where Rt, a threshold value representing aquifer boundary conditions, is determined during the simulation process. It may range from 0 to Rav with 0 indicating an aquifer being closed and Rav implying that the aquifer system is open, perhaps being regulated by spring flow. Note that Eq. (4) reduces to Eq. (2) if rainfall events Ri do not show a trend (Rt = Rav). In this case, cumulative rainfall average would conform to Rav. It is assumed that CRD is the driving force behind a monthly water level change if the other stresses are relatively constant. The groundwater level will rise if the cumulative departure is positive and it will decline if the cumulative departure is negative. Since CRD (h + (Qp + Qout)/(AS)), then rCRD = S(h + (Qp + Qout)/(AS)). After rearrangement, one obtains the following: hi = (r / S ) CRDi (Qpi + Qouti )/ (AS )
1 t

Figure 1 Simulation of groundwater fluctuation using the CRD method based on model generated data
Water level patterns Rain (mm) 4.00 3.50 3.00 2.50 water levels (m) 1.50 1.00 0.50 0.00 -0.50 -1.00 -1.50 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 103 109 Time (months) 115 0.00 50.00 rainfall(mm) 2.00 200.00 150.00 100.00 250.00 Water level dh(new) dh(crd) 300.00

Figure 2 Simulation of groundwater fluctuation using the CRD method based on data from Grootfontein compartment
Recharge Estimated through Simulation of Water level Fluctuation Rain (mm) 7.00 6.00 5.00 water levels (m) 4.00 3.00 2.00 1.00 0.00 -1.00 -2.00 -3.00 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 Time (months) 33 200.00 150.00 100.00 50.00 0.00 Water level dh(new) dh(crd) 450.00 400.00 350.00 rainfall(mm) 300.00 250.00

(i = 0, 1, 2, 3, ... N )

(5)

Term (Qpi + Qouti)/(AS) in Eq. (5) is necessary only if a pumping hole has influence over the study area where water levels were collected. Eq. (5) may be used to estimate the ratio of recharge to aquifer storativity through minimising the difference between calculated and measured hi series. This optimisation is implemented in a user-friendly Excel program called REME (available on request).

Discussion
Analysis of Eq. (2) through Eq. (5) reveals the following facts: If rainfall Ri is constant over time, Ri=Rav and groundwater levels do not fluctuate naturally. Steady state conditions prevail. Term (Qpi + Qouti)/(AS) of Eq. (5) is necessary only if the influence of pumping and/or outflow on water level changes is evident.

Figure 3 Simulation of groundwater fluctuation using the CRD method based on data from Dewetsdorp aquifer

342

ISSN 0378-4738 = Water SA Vol. 27 No. 3 July 2001

Available on website http://www.wrc.org.za

This may be true in cases of highly fractured dolomitic aquifers where high values of transmisivity are encountered. Only the ratio r/S can be determined through water level simulation. Since Eq. (3) makes use of Eq. (2), Eq. (3) cannot accommodate for variable pumping rates. Eq. (5) accounts for changing pumping and outflow rates (Qpi + Qouti). Eq. (3) implicitly assumes that there are no long-term trends in the rainfall.

Water balance based methods are lumped parameter approaches. They do not address parameter variation in space and should be applied with caution.

Case studies
Closed aquifer system A hypothetical aquifer with recharge of 2% of rainfall over a closed area of 5x5 km2 has a borehole at the centre pumping at a rate of 15 000 m3 per month. The aquifer has a storativity of 1x10-3. Water levels over 24 months are generated using Modflow-based software. The water level series is simulated using the computer program REME. Comparison of simulated water levels with the generated ones is shown in Fig. 1 where dh(crd) are water levels calculated using Bredenkamp et al. (1995) Eqs. (2) and (3), while dh(new) are water levels calculated using Eq. (5). The average modelled recharge is 1.79% of rainfall. Dolomite aquifer The Grootfontein aquifer is compartmented by dolerite dykes. The compartment situated in the recharge zone covers an area of 1.25x103 km2. Aquifer storativity has been estimated at 2.39% (Bredenkamp et al., 1995). Both methods were applied to this case. Results are shown in Fig. 2. Based on the revised CRD method (Eq. (5)), a recharge value of 5.71% of rainfall was calculated whereas Bredenkamp et al. (1995) formulae yielded a value of 11%. Karoo aquifer The Karoo aquifer in Dewetsdorp was investigated by Kirchner et al. (1991). It covers an area of 21 km2 with aquifer storativity estimated at 0.19%. Both methods were applied to this case and yielded an average recharge of 1.45% of rainfall as shown in Fig. 3. Applying Eq. (5) produces a better fit.

According to Vegter (1995), the depths to water tables in aquifers with secondary porosity (e.g. fractured and weathered rock) is commonly between 10 and 30 m below ground level. This group includes such primary drainage regions as denoted D, F, G, K, P, R and Q (Midgley et al., 1994). The depths to water tables in fractured and karstic aquifers often vary from 10 to 40 m below ground level. This group includes such primary drainage regions as E, M, S, V and W (Midgley et al., 1994). The depths to water levels in weathered fractured aquifers are often ranging from10 to 125 m below ground level. This group includes such primary drainage regions as A, B, C, H, L, I, J, N, O, T and U (Midgley et al., 1994). In general, the depths range from 10 to 125 ms. If depths within 50 m are termed as shallow aquifers, those beyond 50 m are relatively deep aquifers. If the depth to the groundwater table exceeds 50 m we recommend that consideration of the CRD method be applied with time lags. This is due to the different filtering (delay) effect of rainfall passing through the unsaturated zone (Wu et al., 1996). Fluctuation of water levels in deep aquifers thus may be retarded or smoothened, and consequently fails to correspond to the rainfall signals.

Conclusion
The CRD method is simple but powerful for groundwater recharge estimation. Since > 1, Eq. (3) can yield erroneously high recharge values. The revised formula proposed in this paper is able to account for rainfall series with trends and is more accurate. The ratio r/S can be estimated for shallow aquifers using the CRD method, which does not require a large amount of spatial data. The estimation can be optimised through a Solver built in Excel. The applicability of the CRD method for deep aquifers remains to be verified.

Acknowledgements
The authors would like to thank the Department of Water Affairs and Forestry for permission to publish this paper.

References
BREDENKAMP DB, BOTHA LJ, VAN TONDER GJ and VAN RENSBURG HJ (1995) Manual on Quantitative Estimation of Groundwater Recharge and Aquifer Storativity. WRC Report No TT 73/95 KIRCHNER R, VAN TONDER GJ and LUKAS E (1991) Exploitation Potential of Karoo Aquifers. WRC Report No. 170/1/91. MIDGLEY DC, PITMAN WV and MIDDLETON BJ (1995) The Surface Water Resources of South Africa, Vols. 1 to 6. WRC Report Numbers 298/1.1/94 to 298/6.1/94 (text) and Numbers 298/1.2/94 to 298/6.2/94 (maps). SOPHOCLEOUS MA (1991) Combining the soilwater balance and the water-level fluctuation methods to estimate natural groundwater recharge: Practical aspects. J. Hydrol. 124 229-241. VEGTER JR (1995) An Explanation of a Set of National Groundwater Maps. WRC Report No 74/95. WENZEL LK (1936) Several methods of studying of groundwater levels. Trans. Am. Geophys. Union . Vol. 17. WU JR, ZHANG T and YANG J (1996) Analysis of rainfall-recharge relationship. J. Hydrol. 177 143-160.

Limitations
Most aquifers in South Africa are of a fractured nature with small storativities. Hence, changes in groundwater levels in these aquifers are very sensitive to recharge from rainfall. For this type of aquifers simulation of water levels based on the CRD is less accurate. Another important factor influencing the accuracy of the CRD is the depth to the groundwater table in these aquifers.

Available on website http://www.wrc.org.za

ISSN 0378-4738 = Water SA Vol. 27 No. 3 July 2001

343

344

ISSN 0378-4738 = Water SA Vol. 27 No. 3 July 2001

Available on website http://www.wrc.org.za